Annotation of cell-types is a critical step in the analysis of single-cell RNA sequencing (scRNA-seq) data that allows the study of heterogeneity across multiple cell populations. Currently, this is most commonly done using unsupervised clustering algorithms, which project single-cell expression data into a lower dimensional space and then cluster cells based on their distances from each other. However, as these methods do not use reference datasets, they can only achieve a rough classification of cell-types, and it is difficult to improve the recognition accuracy further. To effectively solve this issue, we propose a novel supervised annotation method, scDeepInsight. The scDeepInsight method is capable of performing manifold assignments. It is competent in executing data integration through batch normalization, performing supervised training on the reference dataset, doing outlier detection and annotating cell-types on query datasets. Moreover, it can help identify active genes or marker genes related to cell-types. The training of the scDeepInsight model is performed in a unique way. Tabular scRNA-seq data are first converted to corresponding images through the DeepInsight methodology. DeepInsight can create a trainable image transformer to convert non-image RNA data to images by comprehensively comparing interrelationships among multiple genes. Subsequently, the converted images are fed into convolutional neural networks such as EfficientNet-b3. This enables automatic feature extraction to identify the cell-types of scRNA-seq samples. We benchmarked scDeepInsight with six other mainstream cell annotation methods. The average accuracy rate of scDeepInsight reached 87.5%, which is more than 7% higher compared with the state-of-the-art methods.

Advances in single-cell multi-omics technology provide an unprecedented opportunity to fully understand cellular heterogeneity. However, integrating omics data from multiple modalities is challenging due to the individual characteristics of each measurement. Here, to solve such a problem, we propose a contrastive and generative deep self-expression model, called single-cell multimodal self-expressive integration (scMSI), which integrates the heterogeneous multimodal data into a unified manifold space. Specifically, scMSI first learns each omics-specific latent representation and self-expression relationship to consider the characteristics of different omics data by deep self-expressive generative model. Then, scMSI combines these omics-specific self-expression relations through contrastive learning. In such a way, scMSI provides a paradigm to integrate multiple omics data even with weak relation, which effectively achieves the representation learning and data integration into a unified framework. We demonstrate that scMSI provides a cohesive solution for a variety of analysis tasks, such as integration analysis, data denoising, batch correction and spatial domain detection. We have applied scMSI on various single-cell and spatial multimodal datasets to validate its high effectiveness and robustness in diverse data types and application scenarios.

Single cell RNA-sequencing (scRNA-seq) technology has significantly advanced the understanding of transcriptomic signatures. Although various statistical models have been used to describe the distribution of gene expression across cells, a comprehensive assessment of the different models is missing. Moreover, the growing number of features associated with scRNA-seq datasets creates new challenges for analytical accuracy and computing speed. Here, we developed a Python-based package (TensorZINB) to solve the zero-inflated negative binomial (ZINB) model using the TensorFlow deep learning framework. We used a sequential initialization method to solve the numerical stability issues associated with hurdle and zero-inflated models. A recursive feature selection protocol was used to optimize feature selections for data processing and downstream differentially expressed gene (DEG) analysis. We proposed a class of hybrid models combining nested models to further improve the model's performance. Additionally, we developed a new method to convert a continuous distribution to its equivalent discrete form, so that statistical models can be fairly compared. Finally, we showed that the proposed TensorFlow algorithm (TensorZINB) was numerically stable and that its computing speed and performance were superior to those of existing ZINB solvers. Moreover, we implemented seven hurdle and zero-inflated statistical models in Python and systematically assessed their performance using a real scRNA-seq dataset. We demonstrated that the ZINB model achieved the lowest Akaike information criterion compared with other models tested. Taken together, TensorZINB was accurate, efficient and scalable for the implementation of ZINB and for large-scale scRNA-seq data analysis with DEG identification.

Drug-drug interaction (DDI) identification is essential to clinical medicine and drug discovery. The two categories of drugs (i.e. chemical drugs and biotech drugs) differ remarkably in molecular properties, action mechanisms, etc. Biotech drugs are up-to-comers but highly promising in modern medicine due to higher specificity and fewer side effects. However, existing DDI prediction methods only consider chemical drugs of small molecules, not biotech drugs of large molecules. Here, we build a large-scale dual-modal graph database named CB-DB and customize a graph-based framework named CB-TIP to reason event-aware DDIs for both chemical and biotech drugs. CB-DB comprehensively integrates various interaction events and two heterogeneous kinds of molecular structures. It imports endogenous proteins founded on the fact that most drugs take effects by interacting with endogenous proteins. In the modality of molecular structure, drugs and endogenous proteins are two heterogeneous kinds of graphs, while in the modality of interaction, they are nodes connected by events (i.e. edges of different relationships). CB-TIP employs graph representation learning methods to generate drug representations from either modality and then contrastively mixes them to predict how likely an event occurs when a drug meets another in an end-to-end manner. Experiments demonstrate CB-TIP's great superiority in DDI prediction and the promising potential of uncovering novel DDIs.

Cytometry enables precise single-cell phenotyping within heterogeneous populations. These cell types are traditionally annotated via manual gating, but this method lacks reproducibility and sensitivity to batch effect. Also, the most recent cytometers-spectral flow or mass cytometers-create rich and high-dimensional data whose analysis via manual gating becomes challenging and time-consuming. To tackle these limitations, we introduce Scyan https://github.com/MICS-Lab/scyan, a Single-cell Cytometry Annotation Network that automatically annotates cell types using only prior expert knowledge about the cytometry panel. For this, it uses a normalizing flow-a type of deep generative model-that maps protein expressions into a biologically relevant latent space. We demonstrate that Scyan significantly outperforms the related state-of-the-art models on multiple public datasets while being faster and interpretable. In addition, Scyan overcomes several complementary tasks, such as batch-effect correction, debarcoding and population discovery. Overall, this model accelerates and eases cell population characterization, quantification and discovery in cytometry.

Here, we present AtacAnnoR, a two-round annotation method for scATAC-seq data using well-annotated scRNA-seq data as reference. We evaluate AtacAnnoR's performance against six competing methods on 11 benchmark datasets. Our results show that AtacAnnoR achieves the highest mean accuracy and the highest mean balanced accuracy and performs particularly well when unpaired scRNA-seq data are used as the reference. Furthermore, AtacAnnoR implements a 'Combine and Discard' strategy to further improve annotation accuracy when annotations of multiple references are available. AtacAnnoR has been implemented in an R package and can be directly integrated into currently popular scATAC-seq analysis pipelines.

Vertical federated learning has gained popularity as a means of enabling collaboration and information sharing between different entities while maintaining data privacy and security. This approach has potential applications in disease healthcare, cancer prognosis prediction, and other industries where data privacy is a major concern. Although using multi-omics data for cancer prognosis prediction provides more information for treatment selection, collecting different types of omics data can be challenging due to their production in various medical institutions. Data owners must comply with strict data protection regulations such as European Union (EU) General Data Protection Regulation. To share patient data across multiple institutions, privacy and security issues must be addressed. Therefore, we propose an adaptive optimized vertical federated-learning-based framework adaptive optimized vertical federated learning for heterogeneous multi-omics data integration (AFEI) to integrate multi-omics data collected from multiple institutions for cancer prognosis prediction. AFEI enables participating parties to build an accurate joint evaluation model for learning more information related to cancer patients from different perspectives, based on the distributed and encrypted multi-omics features shared by multiple institutions. The experimental results demonstrate that AFEI achieves higher prediction accuracy (6.5% on average) than using single omics data by utilizing the encrypted multi-omics data from different institutions, and it performs almost as well as prognosis prediction by directly integrating multi-omics data. Overall, AFEI can be seen as an efficient solution for breaking down barriers to multi-institutional collaboration and promoting the development of cancer prognosis prediction.

Numerous biological studies have shown that considering disease-associated micro RNAs (miRNAs) as potential biomarkers or therapeutic targets offers new avenues for the diagnosis of complex diseases. Computational methods have gradually been introduced to reveal disease-related miRNAs. Considering that previous models have not fused sufficiently diverse similarities, that their inappropriate fusion methods may lead to poor quality of the comprehensive similarity network and that their results are often limited by insufficiently known associations, we propose a computational model called Generative Adversarial Matrix Completion Network based on Multi-source Data Fusion (GAMCNMDF) for miRNA-disease association prediction. We create a diverse network connecting miRNAs and diseases, which is then represented using a matrix. The main task of GAMCNMDF is to complete the matrix and obtain the predicted results. The main innovations of GAMCNMDF are reflected in two aspects: GAMCNMDF integrates diverse data sources and employs a nonlinear fusion approach to update the similarity networks of miRNAs and diseases. Also, some additional information is provided to GAMCNMDF in the form of a 'hint' so that GAMCNMDF can work successfully even when complete data are not available. Compared with other methods, the outcomes of 10-fold cross-validation on two distinct databases validate the superior performance of GAMCNMDF with statistically significant results. It is worth mentioning that we apply GAMCNMDF in the identification of underlying small molecule-related miRNAs, yielding outstanding performance results in this specific domain. In addition, two case studies about two important neoplasms show that GAMCNMDF is a promising prediction method.

Artificial intelligence (AI) systems utilizing deep neural networks and machine learning (ML) algorithms are widely used for solving critical problems in bioinformatics, biomedical informatics and precision medicine. However, complex ML models that are often perceived as opaque and black-box methods make it difficult to understand the reasoning behind their decisions. This lack of transparency can be a challenge for both end-users and decision-makers, as well as AI developers. In sensitive areas such as healthcare, explainability and accountability are not only desirable properties but also legally required for AI systems that can have a significant impact on human lives. Fairness is another growing concern, as algorithmic decisions should not show bias or discrimination towards certain groups or individuals based on sensitive attributes. Explainable AI (XAI) aims to overcome the opaqueness of black-box models and to provide transparency in how AI systems make decisions. Interpretable ML models can explain how they make predictions and identify factors that influence their outcomes. However, the majority of the state-of-the-art interpretable ML methods are domain-agnostic and have evolved from fields such as computer vision, automated reasoning or statistics, making direct application to bioinformatics problems challenging without customization and domain adaptation. In this paper, we discuss the importance of explainability and algorithmic transparency in the context of bioinformatics. We provide an overview of model-specific and model-agnostic interpretable ML methods and tools and outline their potential limitations. We discuss how existing interpretable ML methods can be customized and fit to bioinformatics research problems. Further, through case studies in bioimaging, cancer genomics and text mining, we demonstrate how XAI methods can improve transparency and decision fairness. Our review aims at providing valuable insights and serving as a starting point for researchers wanting to enhance explainability and decision transparency while solving bioinformatics problems. GitHub: https://github.com/rezacsedu/XAI-for-bioinformatics.

The Hi-C experiments have been extensively used for the studies of genomic structures. In the last few years, spatiotemporal Hi-C has largely contributed to the investigation of genome dynamic reorganization. However, computationally modeling and forecasting spatiotemporal Hi-C data still have not been seen in the literature. We present HiC4D for dealing with the problem of forecasting spatiotemporal Hi-C data. We designed and benchmarked a novel network and named it residual ConvLSTM (ResConvLSTM), which is a combination of residual network and convolutional long short-term memory (ConvLSTM). We evaluated our new ResConvLSTM networks and compared them with the other five methods, including a naïve network (NaiveNet) that we designed as a baseline method and four outstanding video-prediction methods from the literature: ConvLSTM, spatiotemporal LSTM (ST-LSTM), self-attention LSTM (SA-LSTM) and simple video prediction (SimVP). We used eight different spatiotemporal Hi-C datasets for the blind test, including two from mouse embryogenesis, one from somatic cell nuclear transfer (SCNT) embryos, three embryogenesis datasets from different species and two non-embryogenesis datasets. Our evaluation results indicate that our ResConvLSTM networks almost always outperform the other methods on the eight blind-test datasets in terms of accurately predicting the Hi-C contact matrices at future time-steps. Our benchmarks also indicate that all of the methods that we benchmarked can successfully recover the boundaries of topologically associating domains called on the experimental Hi-C contact matrices. Taken together, our benchmarks suggest that HiC4D is an effective tool for predicting spatiotemporal Hi-C data. HiC4D is publicly available at both http://dna.cs.miami.edu/HiC4D/ and https://github.com/zwang-bioinformatics/HiC4D/.

Access to accurate viral genomes is important to downstream data analysis. Third-generation sequencing (TGS) has recently become a popular platform for virus sequencing because of its long read length. However, its per-base error rate, which is higher than next-generation sequencing, can lead to genomes with errors. Polishing tools are thus needed to correct errors either before or after sequence assembly. Despite promising results of available polishing tools, there is still room to improve the error correction performance to perform more accurate genome assembly. The errors, particularly those in coding regions, can hamper analysis such as linage identification and variant monitoring. In this work, we developed a novel pipeline, HMMPolish, for correcting (polishing) errors in protein-coding regions of known RNA viruses. This tool can be applied to either raw TGS reads or the assembled sequences of the target virus. By utilizing profile Hidden Markov Models of protein families/domains in known viruses, HMMPolish can correct errors that are ignored by available polishers. We extensively validated HMMPolish on 34 datasets that covered four clinically important viruses, including HIV-1, influenza-A, norovirus, and severe acute respiratory syndrome coronavirus 2. These datasets contain reads with different properties, such as sequencing depth and platforms (PacBio or Nanopore). The benchmark results against popular/representative polishers show that HMMPolish competes favorably on error correction in coding regions of known RNA viruses.

Factor analysis, ranging from principal component analysis to nonnegative matrix factorization, represents a foremost approach in analyzing multi-dimensional data to extract valuable patterns, and is increasingly being applied in the context of multi-dimensional omics datasets represented in tensor form. However, traditional analytical methods are heavily dependent on the format and structure of the data itself, and if these change even slightly, the analyst must change their data analysis strategy and techniques and spend a considerable amount of time on data preprocessing. Additionally, many traditional methods cannot be applied as-is in the presence of missing values in the data. We present a new statistical framework, unified nonnegative matrix factorization (UNMF), for finding informative patterns in messy biological data sets. UNMF is designed for tidy data format and structure, making data analysis easier and simplifying the development of data analysis tools. UNMF can handle a wide range of data structures and formats, and works seamlessly with tensor data including missing observations and repeated measurements. The usefulness of UNMF is demonstrated through its application to several multi-dimensional omics data, offering user-friendly and unified features for analysis and integration. Its application holds great potential for the life science community. UNMF is implemented with R and is available from GitHub (https://github.com/abikoushi/moltenNMF).

Most life activities in organisms are regulated through protein complexes, which are mainly controlled via Protein-Protein Interactions (PPIs). Discovering new interactions between proteins and revealing their biological functions are of great significance for understanding the molecular mechanisms of biological processes and identifying the potential targets in drug discovery. Current experimental methods only capture stable protein interactions, which lead to limited coverage. In addition, expensive cost and time consuming are also the obvious shortcomings. In recent years, various computational methods have been successfully developed for predicting PPIs based only on protein homology, primary sequences of protein or gene ontology information. Computational efficiency and data complexity are still the main bottlenecks for the algorithm generalization. In this study, we proposed a novel computational framework, HNSPPI, to predict PPIs. As a hybrid supervised learning model, HNSPPI comprehensively characterizes the intrinsic relationship between two proteins by integrating amino acid sequence information and connection properties of PPI network. The experimental results show that HNSPPI works very well on six benchmark datasets. Moreover, the comparison analysis proved that our model significantly outperforms other five existing algorithms. Finally, we used the HNSPPI model to explore the SARS-CoV-2-Human interaction system and found several potential regulations. In summary, HNSPPI is a promising model for predicting new protein interactions from known PPI data.

Great efforts have been made to develop precision medicine-based treatments using machine learning. In this field, where the goal is to provide the optimal treatment for each patient based on his/her medical history and genomic characteristics, it is not sufficient to make excellent predictions. The challenge is to understand and trust the model's decisions while also being able to easily implement it. However, one of the issues with machine learning algorithms-particularly deep learning-is their lack of interpretability. This review compares six different machine learning methods to provide guidance for defining interpretability by focusing on accuracy, multi-omics capability, explainability and implementability. Our selection of algorithms includes tree-, regression- and kernel-based methods, which we selected for their ease of interpretation for the clinician. We also included two novel explainable methods in the comparison. No significant differences in accuracy were observed when comparing the methods, but an improvement was observed when using gene expression instead of mutational status as input for these methods. We concentrated on the current intriguing challenge: model comprehension and ease of use. Our comparison suggests that the tree-based methods are the most interpretable of those tested.

Recent advances and achievements of artificial intelligence (AI) as well as deep and graph learning models have established their usefulness in biomedical applications, especially in drug-drug interactions (DDIs). DDIs refer to a change in the effect of one drug to the presence of another drug in the human body, which plays an essential role in drug discovery and clinical research. DDIs prediction through traditional clinical trials and experiments is an expensive and time-consuming process. To correctly apply the advanced AI and deep learning, the developer and user meet various challenges such as the availability and encoding of data resources, and the design of computational methods. This review summarizes chemical structure based, network based, natural language processing based and hybrid methods, providing an updated and accessible guide to the broad researchers and development community with different domain knowledge. We introduce widely used molecular representation and describe the theoretical frameworks of graph neural network models for representing molecular structures. We present the advantages and disadvantages of deep and graph learning methods by performing comparative experiments. We discuss the potential technical challenges and highlight future directions of deep and graph learning models for accelerating DDIs prediction.

Nucleic-acid G-quadruplexes (G4s) play vital roles in many cellular processes. Due to their importance, researchers have developed experimental assays to measure nucleic-acid G4s in high throughput. The generated high-throughput datasets gave rise to unique opportunities to develop machine-learning-based methods, and in particular deep neural networks, to predict G4s in any given nucleic-acid sequence and any species. In this paper, we review the success stories of deep-neural-network applications for G4 prediction. We first cover the experimental technologies that generated the most comprehensive nucleic-acid G4 high-throughput datasets in recent years. We then review classic rule-based methods for G4 prediction. We proceed by reviewing the major machine-learning and deep-neural-network applications to nucleic-acid G4 datasets and report a novel comparison between them. Next, we present the interpretability techniques used on the trained neural networks to learn key molecular principles underlying nucleic-acid G4 folding. As a new result, we calculate the overlap between measured DNA and RNA G4s and compare the performance of DNA- and RNA-G4 predictors on RNA- and DNA-G4 datasets, respectively, to demonstrate the potential of transfer learning from DNA G4s to RNA G4s. Last, we conclude with open questions in the field of nucleic-acid G4 prediction and computational modeling.

Liquid-liquid phase separation (LLPS) of proteins and nucleic acids underlies the formation of biomolecular condensates in cell. Dysregulation of protein LLPS is closely implicated in a range of intractable diseases. A variety of tools for predicting phase-separating proteins (PSPs) have been developed with the increasing experimental data accumulated and several related databases released. Comparing their performance directly can be challenging due to they were built on different algorithms and datasets. In this study, we evaluate eleven available PSPs predictors using negative testing datasets, including folded proteins, the human proteome, and non-PSPs under near physiological conditions, based on our recently updated LLPSDB v2.0 database. Our results show that the new generation predictors FuzDrop, DeePhase and PSPredictor perform better on folded proteins as a negative test set, while LLPhyScore outperforms other tools on the human proteome. However, none of the predictors could accurately identify experimentally verified non-PSPs. Furthermore, the correlation between predicted scores and experimentally measured saturation concentrations of protein A1-LCD and its mutants suggests that, these predictors could not consistently predict the protein LLPS propensity rationally. Further investigation with more diverse sequences for training, as well as considering features such as refined sequence pattern characterization that comprehensively reflects molecular physiochemical interactions, may improve the performance of PSPs prediction.

Loss of solubility usually leads to the detrimental elimination of protein function. In some cases, the protein aggregation is also required for beneficial functions. Given the duality of this phenomenon, it remains a fundamental question how natural selection controls the aggregation. The exponential growth of genomic sequence data and recent progress with in silico predictors of the aggregation allows approaching this problem by a large-scale bioinformatics analysis. Most of the aggregation-prone regions are hidden within the 3D structure, rendering them inaccessible for the intermolecular interactions responsible for aggregation. Thus, the most realistic census of the aggregation-prone regions requires crossing aggregation prediction with information about the location of the natively unfolded regions. This allows us to detect so-called 'exposed aggregation-prone regions' (EARs). Here, we analyzed the occurrence and distribution of the EARs in 76 reference proteomes from the three kingdoms of life. For this purpose, we used a bioinformatics pipeline, which provides a consensual result based on several predictors of aggregation. Our analysis revealed a number of new statistically significant correlations about the presence of EARs in different organisms, their dependence on protein length, cellular localizations, co-occurrence with short linear motifs and the level of protein expression. We also obtained a list of proteins with the conserved aggregation-prone sequences for further experimental tests. Insights gained from this work led to a deeper understanding of the relationship between protein evolution and aggregation.

Major histocompatibility complex (MHC)-peptide binding is a critical step in enabling a peptide to serve as an antigen for T-cell recognition. Accurate prediction of this binding can facilitate various applications in immunotherapy. While many existing methods offer good predictive power for the binding affinity of a peptide to a specific MHC, few models attempt to infer the binding threshold that distinguishes binding sequences. These models often rely on experience-based ad hoc criteria, such as 500 or 1000nM. However, different MHCs may have different binding thresholds. As such, there is a need for an automatic, data-driven method to determine an accurate binding threshold. In this study, we proposed a Bayesian model that jointly infers core locations (binding sites), the binding affinity and the binding threshold. Our model provided the posterior distribution of the binding threshold, enabling accurate determination of an appropriate threshold for each MHC. To evaluate the performance of our method under different scenarios, we conducted simulation studies with varying dominant levels of motif distributions and proportions of random sequences. These simulation studies showed desirable estimation accuracy and robustness of our model. Additionally, when applied to real data, our results outperformed commonly used thresholds.

The rapid growth of omics-based data has revolutionized biomedical research and precision medicine, allowing machine learning models to be developed for cutting-edge performance. However, despite the wealth of high-throughput data available, the performance of these models is hindered by the lack of sufficient training data, particularly in clinical research (in vivo experiments). As a result, translating this knowledge into clinical practice, such as predicting drug responses, remains a challenging task. Transfer learning is a promising tool that bridges the gap between data domains by transferring knowledge from the source to the target domain. Researchers have proposed transfer learning to predict clinical outcomes by leveraging pre-clinical data (mouse, zebrafish), highlighting its vast potential. In this work, we present a comprehensive literature review of deep transfer learning methods for health informatics and clinical decision-making, focusing on high-throughput molecular data. Previous reviews mostly covered image-based transfer learning works, while we present a more detailed analysis of transfer learning papers. Furthermore, we evaluated original studies based on different evaluation settings across cross-validations, data splits and model architectures. The result shows that those transfer learning methods have great potential; high-throughput sequencing data and state-of-the-art deep learning models lead to significant insights and conclusions. Additionally, we explored various datasets in transfer learning papers with statistics and visualization.

The importance of regulatory features in health and disease is increasing, making it crucial to identify the hallmarks of these features. Self-attention networks (SAN) have given rise to numerous models for the prediction of complex phenomena. But the potential of SANs in biological models was limited because of high memory requirement proportional to input token length and lack of interpretability of self-attention scores. To overcome these constraints, we propose a deep learning model named Interpretable Self-Attention Network for REGulatory interactions (ISANREG) that combines both block self-attention and attention-attribution mechanisms. This model predicts transcription factor-bound motif instances and DNA-mediated TF-TF interactions using self-attention attribution scores derived from the network, overcoming the limitations of previous deep learning models. ISANREG will serve as a framework for other biological models in interpreting the contribution of the input with single-nucleotide resolution.

Pathogen detection from biological and environmental samples is important for global disease control. Despite advances in pathogen detection using deep learning, current algorithms have limitations in processing long genomic sequences. Through the deep cross-fusion of cross, residual and deep neural networks, we developed DCiPatho for accurate pathogen detection based on the integrated frequency features of 3-to-7 k-mers. Compared with the existing state-of-the-art algorithms, DCiPatho can be used to accurately identify distinct pathogenic bacteria infecting humans, animals and plants. We evaluated DCiPatho on both learned and unlearned pathogen species using both genomics and metagenomics datasets. DCiPatho is an effective tool for the genomic-scale identification of pathogens by integrating the frequency of k-mers into deep cross-fusion networks. The source code is publicly available at https://github.com/LorMeBioAI/DCiPatho.

Human prescription drug labeling contains a summary of the essential scientific information needed for the safe and effective use of the drug and includes the Prescribing Information, FDA-approved patient labeling (Medication Guides, Patient Package Inserts and/or Instructions for Use), and/or carton and container labeling. Drug labeling contains critical information about drug products, such as pharmacokinetics and adverse events. Automatic information extraction from drug labels may facilitate finding the adverse reaction of the drugs or finding the interaction of one drug with another drug. Natural language processing (NLP) techniques, especially recently developed Bidirectional Encoder Representations from Transformers (BERT), have exhibited exceptional merits in text-based information extraction. A common paradigm in training BERT is to pretrain the model on large unlabeled generic language corpora, so that the model learns the distribution of the words in the language, and then fine-tune on a downstream task. In this paper, first, we show the uniqueness of language used in drug labels, which therefore cannot be optimally handled by other BERT models. Then, we present the developed PharmBERT, which is a BERT model specifically pretrained on the drug labels (publicly available at Hugging Face). We demonstrate that our model outperforms the vanilla BERT, ClinicalBERT and BioBERT in multiple NLP tasks in the drug label domain. Moreover, how the domain-specific pretraining has contributed to the superior performance of PharmBERT is demonstrated by analyzing different layers of PharmBERT, and more insight into how it understands different linguistic aspects of the data is gained.

As the volume of protein sequence and structure data grows rapidly, the functions of the overwhelming majority of proteins cannot be experimentally determined. Automated annotation of protein function at a large scale is becoming increasingly important. Existing computational prediction methods are typically based on expanding the relatively small number of experimentally determined functions to large collections of proteins with various clues, including sequence homology, protein-protein interaction, gene co-expression, etc. Although there has been some progress in protein function prediction in recent years, the development of accurate and reliable solutions still has a long way to go. Here we exploit AlphaFold predicted three-dimensional structural information, together with other non-structural clues, to develop a large-scale approach termed PredGO to annotate Gene Ontology (GO) functions for proteins. We use a pre-trained language model, geometric vector perceptrons and attention mechanisms to extract heterogeneous features of proteins and fuse these features for function prediction. The computational results demonstrate that the proposed method outperforms other state-of-the-art approaches for predicting GO functions of proteins in terms of both coverage and accuracy. The improvement of coverage is because the number of structures predicted by AlphaFold is greatly increased, and on the other hand, PredGO can extensively use non-structural information for functional prediction. Moreover, we show that over 205 000 ($\sim $100%) entries in UniProt for human are annotated by PredGO, over 186 000 ($\sim $90%) of which are based on predicted structure. The webserver and database are available at http://predgo.denglab.org/.

Missing values (MVs) can adversely impact data analysis and machine-learning model development. We propose a novel mixed-model method for missing value imputation (MVI). This method, ProJect (short for Protein inJection), is a powerful and meaningful improvement over existing MVI methods such as Bayesian principal component analysis (PCA), probabilistic PCA, local least squares and quantile regression imputation of left-censored data. We rigorously tested ProJect on various high-throughput data types, including genomics and mass spectrometry (MS)-based proteomics. Specifically, we utilized renal cancer (RC) data acquired using DIA-SWATH, ovarian cancer (OC) data acquired using DIA-MS, bladder (BladderBatch) and glioblastoma (GBM) microarray gene expression dataset. Our results demonstrate that ProJect consistently performs better than other referenced MVI methods. It achieves the lowest normalized root mean square error (on average, scoring 45.92% less error in RC_C, 27.37% in RC_full, 29.22% in OC, 23.65% in BladderBatch and 20.20% in GBM relative to the closest competing method) and the Procrustes sum of squared error (Procrustes SS) (exhibits 79.71% less error in RC_C, 38.36% in RC full, 18.13% in OC, 74.74% in BladderBatch and 30.79% in GBM compared to the next best method). ProJect also leads with the highest correlation coefficient among all types of MV combinations (0.64% higher in RC_C, 0.24% in RC full, 0.55% in OC, 0.39% in BladderBatch and 0.27% in GBM versus the second-best performing method). ProJect's key strength is its ability to handle different types of MVs commonly found in real-world data. Unlike most MVI methods that are designed to handle only one type of MV, ProJect employs a decision-making algorithm that first determines if an MV is missing at random or missing not at random. It then employs targeted imputation strategies for each MV type, resulting in more accurate and reliable imputation outcomes. An R implementation of ProJect is available at https://github.com/miaomiao6606/ProJect.

AlphaFold-Multimer has greatly improved the protein complex structure prediction, but its accuracy also depends on the quality of the multiple sequence alignment (MSA) formed by the interacting homologs (i.e. interologs) of the complex under prediction. Here we propose a novel method, ESMPair, that can identify interologs of a complex using protein language models. We show that ESMPair can generate better interologs than the default MSA generation method in AlphaFold-Multimer. Our method results in better complex structure prediction than AlphaFold-Multimer by a large margin (+10.7% in terms of the Top-5 best DockQ), especially when the predicted complex structures have low confidence. We further show that by combining several MSA generation methods, we may yield even better complex structure prediction accuracy than Alphafold-Multimer (+22% in terms of the Top-5 best DockQ). By systematically analyzing the impact factors of our algorithm we find that the diversity of MSA of interologs significantly affects the prediction accuracy. Moreover, we show that ESMPair performs particularly well on complexes in eucaryotes.

Neurodegenerative diseases (NDs) usually connect with aggregation and molecular interactions of pathological proteins. The integration of accumulative data from clinical and biomedical research will allow for the excavation of pathological proteins and related interactors. It is also important to systematically study their interacting proteins in order to find more related proteins and potential therapeutic targets. Understanding binding regions in protein interactions will help functional proteomics and provide an alternative method for predicting novel interactions. This study integrated data from biomedical research to achieve systematic mining and analysis of pathogenic proteins and their interaction network. A workflow has been built as a solution for the collective information of proteins involved in NDs, related protein-protein interactions (PPIs) and interactive visualizations. It also included protein isoforms and mapped them in a disease-related PPI network to illuminate the impact of alternative splicing on protein binding. The interacting proteins enriched by diseases and biological processes (BPs) revealed possible regulatory modules. A high-resolution network with structural affinity information was generated. Finally, Neurodegenerative Disease Atlas (NDAtlas) was constructed with an interactive and intuitive view of protein docking with 3D molecular graphics beyond the traditional 2D network. NDAtlas is available at http://bis.zju.edu.cn/ndatlas.

Clustering methods have been widely used in single-cell RNA-seq data for investigating tumor heterogeneity. Since traditional clustering methods fail to capture the high-dimension methods, deep clustering methods have drawn increasing attention these years due to their promising strengths on the task. However, existing methods consider either the attribute information of each cell or the structure information between different cells. In other words, they cannot sufficiently make use of all of this information simultaneously. To this end, we propose a novel single-cell deep fusion clustering model, which contains two modules, i.e. an attributed feature clustering module and a structure-attention feature clustering module. More concretely, two elegantly designed autoencoders are built to handle both features regardless of their data types. Experiments have demonstrated the validity of the proposed approach, showing that it is efficient to fuse attributes, structure, and attention information on single-cell RNA-seq data. This work will be further beneficial for investigating cell subpopulations and tumor microenvironment. The Python implementation of our work is now freely available at https://github.com/DayuHuu/scDFC.

The prediction of drug-drug interactions (DDIs) is essential for the development and repositioning of new drugs. Meanwhile, they play a vital role in the fields of biopharmaceuticals, disease diagnosis and pharmacological treatment. This article proposes a new method called DBGRU-SE for predicting DDIs. Firstly, FP3 fingerprints, MACCS fingerprints, Pubchem fingerprints and 1D and 2D molecular descriptors are used to extract the feature information of the drugs. Secondly, Group Lasso is used to remove redundant features. Then, SMOTE-ENN is applied to balance the data to obtain the best feature vectors. Finally, the best feature vectors are fed into the classifier combining BiGRU and squeeze-and-excitation (SE) attention mechanisms to predict DDIs. After applying five-fold cross-validation, The ACC values of DBGRU-SE model on the two datasets are 97.51 and 94.98%, and the AUC are 99.60 and 98.85%, respectively. The results showed that DBGRU-SE had good predictive performance for drug-drug interactions.

Hi-C is a widely applied chromosome conformation capture (3C)-based technique, which has produced a large number of genomic contact maps with high sequencing depths for a wide range of cell types, enabling comprehensive analyses of the relationships between biological functionalities (e.g. gene regulation and expression) and the three-dimensional genome structure. Comparative analyses play significant roles in Hi-C data studies, which are designed to make comparisons between Hi-C contact maps, thus evaluating the consistency of replicate Hi-C experiments (i.e. reproducibility measurement) and detecting statistically differential interacting regions with biological significance (i.e. differential chromatin interaction detection). However, due to the complex and hierarchical nature of Hi-C contact maps, it remains challenging to conduct systematic and reliable comparative analyses of Hi-C data. Here, we proposed sslHiC, a contrastive self-supervised representation learning framework, for precisely modeling the multi-level features of chromosome conformation and automatically producing informative feature embeddings for genomic loci and their interactions to facilitate comparative analyses of Hi-C contact maps. Comprehensive computational experiments on both simulated and real datasets demonstrated that our method consistently outperformed the state-of-the-art baseline methods in providing reliable measurements of reproducibility and detecting differential interactions with biological meanings.

Three-dimensional (3D) genome architecture is characterized by multi-scale patterns and plays an essential role in gene regulation. Chromatin conformation capturing experiments have revealed many properties underlying 3D genome architecture, such as the compartmentalization of chromatin based on transcriptional states. However, they are complex, costly and time consuming, and therefore only a limited number of cell types have been examined using these techniques. Increasing effort is being directed towards deriving computational methods that can predict chromatin conformation and associated structures. Here we present DNA-delay differential analysis (DDA), a purely sequence-based method based on chaos theory to predict genome-wide A and B compartments. We show that DNA-DDA models derived from a 20 Mb sequence are sufficient to predict genome wide compartmentalization at the scale of 100 kb in four different cell types. Although this is a proof-of-concept study, our method shows promise in elucidating the mechanisms responsible for genome folding as well as modeling the impact of genetic variation on 3D genome architecture and the processes regulated thereby.

The three-dimensional structure of RNA molecules plays a critical role in a wide range of cellular processes encompassing functions from riboswitches to epigenetic regulation. These RNA structures are incredibly dynamic and can indeed be described aptly as an ensemble of structures that shifts in distribution depending on different cellular conditions. Thus, the computational prediction of RNA structure poses a unique challenge, even as computational protein folding has seen great advances. In this review, we focus on a variety of machine learning-based methods that have been developed to predict RNA molecules' secondary structure, as well as more complex tertiary structures. We survey commonly used modeling strategies, and how many are inspired by or incorporate thermodynamic principles. We discuss the shortcomings that various design decisions entail and propose future directions that could build off these methods to yield more robust, accurate RNA structure predictions.

Allergies have become an emerging public health problem worldwide. The most effective way to prevent allergies is to find the causative allergen at the source and avoid re-exposure. However, most of the current computational methods used to identify allergens were based on homology or conventional machine learning methods, which were inefficient and still had room to be improved for the detection of allergens with low homology. In addition, few methods based on deep learning were reported, although deep learning has been successfully applied to several tasks in protein sequence analysis. In the present work, a deep neural network-based model, called DeepAlgPro, was proposed to identify allergens. We showed its great accuracy and applicability to large-scale forecasts by comparing it to other available tools. Additionally, we used ablation experiments to demonstrate the critical importance of the convolutional module in our model. Moreover, further analyses showed that epitope features contributed to model decision-making, thus improving the model's interpretability. Finally, we found that DeepAlgPro was capable of detecting potential new allergens. Overall, DeepAlgPro can serve as powerful software for identifying allergens.

Recently a biochemistry experiment named methyl-3C was developed to simultaneously capture the chromosomal conformations and DNA methylation levels on individual single cells. However, the number of data sets generated from this experiment is still small in the scientific community compared with the greater amount of single-cell Hi-C data generated from separate single cells. Therefore, a computational tool to predict single-cell methylation levels based on single-cell Hi-C data on the same individual cells is needed. We developed a graph transformer named scHiMe to accurately predict the base-pair-specific (bp-specific) methylation levels based on both single-cell Hi-C data and DNA nucleotide sequences. We benchmarked scHiMe for predicting the bp-specific methylation levels on all of the promoters of the human genome, all of the promoter regions together with the corresponding first exon and intron regions, and random regions on the whole genome. Our evaluation showed a high consistency between the predicted and methyl-3C-detected methylation levels. Moreover, the predicted DNA methylation levels resulted in accurate classifications of cells into different cell types, which indicated that our algorithm successfully captured the cell-to-cell variability in the single-cell Hi-C data. scHiMe is freely available at http://dna.cs.miami.edu/scHiMe/.

In modern precision medicine, it is an important research topic to predict cancer drug response. Due to incomplete chemical structures and complex gene features, however, it is an ongoing work to design efficient data-driven methods for predicting drug response. Moreover, since the clinical data cannot be easily obtained all at once, the data-driven methods may require relearning when new data are available, resulting in increased time consumption and cost. To address these issues, an incremental broad Transformer network (iBT-Net) is proposed for cancer drug response prediction. Different from the gene expression features learning from cancer cell lines, structural features are further extracted from drugs by Transformer. Broad learning system is then designed to integrate the learned gene features and structural features of drugs to predict the response. With the capability of incremental learning, the proposed method can further use new data to improve its prediction performance without retraining totally. Experiments and comparison studies demonstrate the effectiveness and superiority of iBT-Net under different experimental configurations and continuous data learning.

Nucleic acid-binding proteins are proteins that interact with DNA and RNA to regulate gene expression and transcriptional control. The pathogenesis of many human diseases is related to abnormal gene expression. Therefore, recognizing nucleic acid-binding proteins accurately and efficiently has important implications for disease research. To address this question, some scientists have proposed the method of using sequence information to identify nucleic acid-binding proteins. However, different types of nucleic acid-binding proteins have different subfunctions, and these methods ignore their internal differences, so the performance of the predictor can be further improved. In this study, we proposed a new method, called iDRPro-SC, to predict the type of nucleic acid-binding proteins based on the sequence information. iDRPro-SC considers the internal differences of nucleic acid-binding proteins and combines their subfunctions to build a complete dataset. Additionally, we used an ensemble learning to characterize and predict nucleic acid-binding proteins. The results of the test dataset showed that iDRPro-SC achieved the best prediction performance and was superior to the other existing nucleic acid-binding protein prediction methods. We have established a web server that can be accessed online: http://bliulab.net/iDRPro-SC.

Drug-drug interactions (DDI) may lead to adverse reactions in human body and accurate prediction of DDI can mitigate the medical risk. Currently, most of computer-aided DDI prediction methods construct models based on drug-associated features or DDI network, ignoring the potential information contained in drug-related biological entities such as targets and genes. Besides, existing DDI network-based models could not make effective predictions for drugs without any known DDI records. To address the above limitations, we propose an attention-based cross domain graph neural network (ACDGNN) for DDI prediction, which considers the drug-related different entities and propagate information through cross domain operation. Different from the existing methods, ACDGNN not only considers rich information contained in drug-related biomedical entities in biological heterogeneous network, but also adopts cross-domain transformation to eliminate heterogeneity between different types of entities. ACDGNN can be used in the prediction of DDIs in both transductive and inductive setting. By conducting experiments on real-world dataset, we compare the performance of ACDGNN with several state-of-the-art methods. The experimental results show that ACDGNN can effectively predict DDIs and outperform the comparison models.

Determining cancer subtypes and estimating patient prognosis are crucial for cancer research. The massive amount of multi-omics data generated by high-throughput sequencing technology is an important resource for cancer prognosis. Deep learning methods can integrate such data to accurately identify more cancer subtypes. We propose a prognostic model based on a convolutional autoencoder (ProgCAE) that can predict cancer subtypes associated with survival using multi-omics data. We demonstrated that ProgCAE predicted subtypes of 12 cancer types with significant survival differences and outperformed traditional statistical methods for predicting the survival of most patients with cancer. Supervised classifiers can be constructed based on subtypes predicted by robust ProgCAE.

Circular RNA (circRNA) is closely associated with human diseases. Accordingly, identifying the associations between human diseases and circRNA can help in disease prevention, diagnosis and treatment. Traditional methods are time consuming and laborious. Meanwhile, computational models can effectively predict potential circRNA-disease associations (CDAs), but are restricted by limited data, resulting in data with high dimension and imbalance. In this study, we propose a model based on automatically selected meta-path and contrastive learning, called the MPCLCDA model. First, the model constructs a new heterogeneous network based on circRNA similarity, disease similarity and known association, via automatically selected meta-path and obtains the low-dimensional fusion features of nodes via graph convolutional networks. Then, contrastive learning is used to optimize the fusion features further, and obtain the node features that make the distinction between positive and negative samples more evident. Finally, circRNA-disease scores are predicted through a multilayer perceptron. The proposed method is compared with advanced methods on four datasets. The average area under the receiver operating characteristic curve, area under the precision-recall curve and F1 score under 5-fold cross-validation reached 0.9752, 0.9831 and 0.9745, respectively. Simultaneously, case studies on human diseases further prove the predictive ability and application value of this method.

Adverse drug-drug interactions (DDIs) have become an increasingly serious problem in the medical and health system. Recently, the effective application of deep learning and biomedical knowledge graphs (KGs) have improved the DDI prediction performance of computational models. However, the problems of feature redundancy and KG noise also arise, bringing new challenges for researchers. To overcome these challenges, we proposed a Multi-Channel Feature Fusion model for multi-typed DDI prediction (MCFF-MTDDI). Specifically, we first extracted drug chemical structure features, drug pairs' extra label features, and KG features of drugs. Then, these different features were effectively fused by a multi-channel feature fusion module. Finally, multi-typed DDIs were predicted through the fully connected neural network. To our knowledge, we are the first to integrate the extra label information into KG-based multi-typed DDI prediction; besides, we innovatively proposed a novel KG feature learning method and a State Encoder to obtain target drug pairs' KG-based features which contained more abundant and more key drug-related KG information with less noise; furthermore, a Gated Recurrent Unit-based multi-channel feature fusion module was proposed in an innovative way to yield more comprehensive feature information about drug pairs, effectively alleviating the problem of feature redundancy. We experimented with four datasets in the multi-class and the multi-label prediction tasks to comprehensively evaluate the performance of MCFF-MTDDI for predicting interactions of known-known drugs, known-new drugs and new-new drugs. In addition, we further conducted ablation studies and case studies. All the results fully demonstrated the effectiveness of MCFF-MTDDI.

MicroRNAs (miRNAs) have significant implications in diverse human diseases and have proven to be effectively targeted by small molecules (SMs) for therapeutic interventions. However, current SM-miRNA association prediction models do not adequately capture SM/miRNA similarity. Matrix completion is an effective method for association prediction, but existing models use nuclear norm instead of rank function, which has some drawbacks. Therefore, we proposed a new approach for predicting SM-miRNA associations by utilizing the truncated schatten p-norm (TSPN). First, the SM/miRNA similarity was preprocessed by incorporating the Gaussian interaction profile kernel similarity method. This identified more SM/miRNA similarities and significantly improved the SM-miRNA prediction accuracy. Next, we constructed a heterogeneous SM-miRNA network by combining biological information from three matrices and represented the network with its adjacency matrix. Finally, we constructed the prediction model by minimizing the truncated schatten p-norm of this adjacency matrix and we developed an efficient iterative algorithmic framework to solve the model. In this framework, we also used a weighted singular value shrinkage algorithm to avoid the problem of excessive singular value shrinkage. The truncated schatten p-norm approximates the rank function more closely than the nuclear norm, so the predictions are more accurate. We performed four different cross-validation experiments on two separate datasets, and TSPN outperformed various most advanced methods. In addition, public literature confirms a large number of predictive associations of TSPN in four case studies. Therefore, TSPN is a reliable model for SM-miRNA association prediction.

The T-cell receptor (TCR) repertoire is highly diverse among the population and plays an essential role in initiating multiple immune processes. TCR sequencing (TCR-seq) has been developed to profile the T cell repertoire. Similar to other high-throughput experiments, contamination can happen during several steps of TCR-seq, including sample collection, preparation and sequencing. Such contamination creates artifacts in the data, leading to inaccurate or even biased results. Most existing methods assume 'clean' TCR-seq data as the starting point with no ability to handle data contamination. Here, we develop a novel statistical model to systematically detect and remove contamination in TCR-seq data. We summarize the observed contamination into two sources, pairwise and cross-cohort. For both sources, we provide visualizations and summary statistics to help users assess the severity of the contamination. Incorporating prior information from 14 existing TCR-seq datasets with minimum contamination, we develop a straightforward Bayesian model to statistically identify contaminated samples. We further provide strategies for removing the impacted sequences to allow for downstream analysis, thus avoiding any need to repeat experiments. Our proposed model shows robustness in contamination detection compared with a few off-the-shelf detection methods in simulation studies. We illustrate the use of our proposed method on two TCR-seq datasets generated locally.

Enzymatic reactions are crucial to explore the mechanistic function of metabolites and proteins in cellular processes and to understand the etiology of diseases. The increasing number of interconnected metabolic reactions allows the development of in silico deep learning-based methods to discover new enzymatic reaction links between metabolites and proteins to further expand the landscape of existing metabolite-protein interactome. Computational approaches to predict the enzymatic reaction link by metabolite-protein interaction (MPI) prediction are still very limited. In this study, we developed a Variational Graph Autoencoders (VGAE)-based framework to predict MPI in genome-scale heterogeneous enzymatic reaction networks across ten organisms. By incorporating molecular features of metabolites and proteins as well as neighboring information in the MPI networks, our MPI-VGAE predictor achieved the best predictive performance compared to other machine learning methods. Moreover, when applying the MPI-VGAE framework to reconstruct hundreds of metabolic pathways, functional enzymatic reaction networks and a metabolite-metabolite interaction network, our method showed the most robust performance among all scenarios. To the best of our knowledge, this is the first MPI predictor by VGAE for enzymatic reaction link prediction. Furthermore, we implemented the MPI-VGAE framework to reconstruct the disease-specific MPI network based on the disrupted metabolites and proteins in Alzheimer's disease and colorectal cancer, respectively. A substantial number of novel enzymatic reaction links were identified. We further validated and explored the interactions of these enzymatic reactions using molecular docking. These results highlight the potential of the MPI-VGAE framework for the discovery of novel disease-related enzymatic reactions and facilitate the study of the disrupted metabolisms in diseases.

Single-cell omics technologies have made it possible to analyze the individual cells within a biological sample, providing a more detailed understanding of biological systems. Accurately determining the cell type of each cell is a crucial goal in single-cell RNA-seq (scRNA-seq) analysis. Apart from overcoming the batch effects arising from various factors, single-cell annotation methods also face the challenge of effectively processing large-scale datasets. With the availability of an increase in the scRNA-seq datasets, integrating multiple datasets and addressing batch effects originating from diverse sources are also challenges in cell-type annotation. In this work, to overcome the challenges, we developed a supervised method called CIForm based on the Transformer for cell-type annotation of large-scale scRNA-seq data. To assess the effectiveness and robustness of CIForm, we have compared it with some leading tools on benchmark datasets. Through the systematic comparisons under various cell-type annotation scenarios, we exhibit that the effectiveness of CIForm is particularly pronounced in cell-type annotation. The source code and data are available at https://github.com/zhanglab-wbgcas/CIForm.

Deoxyribonucleic acid (DNA) methylation (DNAm) is an important epigenetic mechanism that plays a role in chromatin structure and transcriptional regulation. Elucidating the relationship between DNAm and gene expression is of great importance for understanding its role in transcriptional regulation. The conventional approach is to construct machine-learning-based methods to predict gene expression based on mean methylation signals in promoter regions. However, this type of strategy only explains about 25% of gene expression variation, and hence is inadequate in elucidating the relationship between DNAm and transcriptional activity. In addition, using mean methylation as input features neglects the heterogeneity of cell populations that can be reflected by DNAm haplotypes. We here developed TRAmaHap, a novel deep-learning framework that predicts gene expression by utilizing the characteristics of DNAm haplotypes in proximal promoters and distal enhancers. Using benchmark data of human and mouse normal tissues, TRAmHap shows much higher accuracy than existing machine-learning based methods, by explaining 60~80% of gene expression variation across tissue types and disease conditions. Our model demonstrated that gene expression can be accurately predicted by DNAm patterns in promoters and long-range enhancers as far as 25 kb away from transcription start site, especially in the presence of intra-gene chromatin interactions.

The rational design of chemical entities with desired properties for a specific target is a long-standing challenge in drug design. Generative neural networks have emerged as a powerful approach to sample novel molecules with specific properties, termed as inverse drug design. However, generating molecules with biological activity against certain targets and predefined drug properties still remains challenging. Here, we propose a conditional molecular generation net (CMGN), the backbone of which is a bidirectional and autoregressive transformer. CMGN applies large-scale pretraining for molecular understanding and navigates the chemical space for specified targets by fine-tuning with corresponding datasets. Additionally, fragments and properties were trained to recover molecules to learn the structure-properties relationships. Our model crisscrosses the chemical space for specific targets and properties that control fragment-growth processes. Case studies demonstrated the advantages and utility of our model in fragment-to-lead processes and multi-objective lead optimization. The results presented in this paper illustrate that CMGN has the potential to accelerate the drug discovery process.

Determining intrinsically disordered regions of proteins is essential for elucidating protein biological functions and the mechanisms of their associated diseases. As the gap between the number of experimentally determined protein structures and the number of protein sequences continues to grow exponentially, there is a need for developing an accurate and computationally efficient disorder predictor. However, current single-sequence-based methods are of low accuracy, while evolutionary profile-based methods are computationally intensive. Here, we proposed a fast and accurate protein disorder predictor LMDisorder that employed embedding generated by unsupervised pretrained language models as features. We showed that LMDisorder performs best in all single-sequence-based methods and is comparable or better than another language-model-based technique in four independent test sets, respectively. Furthermore, LMDisorder showed equivalent or even better performance than the state-of-the-art profile-based technique SPOT-Disorder2. In addition, the high computation efficiency of LMDisorder enabled proteome-scale analysis of human, showing that proteins with high predicted disorder content were associated with specific biological functions. The datasets, the source codes, and the trained model are available at https://github.com/biomed-AI/LMDisorder.

Although sequencing-based high-throughput chromatin interaction data are widely used to uncover genome-wide three-dimensional chromatin architecture, their sparseness and high signal-noise-ratio greatly restrict the precision of the obtained structural elements. To improve data quality, we here present iEnhance (chromatin interaction data resolution enhancement), a multi-scale spatial projection and encoding network, to predict high-resolution chromatin interaction matrices from low-resolution and noisy input data. Specifically, iEnhance projects the input data into matrix spaces to extract multi-scale global and local feature sets, then hierarchically fused these features by attention mechanism. After that, dense channel encoding and residual channel decoding are used to effectively infer robust chromatin interaction maps. iEnhance outperforms state-of-the-art Hi-C resolution enhancement tools in both visual and quantitative evaluation. Comprehensive analysis shows that unlike other tools, iEnhance can recover both short-range structural elements and long-range interaction patterns precisely. More importantly, iEnhance can be transferred to data enhancement of other tissues or cell lines of unknown resolution. Furthermore, iEnhance performs robustly in enhancement of diverse chromatin interaction data including those from single-cell Hi-C and Micro-C experiments.

Computational protein design has been demonstrated to be the most powerful tool in the last few years among protein designing and repacking tasks. In practice, these two tasks are strongly related but often treated separately. Besides, state-of-the-art deep-learning-based methods cannot provide interpretability from an energy perspective, affecting the accuracy of the design. Here we propose a new systematic approach, including both a posterior probability and a joint probability parts, to solve the two essential questions once for all. This approach takes the physicochemical property of amino acids into consideration and uses the joint probability model to ensure the convergence between structure and amino acid type. Our results demonstrated that this method could generate feasible, high-confidence sequences with low-energy side conformations. The designed sequences can fold into target structures with high confidence and maintain relatively stable biochemical properties. The side chain conformation has a significantly lower energy landscape without delegating to a rotamer library or performing the expensive conformational searches. Overall, we propose an end-to-end method that combines the advantages of both deep learning and energy-based methods. The design results of this model demonstrate high efficiency, and precision, as well as a low energy state and good interpretability.

Spatially resolved transcriptomics (SRT) enable the comprehensive characterization of transcriptomic profiles in the context of tissue microenvironments. Unveiling spatial transcriptional heterogeneity needs to effectively incorporate spatial information accounting for the substantial spatial correlation of expression measurements. Here, we develop a computational method, SpaSRL (spatially aware self-representation learning), which flexibly enhances and decodes spatial transcriptional signals to simultaneously achieve spatial domain detection and spatial functional genes identification. This novel tunable spatially aware strategy of SpaSRL not only balances spatial and transcriptional coherence for the two tasks, but also can transfer spatial correlation constraint between them based on a unified model. In addition, this joint analysis by SpaSRL deciphers accurate and fine-grained tissue structures and ensures the effective extraction of biologically informative genes underlying spatial architecture. We verified the superiority of SpaSRL on spatial domain detection, spatial functional genes identification and data denoising using multiple SRT datasets obtained by different platforms and tissue sections. Our results illustrate SpaSRL's utility in flexible integration of spatial information and novel discovery of biological insights from spatial transcriptomic datasets.

Identifying topologically associating domains (TADs), which are considered as the basic units of chromosome structure and function, can facilitate the exploration of the 3D-structure of chromosomes. Methods have been proposed to identify TADs by detecting the boundaries of TADs or identifying the closely interacted regions as TADs, while the possible inner structure of TADs is seldom investigated. In this study, we assume that a TAD is composed of a core and its surrounding attachments, and propose a method, named CATAD, to identify TADs based on the core-attachment structure model. In CATAD, the cores of TADs are identified based on the local density and cosine similarity, and the surrounding attachments are determined based on boundary insulation. CATAD was applied to the Hi-C data of two human cell lines and two mouse cell lines, and the results show that the boundaries of TADs identified by CATAD are significantly enriched by structural proteins, histone modifications, transcription start sites and enzymes. Furthermore, CATAD outperforms other methods in many cases, in terms of the average peak, boundary tagged ratio and fold change. In addition, CATAD is robust and rarely affected by the different resolutions of Hi-C matrices. Conclusively, identifying TADs based on the core-attachment structure is useful, which may inspire researchers to explore TADs from the angles of possible spatial structures and formation process.

The volume of ribonucleic acid (RNA)-seq data has increased exponentially, providing numerous new insights into various biological processes. However, due to significant practical challenges, such as data heterogeneity, it is still difficult to ensure the quality of these data when integrated. Although some quality control methods have been developed, sample consistency is rarely considered and these methods are susceptible to artificial factors. Here, we developed MassiveQC, an unsupervised machine learning-based approach, to automatically download and filter large-scale high-throughput data. In addition to the read quality used in other tools, MassiveQC also uses the alignment and expression quality as model features. Meanwhile, it is user-friendly since the cutoff is generated from self-reporting and is applicable to multimodal data. To explore its value, we applied MassiveQC to Drosophila RNA-seq data and generated a comprehensive transcriptome atlas across 28 tissues from embryogenesis to adulthood. We systematically characterized fly gene expression dynamics and found that genes with high expression dynamics were likely to be evolutionarily young and expressed at late developmental stages, exhibiting high nonsynonymous substitution rates and low phenotypic severity, and they were involved in simple regulatory programs. We also discovered that human and Drosophila had strong positive correlations in gene expression in orthologous organs, revealing the great potential of the Drosophila system for studying human development and disease.

Single-cell RNA sequencing (scRNA-seq) measures transcriptome-wide gene expression at single-cell resolution. Clustering analysis of scRNA-seq data enables researchers to characterize cell types and states, shedding new light on cell-to-cell heterogeneity in complex tissues. Recently, self-supervised contrastive learning has become a prominent technique for underlying feature representation learning. However, for the noisy, high-dimensional and sparse scRNA-seq data, existing methods still encounter difficulties in capturing the intrinsic patterns and structures of cells, and seldom utilize prior knowledge, resulting in clusters that mismatch with the real situation. To this end, we propose scDECL, a novel deep enhanced constraint clustering algorithm for scRNA-seq data analysis based on contrastive learning and pairwise constraints. Specifically, based on interpolated contrastive learning, a pre-training model is trained to learn the feature embedding, and then perform clustering according to the constructed enhanced pairwise constraint. In the pre-training stage, a mixup data augmentation strategy and interpolation loss is introduced to improve the diversity of the dataset and the robustness of the model. In the clustering stage, the prior information is converted into enhanced pairwise constraints to guide the clustering. To validate the performance of scDECL, we compare it with six state-of-the-art algorithms on six real scRNA-seq datasets. The experimental results demonstrate the proposed algorithm outperforms the six competing methods. In addition, the ablation studies on each module of the algorithm indicate that these modules are complementary to each other and effective in improving the performance of the proposed algorithm. Our method scDECL is implemented in Python using the Pytorch machine-learning library, and it is freely available at https://github.com/DBLABDHU/scDECL.

A phenotype may be associated with multiple genes that interact with each other in the form of a gene module or network. How to identify these relationships is one important aspect of comparative transcriptomics. However, it is still a challenge to align gene modules associated with different phenotypes. Although several studies attempted to address this issue in different aspects, a general framework is still needed. In this study, we introduce Module Alignment of TranscripTomE (MATTE), a novel approach to analyze transcriptomics data and identify differences in a modular manner. MATTE assumes that gene interactions modulate a phenotype and models phenotype differences as gene location changes. Specifically, we first represented genes by a relative differential expression to reduce the influence of noise in omics data. Meanwhile, clustering and aligning are combined to depict gene differences in a modular way robustly. The results show that MATTE outperformed state-of-the-art methods in identifying differentially expressed genes under noise in gene expression. In particular, MATTE could also deal with single-cell ribonucleic acid-seq data to extract the best cell-type marker genes compared to other methods. Additionally, we demonstrate how MATTE supports the discovery of biologically significant genes and modules, and facilitates downstream analyses to gain insight into breast cancer. The source code of MATTE and case analysis are available at https://github.com/zjupgx/MATTE.

Protein is the most important component in organisms and plays an indispensable role in life activities. In recent years, a large number of intelligent methods have been proposed to predict protein function. These methods obtain different types of protein information, including sequence, structure and interaction network. Among them, protein sequences have gained significant attention where methods are investigated to extract the information from different views of features. However, how to fully exploit the views for effective protein sequence analysis remains a challenge. In this regard, we propose a multi-view, multi-scale and multi-attention deep neural model (MMSMA) for protein function prediction. First, MMSMA extracts multi-view features from protein sequences, including one-hot encoding features, evolutionary information features, deep semantic features and overlapping property features based on physiochemistry. Second, a specific multi-scale multi-attention deep network model (MSMA) is built for each view to realize the deep feature learning and preliminary classification. In MSMA, both multi-scale local patterns and long-range dependence from protein sequences can be captured. Third, a multi-view adaptive decision mechanism is developed to make a comprehensive decision based on the classification results of all the views. To further improve the prediction performance, an extended version of MMSMA, MMSMAPlus, is proposed to integrate homology-based protein prediction under the framework of multi-view deep neural model. Experimental results show that the MMSMAPlus has promising performance and is significantly superior to the state-of-the-art methods. The source code can be found at https://github.com/wzy-2020/MMSMAPlus.

Accurately predicting the antigen-binding specificity of adaptive immune receptors (AIRs), such as T-cell receptors (TCRs) and B-cell receptors (BCRs), is essential for discovering new immune therapies. However, the diversity of AIR chain sequences limits the accuracy of current prediction methods. This study introduces SC-AIR-BERT, a pre-trained model that learns comprehensive sequence representations of paired AIR chains to improve binding specificity prediction. SC-AIR-BERT first learns the 'language' of AIR sequences through self-supervised pre-training on a large cohort of paired AIR chains from multiple single-cell resources. The model is then fine-tuned with a multilayer perceptron head for binding specificity prediction, employing the K-mer strategy to enhance sequence representation learning. Extensive experiments demonstrate the superior AUC performance of SC-AIR-BERT compared with current methods for TCR- and BCR-binding specificity prediction.

Antimicrobial peptides (AMPs) are short peptides that play crucial roles in diverse biological processes and have various functional activities against target organisms. Due to the abuse of chemical antibiotics and microbial pathogens' increasing resistance to antibiotics, AMPs have the potential to be alternatives to antibiotics. As such, the identification of AMPs has become a widely discussed topic. A variety of computational approaches have been developed to identify AMPs based on machine learning algorithms. However, most of them are not capable of predicting the functional activities of AMPs, and those predictors that can specify activities only focus on a few of them. In this study, we first surveyed 10 predictors that can identify AMPs and their functional activities in terms of the features they employed and the algorithms they utilized. Then, we constructed comprehensive AMP datasets and proposed a new deep learning-based framework, iAMPCN (identification of AMPs based on CNNs), to identify AMPs and their related 22 functional activities. Our experiments demonstrate that iAMPCN significantly improved the prediction performance of AMPs and their corresponding functional activities based on four types of sequence features. Benchmarking experiments on the independent test datasets showed that iAMPCN outperformed a number of state-of-the-art approaches for predicting AMPs and their functional activities. Furthermore, we analyzed the amino acid preferences of different AMP activities and evaluated the model on datasets of varying sequence redundancy thresholds. To facilitate the community-wide identification of AMPs and their corresponding functional types, we have made the source codes of iAMPCN publicly available at https://github.com/joy50706/iAMPCN/tree/master. We anticipate that iAMPCN can be explored as a valuable tool for identifying potential AMPs with specific functional activities for further experimental validation.

Studies have confirmed that the occurrence of many complex diseases in the human body is closely related to the microbial community, and microbes can affect tumorigenesis and metastasis by regulating the tumor microenvironment. However, there are still large gaps in the clinical observation of the microbiota in disease. Although biological experiments are accurate in identifying disease-associated microbes, they are also time-consuming and expensive. The computational models for effective identification of diseases related microbes can shorten this process, and reduce capital and time costs. Based on this, in the paper, a model named DSAE_RF is presented to predict latent microbe-disease associations by combining multi-source features and deep learning. DSAE_RF calculates four similarities between microbes and diseases, which are then used as feature vectors for the disease-microbe pairs. Later, reliable negative samples are screened by k-means clustering, and a deep sparse autoencoder neural network is further used to extract effective features of the disease-microbe pairs. In this foundation, a random forest classifier is presented to predict the associations between microbes and diseases. To assess the performance of the model in this paper, 10-fold cross-validation is implemented on the same dataset. As a result, the AUC and AUPR of the model are 0.9448 and 0.9431, respectively. Furthermore, we also conduct a variety of experiments, including comparison of negative sample selection methods, comparison with different models and classifiers, Kolmogorov-Smirnov test and t-test, ablation experiments, robustness analysis, and case studies on Covid-19 and colorectal cancer. The results fully demonstrate the reliability and availability of our model.

The scoring models used for protein structure modeling and ranking are mainly divided into unified field and protein-specific scoring functions. Although protein structure prediction has made tremendous progress since CASP14, the modeling accuracy still cannot meet the requirements to a certain extent. Especially, accurate modeling of multi-domain and orphan proteins remains a challenge. Therefore, an accurate and efficient protein scoring model should be developed urgently to guide the protein structure folding or ranking through deep learning. In this work, we propose a protein structure global scoring model based on equivariant graph neural network (EGNN), named GraphGPSM, to guide protein structure modeling and ranking. We construct an EGNN architecture, and a message passing mechanism is designed to update and transmit information between nodes and edges of the graph. Finally, the global score of the protein model is output through a multilayer perceptron. Residue-level ultrafast shape recognition is used to describe the relationship between residues and the overall structure topology, and distance and direction encoded by Gaussian radial basis functions are designed to represent the overall topology of the protein backbone. These two features are combined with Rosetta energy terms, backbone dihedral angles and inter-residue distance and orientations to represent the protein model and embedded into the nodes and edges of the graph neural network. The experimental results on the CASP13, CASP14 and CAMEO test sets show that the scores of our developed GraphGPSM have a strong correlation with the TM-score of the models, which are significantly better than those of the unified field score function REF2015 and the state-of-the-art local lDDT-based scoring models ModFOLD8, ProQ3D and DeepAccNet, etc. The modeling experimental results on 484 test proteins demonstrate that GraphGPSM can greatly improve the modeling accuracy. GraphGPSM is further used to model 35 orphan proteins and 57 multi-domain proteins. The results show that the average TM-score of the models predicted by GraphGPSM is 13.2 and 7.1% higher than that of the models predicted by AlphaFold2. GraphGPSM also participates in CASP15 and achieves competitive performance in global accuracy estimation.

The ability to identify and track T-cell receptor (TCR) sequences from patient samples is becoming central to the field of cancer research and immunotherapy. Tracking genetically engineered T cells expressing TCRs that target specific tumor antigens is important to determine the persistence of these cells and quantify tumor responses. The available high-throughput method to profile TCR repertoires is generally referred to as TCR sequencing (TCR-Seq). However, the available TCR-Seq data are limited compared with RNA sequencing (RNA-Seq). In this paper, we have benchmarked the ability of RNA-Seq-based methods to profile TCR repertoires by examining 19 bulk RNA-Seq samples across 4 cancer cohorts including both T-cell-rich and T-cell-poor tissue types. We have performed a comprehensive evaluation of the existing RNA-Seq-based repertoire profiling methods using targeted TCR-Seq as the gold standard. We also highlighted scenarios under which the RNA-Seq approach is suitable and can provide comparable accuracy to the TCR-Seq approach. Our results show that RNA-Seq-based methods are able to effectively capture the clonotypes and estimate the diversity of TCR repertoires, as well as provide relative frequencies of clonotypes in T-cell-rich tissues and low-diversity repertoires. However, RNA-Seq-based TCR profiling methods have limited power in T-cell-poor tissues, especially in highly diverse repertoires of T-cell-poor tissues. The results of our benchmarking provide an additional appealing argument to incorporate RNA-Seq into the immune repertoire screening of cancer patients as it offers broader knowledge into the transcriptomic changes that exceed the limited information provided by TCR-Seq.

Recent advances in long read technologies not only enable large consortia to aim to sequence all eukaryotes on Earth, but they also allow individual laboratories to sequence their species of interest with relatively low investment. Long read technologies embody the promise of overcoming scaffolding problems associated with repeats and low complexity sequences, but the number of contigs often far exceeds the number of chromosomes and they may contain many insertion and deletion errors around homopolymer tracts. To overcome these issues, we have implemented the ILRA pipeline to correct long read-based assemblies. Contigs are first reordered, renamed, merged, circularized, or filtered if erroneous or contaminated. Illumina short reads are used subsequently to correct homopolymer errors. We successfully tested our approach by improving the genome sequences of Homo sapiens, Trypanosoma brucei, and Leptosphaeria spp., and by generating four novel Plasmodium falciparum assemblies from field samples. We found that correcting homopolymer tracts reduced the number of genes incorrectly annotated as pseudogenes, but an iterative approach seems to be required to correct more sequencing errors. In summary, we describe and benchmark the performance of our new tool, which improved the quality of novel long read assemblies up to 1 Gbp. The pipeline is available at GitHub: https://github.com/ThomasDOtto/ILRA.

MicroRNAs (miRNAs) are a family of non-coding RNA molecules with vital roles in regulating gene expression. Although researchers have recognized the importance of miRNAs in the development of human diseases, it is very resource-consuming to use experimental methods for identifying which dysregulated miRNA is associated with a specific disease. To reduce the cost of human effort, a growing body of studies has leveraged computational methods for predicting the potential miRNA-disease associations. However, the extant computational methods usually ignore the crucial mediating role of genes and suffer from the data sparsity problem. To address this limitation, we introduce the multi-task learning technique and develop a new model called MTLMDA (Multi-Task Learning model for predicting potential MicroRNA-Disease Associations). Different from existing models that only learn from the miRNA-disease network, our MTLMDA model exploits both miRNA-disease and gene-disease networks for improving the identification of miRNA-disease associations. To evaluate model performance, we compare our model with competitive baselines on a real-world dataset of experimentally supported miRNA-disease associations. Empirical results show that our model performs best using various performance metrics. We also examine the effectiveness of model components via ablation study and further showcase the predictive power of our model for six types of common cancers. The data and source code are available from https://github.com/qwslle/MTLMDA.

Untargeted metabolomics is gaining widespread applications. The key aspects of the data analysis include modeling complex activities of the metabolic network, selecting metabolites associated with clinical outcome and finding critical metabolic pathways to reveal biological mechanisms. One of the key roadblocks in data analysis is not well-addressed, which is the problem of matching uncertainty between data features and known metabolites. Given the limitations of the experimental technology, the identities of data features cannot be directly revealed in the data. The predominant approach for mapping features to metabolites is to match the mass-to-charge ratio (m/z) of data features to those derived from theoretical values of known metabolites. The relationship between features and metabolites is not one-to-one since some metabolites share molecular composition, and various adduct ions can be derived from the same metabolite. This matching uncertainty causes unreliable metabolite selection and functional analysis results. Here we introduce an integrated deep learning framework for metabolomics data that take matching uncertainty into consideration. The model is devised with a gradual sparsification neural network based on the known metabolic network and the annotation relationship between features and metabolites. This architecture characterizes metabolomics data and reflects the modular structure of biological system. Three goals can be achieved simultaneously without requiring much complex inference and additional assumptions: (1) evaluate metabolite importance, (2) infer feature-metabolite matching likelihood and (3) select disease sub-networks. When applied to a COVID metabolomics dataset and an aging mouse brain dataset, our method found metabolic sub-networks that were easily interpretable.

Studies have shown that the mechanism of action of many drugs is related to miRNA. In-depth research on the relationship between miRNA and drugs can provide theoretical foundations and practical approaches for various areas, such as drug target discovery, drug repositioning and biomarker research. Traditional biological experiments to test miRNA-drug susceptibility are costly and time-consuming. Thus, sequence- or topology-based deep learning methods are recognized in this field for their efficiency and accuracy. However, these methods have limitations in dealing with sparse topologies and higher-order information of miRNA (drug) feature. In this work, we propose GCFMCL, a model for multi-view contrastive learning based on graph collaborative filtering. To the best of our knowledge, this is the first attempt that incorporates contrastive learning strategy into the graph collaborative filtering framework to predict the sensitivity relationships between miRNA and drug. The proposed multi-view contrastive learning method is divided into topological contrastive objective and feature contrastive objective: (1) For the homogeneous neighbors of the topological graph, we propose a novel topological contrastive learning method via constructing the contrastive target through the topological neighborhood information of nodes. (2) The proposed model obtains feature contrastive targets from high-order feature information according to the correlation of node features, and mines potential neighborhood relationships in the feature space. The proposed multi-view comparative learning effectively alleviates the impact of heterogeneous node noise and graph data sparsity in graph collaborative filtering, and significantly enhances the performance of the model. Our study employs a dataset derived from the NoncoRNA and ncDR databases, encompassing 2049 experimentally validated miRNA-drug sensitivity associations. Five-fold cross-validation shows that the Area Under the Curve (AUC), Area Under the Precision-Recall Curve (AUPR) and F1-score (F1) of GCFMCL reach 95.28%, 95.66% and 89.77%, which outperforms the state-of-the-art (SOTA) method by the margin of 2.73%, 3.42% and 4.96%, respectively. Our code and data can be accessed at https://github.com/kkkayle/GCFMCL.

Protein complexes are key functional units in cellular processes. High-throughput techniques, such as co-fractionation coupled with mass spectrometry (CF-MS), have advanced protein complex studies by enabling global interactome inference. However, dealing with complex fractionation characteristics to define true interactions is not a simple task, since CF-MS is prone to false positives due to the co-elution of non-interacting proteins by chance. Several computational methods have been designed to analyze CF-MS data and construct probabilistic protein-protein interaction (PPI) networks. Current methods usually first infer PPIs based on handcrafted CF-MS features, and then use clustering algorithms to form potential protein complexes. While powerful, these methods suffer from the potential bias of handcrafted features and severely imbalanced data distribution. However, the handcrafted features based on domain knowledge might introduce bias, and current methods also tend to overfit due to the severely imbalanced PPI data. To address these issues, we present a balanced end-to-end learning architecture, Software for Prediction of Interactome with Feature-extraction Free Elution Data (SPIFFED), to integrate feature representation from raw CF-MS data and interactome prediction by convolutional neural network. SPIFFED outperforms the state-of-the-art methods in predicting PPIs under the conventional imbalanced training. When trained with balanced data, SPIFFED had greatly improved sensitivity for true PPIs. Moreover, the ensemble SPIFFED model provides different voting schemes to integrate predicted PPIs from multiple CF-MS data. Using the clustering software (i.e. ClusterONE), SPIFFED allows users to infer high-confidence protein complexes depending on the CF-MS experimental designs. The source code of SPIFFED is freely available at: https://github.com/bio-it-station/SPIFFED.

BACKGROUND: Promoters are DNA regions that initiate the transcription of specific genes near the transcription start sites. In bacteria, promoters are recognized by RNA polymerases and associated sigma factors. Effective promoter recognition is essential for synthesizing the gene-encoded products by bacteria to grow and adapt to different environmental conditions. A variety of machine learning-based predictors for bacterial promoters have been developed; however, most of them were designed specifically for a particular species. To date, only a few predictors are available for identifying general bacterial promoters with limited predictive performance. RESULTS: In this study, we developed TIMER, a Siamese neural network-based approach for identifying both general and species-specific bacterial promoters. Specifically, TIMER uses DNA sequences as the input and employs three Siamese neural networks with the attention layers to train and optimize the models for a total of 13 species-specific and general bacterial promoters. Extensive 10-fold cross-validation and independent tests demonstrated that TIMER achieves a competitive performance and outperforms several existing methods on both general and species-specific promoter prediction. As an implementation of the proposed method, the web server of TIMER is publicly accessible at http://web.unimelb-bioinfortools.cloud.edu.au/TIMER/.

DNA methylation plays a crucial role in transcriptional regulation. Reduced representation bisulfite sequencing (RRBS) is a technique of increasing use for analyzing genome-wide methylation profiles. Many computational tools such as Metilene, MethylKit, BiSeq and DMRfinder have been developed to use RRBS data for the detection of the differentially methylated regions (DMRs) potentially involved in epigenetic regulations of gene expression. For DMR detection tools, as for countless other medical applications, P-values and their adjustments are among the most standard reporting statistics used to assess the statistical significance of biological findings. However, P-values are coming under increasing criticism relating to their questionable accuracy and relatively high levels of false positive or negative indications. Here, we propose a method to calculate E-values, as likelihood ratios falling into the null hypothesis over the entire parameter space, for DMR detection in RRBS data. We also provide the R package 'metevalue' as a user-friendly interface to implement E-value calculations into various DMR detection tools. To evaluate the performance of E-values, we generated various RRBS benchmarking datasets using our simulator 'RRBSsim' with eight samples in each experimental group. Our comprehensive benchmarking analyses showed that using E-values not only significantly improved accuracy, area under ROC curve and power, over that of P-values or adjusted P-values, but also reduced false discovery rates and type I errors. In applications using real RRBS data of CRL rats and a clinical trial on low-salt diet, the use of E-values detected biologically more relevant DMRs and also improved the negative association between DNA methylation and gene expression.

CD8+ T cells can recognize peptides presented by class I human leukocyte antigen (HLA-I) of nucleated cells. Exploring this immune mechanism is essential for identifying T-cell vaccine targets in cancer immunotherapy. Over the past decade, the wealth of data generated by experiments has spawned many computational approaches for predicting HLA-I binding, antigen presentation and T-cell immune responses. Nevertheless, existing HLA-I binding and antigen presentation prediction approaches suffer from low precision due to the absence of T-cell receptor (TCR) recognition. Direct modeling of T-cell immune responses is less effective as TCR recognition's mechanism still remains underexplored. Therefore, directly applying these existing methods to screen cancer neoantigens is still challenging. Here, we propose a novel immune epitope prediction method termed IEPAPI by effectively incorporating antigen presentation and immunogenicity. First, IEPAPI employs a transformer-based feature extraction block to acquire representations of peptides and HLA-I proteins. Second, IEPAPI integrates the prediction of antigen presentation prediction into the input of immunogenicity prediction branch to simulate the connection between the biological processes in the T-cell immune response. Quantitative comparison results on an independent antigen presentation test dataset exhibit that IEPAPI outperformed the current state-of-the-art approaches NetMHCpan4.1 and mhcflurry2.0 on 100 (25/25) and 76% (19/25) of the HLA subtypes, respectively. Furthermore, IEPAPI demonstrates the best precision on two independent neoantigen datasets when compared with existing approaches, suggesting that IEPAPI provides a vital tool for T-cell vaccine design.

MicroRNAs (miRNAs) are human post-transcriptional regulators in humans, which are involved in regulating various physiological processes by regulating the gene expression. The subcellular localization of miRNAs plays a crucial role in the discovery of their biological functions. Although several computational methods based on miRNA functional similarity networks have been presented to identify the subcellular localization of miRNAs, it remains difficult for these approaches to effectively extract well-referenced miRNA functional representations due to insufficient miRNA-disease association representation and disease semantic representation. Currently, there has been a significant amount of research on miRNA-disease associations, making it possible to address the issue of insufficient miRNA functional representation. In this work, a novel model is established, named DAmiRLocGNet, based on graph convolutional network (GCN) and autoencoder (AE) for identifying the subcellular localizations of miRNA. The DAmiRLocGNet constructs the features based on miRNA sequence information, miRNA-disease association information and disease semantic information. GCN is utilized to gather the information of neighboring nodes and capture the implicit information of network structures from miRNA-disease association information and disease semantic information. AE is employed to capture sequence semantics from sequence similarity networks. The evaluation demonstrates that the performance of DAmiRLocGNet is superior to other competing computational approaches, benefiting from implicit features captured by using GCNs. The DAmiRLocGNet has the potential to be applied to the identification of subcellular localization of other non-coding RNAs. Moreover, it can facilitate further investigation into the functional mechanisms underlying miRNA localization. The source code and datasets are accessed at http://bliulab.net/DAmiRLocGNet.

In only a few years, as a breakthrough technology, clustered regularly interspaced short palindromic repeats/CRISPR-associated protein (CRISPR/Cas) gene-editing systems have ushered in the era of genome engineering with a plethora of applications. One of the most promising CRISPR tools, so-called base editors, opened an exciting avenue for exploring new therapeutic approaches through controlled mutagenesis. However, the efficiency of a base editor guide varies depending on several biological determinants, such as chromatin accessibility, DNA repair proteins, transcriptional activity, factors related to local sequence context and so on. Thus, the success of genetic perturbation directed by CRISPR/Cas base-editing systems relies on an optimal single guide RNA (sgRNA) design, taking those determinants into account. Although there is 11 commonly used software to design guides specifically for base editors, only three of them investigated and implemented those biological determinants into their models. This review presents the key features, capabilities and limitations of all currently available software with a particular focus on predictive model-based algorithms. Here, we summarize existing software for sgRNA design and provide a base for improving the efficiency of existing available software suites for precise target base editing.

Single-cell RNA sequencing (scRNA-seq) detects whole transcriptome signals for large amounts of individual cells and is powerful for determining cell-to-cell differences and investigating the functional characteristics of various cell types. scRNA-seq datasets are usually sparse and highly noisy. Many steps in the scRNA-seq analysis workflow, including reasonable gene selection, cell clustering and annotation, as well as discovering the underlying biological mechanisms from such datasets, are difficult. In this study, we proposed an scRNA-seq analysis method based on the latent Dirichlet allocation (LDA) model. The LDA model estimates a series of latent variables, i.e. putative functions (PFs), from the input raw cell-gene data. Thus, we incorporated the 'cell-function-gene' three-layer framework into scRNA-seq analysis, as this framework is capable of discovering latent and complex gene expression patterns via a built-in model approach and obtaining biologically meaningful results through a data-driven functional interpretation process. We compared our method with four classic methods on seven benchmark scRNA-seq datasets. The LDA-based method performed best in the cell clustering test in terms of both accuracy and purity. By analysing three complex public datasets, we demonstrated that our method could distinguish cell types with multiple levels of functional specialization, and precisely reconstruct cell development trajectories. Moreover, the LDA-based method accurately identified the representative PFs and the representative genes for the cell types/cell stages, enabling data-driven cell cluster annotation and functional interpretation. According to the literature, most of the previously reported marker/functionally relevant genes were recognized.

Binding free energy calculation of a ligand to a protein receptor is a fundamental objective in drug discovery. Molecular mechanics/Generalized-Born (Poisson-Boltzmann) surface area (MM/GB(PB)SA) is one of the most popular methods for binding free energy calculations. It is more accurate than most scoring functions and more computationally efficient than alchemical free energy methods. Several open-source tools for performing MM/GB(PB)SA calculations have been developed, but they have limitations and high entry barriers to users. Here, we introduce Uni-GBSA, a user-friendly automatic workflow to perform MM/GB(PB)SA calculations, which can perform topology preparation, structure optimization, binding free energy calculation and parameter scanning for MM/GB(PB)SA calculations. It also offers a batch mode that evaluates thousands of molecules against one protein target in parallel for efficient application in virtual screening. The default parameters are selected after systematic testing on the PDBBind-2011 refined dataset. In our case studies, Uni-GBSA produced a satisfactory correlation with the experimental binding affinities and outperformed AutoDock Vina in molecular enrichment. Uni-GBSA is available as an open-source package at https://github.com/dptech-corp/Uni-GBSA. It can also be accessed for virtual screening from the Hermite web platform at https://hermite.dp.tech. A free Uni-GBSA web server of a lab version is available at https://labs.dp.tech/projects/uni-gbsa/. This increases user-friendliness because the web server frees users from package installations and provides users with validated workflows for input data and parameter settings, cloud computing resources for efficient job completions, a user-friendly interface and professional support and maintenance.

Polygenic risk score (PRS) has been recently developed for predicting complex traits and drug responses. It remains unknown whether multi-trait PRS (mtPRS) methods, by integrating information from multiple genetically correlated traits, can improve prediction accuracy and power for PRS analysis compared with single-trait PRS (stPRS) methods. In this paper, we first review commonly used mtPRS methods and find that they do not directly model the underlying genetic correlations among traits, which has been shown to be useful in guiding multi-trait association analysis in the literature. To overcome this limitation, we propose a mtPRS-PCA method to combine PRSs from multiple traits with weights obtained from performing principal component analysis (PCA) on the genetic correlation matrix. To accommodate various genetic architectures covering different effect directions, signal sparseness and across-trait correlation structures, we further propose an omnibus mtPRS method (mtPRS-O) by combining P values from mtPRS-PCA, mtPRS-ML (mtPRS based on machine learning) and stPRSs using Cauchy Combination Test. Our extensive simulation studies show that mtPRS-PCA outperforms other mtPRS methods in both disease and pharmacogenomics (PGx) genome-wide association studies (GWAS) contexts when traits are similarly correlated, with dense signal effects and in similar effect directions, and mtPRS-O is consistently superior to most other methods due to its robustness under various genetic architectures. We further apply mtPRS-PCA, mtPRS-O and other methods to PGx GWAS data from a randomized clinical trial in the cardiovascular domain and demonstrate performance improvement of mtPRS-PCA in both prediction accuracy and patient stratification as well as the robustness of mtPRS-O in PRS association test.

In recent years, protein structure problems have become a hotspot for understanding protein folding and function mechanisms. It has been observed that most of the protein structure works rely on and benefit from co-evolutionary information obtained by multiple sequence alignment (MSA). As an example, AlphaFold2 (AF2) is a typical MSA-based protein structure tool which is famous for its high accuracy. As a consequence, these MSA-based methods are limited by the quality of the MSAs. Especially for orphan proteins that have no homologous sequence, AlphaFold2 performs unsatisfactorily as MSA depth decreases, which may pose a barrier to its widespread application in protein mutation and design problems in which there are no rich homologous sequences and rapid prediction is needed. In this paper, we constructed two standard datasets for orphan and de novo proteins which have insufficient/none homology information, called Orphan62 and Design204, respectively, to fairly evaluate the performance of the various methods in this case. Then, depending on whether or not utilizing scarce MSA information, we summarized two approaches, MSA-enhanced and MSA-free methods, to effectively solve the issue without sufficient MSAs. MSA-enhanced model aims to improve poor MSA quality from the data source by knowledge distillation and generation models. MSA-free model directly learns the relationship between residues on enormous protein sequences from pre-trained models, bypassing the step of extracting the residue pair representation from MSA. Next, we evaluated the performance of four MSA-free methods (trRosettaX-Single, TRFold, ESMFold and ProtT5) and MSA-enhanced (Bagging MSA) method compared with a traditional MSA-based method AlphaFold2, in two protein structure-related prediction tasks, respectively. Comparison analyses show that trRosettaX-Single and ESMFold which belong to MSA-free method can achieve fast prediction ($\sim\! 40$s) and comparable performance compared with AF2 in tertiary structure prediction, especially for short peptides, $\alpha $-helical segments and targets with few homologous sequences. Bagging MSA utilizing MSA enhancement improves the accuracy of our trained base model which is an MSA-based method when poor homology information exists in secondary structure prediction. Our study provides biologists an insight of how to select rapid and appropriate prediction tools for enzyme engineering and peptide drug development. CONTACT: guofei@csu.edu.cn, jj.tang@siat.ac.cn.

Classifying epitopes is essential since they can be applied in various fields, including therapeutics, diagnostics and peptide-based vaccines. To determine the epitope or peptide against an antibody, epitope mapping with peptides is the most extensively used method. However, this method is more time-consuming and inefficient than using present methods. The ability to retrieve data on protein sequences through laboratory procedures has led to the development of computational models that predict epitope binding based on machine learning and deep learning (DL). It has also evolved to become a crucial part of developing effective cancer immunotherapies. This paper proposes an architecture to generalize this case since various research strives to solve a low-performance classification problem. A proposed DL model is the fusion architecture, which combines two architectures: Transformer architecture and convolutional neural network (CNN), called MITNet and MITNet-Fusion. Combining these two architectures enriches feature space to correlate epitope labels with the binary classification method. The selected epitope-T-cell receptor (TCR) interactions are GILG, GLCT and NLVP, acquired from three databases: IEDB, VDJdb and McPAS-TCR. The previous input data was extracted using amino acid composition, dipeptide composition, spectrum descriptor and the combination of all those features called AADIP composition to encode the input data to DL architecture. For ensuring consistency, fivefold cross-validations were performed using the area under curve metric. Results showed that GILG, GLCT and NLVP received scores of 0.85, 0.87 and 0.86, respectively. Those results were compared to prior architecture and outperformed other similar deep learning models.

RNAs can interact with other molecules in their environment, such as ions, proteins or other RNAs, to form complexes with important biological roles. The prediction of the structure of these complexes is therefore an important issue and a difficult task. We are interested in RNA complexes composed of several (more than two) interacting RNAs. We show how available knowledge on the considered RNAs can help predict their secondary structure. We propose an interactive tool for the prediction of RNA complexes, called C-RCPRed, that considers user knowledge and probing data (which can be generated experimentally or artificially). C-RCPred is based on a multi-objective optimization algorithm. Through an extensive benchmarking procedure, which includes state-of-the-art methods, we show the efficiency of the multi-objective approach and the positive impact of considering user knowledge and probing data on the prediction results. C-RCPred is freely available as an open-source program and web server on the EvryRNA website (https://evryrna.ibisc.univ-evry.fr).

We report the structure-based pathogenicity relationship identifier (SPRI), a novel computational tool for accurate evaluation of pathological effects of missense single mutations and prediction of higher-order spatially organized units of mutational clusters. SPRI can effectively extract properties determining pathogenicity encoded in protein structures, and can identify deleterious missense mutations of germ line origin associated with Mendelian diseases, as well as mutations of somatic origin associated with cancer drivers. It compares favorably to other methods in predicting deleterious mutations. Furthermore, SPRI can discover spatially organized pathogenic higher-order spatial clusters (patHOS) of deleterious mutations, including those of low recurrence, and can be used for discovery of candidate cancer driver genes and driver mutations. We further demonstrate that SPRI can take advantage of AlphaFold2 predicted structures and can be deployed for saturation mutation analysis of the whole human proteome.

Trans-ethnic genome-wide association studies have revealed that many loci identified in European populations can be reproducible in non-European populations, indicating widespread trans-ethnic genetic similarity. However, how to leverage such shared information more efficiently in association analysis is less investigated for traits in underrepresented populations. We here propose a statistical framework, trans-ethnic genetic risk score informed gene-based association mixed model (GAMM), by hierarchically modeling single-nucleotide polymorphism effects in the target population as a function of effects of the same trait in well-studied populations. GAMM powerfully integrates genetic similarity across distinct ancestral groups to enhance power in understudied populations, as confirmed by extensive simulations. We illustrate the usefulness of GAMM via the application to 13 blood cell traits (i.e. basophil count, eosinophil count, hematocrit, hemoglobin concentration, lymphocyte count, mean corpuscular hemoglobin, mean corpuscular hemoglobin concentration, mean corpuscular volume, monocyte count, neutrophil count, platelet count, red blood cell count and total white blood cell count) in Africans of the UK Biobank (n = 3204) while utilizing genetic overlap shared in Europeans (n = 746 667) and East Asians (n = 162 255). We discovered multiple new associated genes, which had otherwise been missed by existing methods, and revealed that the trans-ethnic information indirectly contributed much to the phenotypic variance. Overall, GAMM represents a flexible and powerful statistical framework of association analysis for complex traits in underrepresented populations by integrating trans-ethnic genetic similarity across well-studied populations, and helps attenuate health inequities in current genetics research for people of minority populations.

Human Phenotype Ontology (HPO)-based approaches have gained popularity in recent times as a tool for genomic diagnostics of rare diseases. However, these approaches do not make full use of the available information on disease and patient phenotypes. We present a new method called Phen2Disease, which utilizes the bidirectional maximum matching semantic similarity between two phenotype sets of patients and diseases to prioritize diseases and genes. Our comprehensive experiments have been conducted on six real data cohorts with 2051 cases (Cohort 1, n = 384; Cohort 2, n = 281; Cohort 3, n = 185; Cohort 4, n = 784; Cohort 5, n = 208; and Cohort 6, n = 209) and two simulated data cohorts with 1000 cases. The results of the experiments showed that Phen2Disease outperforms the three state-of-the-art methods when only phenotype information and HPO knowledge base are used, particularly in cohorts with fewer average numbers of HPO terms. We also observed that patients with higher information content scores have more specific information, leading to more accurate predictions. Moreover, Phen2Disease provides high interpretability with ranked diseases and patient HPO terms presented. Our method provides a novel approach to utilizing phenotype data for genomic diagnostics of rare diseases, with potential for clinical impact. Phen2Disease is freely available on GitHub at https://github.com/ZhuLab-Fudan/Phen2Disease.

Precise targeting of transcription factor binding sites (TFBSs) is essential to comprehending transcriptional regulatory processes and investigating cellular function. Although several deep learning algorithms have been created to predict TFBSs, the models' intrinsic mechanisms and prediction results are difficult to explain. There is still room for improvement in prediction performance. We present DeepSTF, a unique deep-learning architecture for predicting TFBSs by integrating DNA sequence and shape profiles. We use the improved transformer encoder structure for the first time in the TFBSs prediction approach. DeepSTF extracts DNA higher-order sequence features using stacked convolutional neural networks (CNNs), whereas rich DNA shape profiles are extracted by combining improved transformer encoder structure and bidirectional long short-term memory (Bi-LSTM), and, finally, the derived higher-order sequence features and representative shape profiles are integrated into the channel dimension to achieve accurate TFBSs prediction. Experiments on 165 ENCODE chromatin immunoprecipitation sequencing (ChIP-seq) datasets show that DeepSTF considerably outperforms several state-of-the-art algorithms in predicting TFBSs, and we explain the usefulness of the transformer encoder structure and the combined strategy using sequence features and shape profiles in capturing multiple dependencies and learning essential features. In addition, this paper examines the significance of DNA shape features predicting TFBSs. The source code of DeepSTF is available at https://github.com/YuBinLab-QUST/DeepSTF/.

Adverse drug events (ADEs) are common in clinical practice and can cause significant harm to patients and increase resource use. Natural language processing (NLP) has been applied to automate ADE detection, but NLP systems become less adaptable when drug entities are missing or multiple medications are specified in clinical narratives. Additionally, no Chinese-language NLP system has been developed for ADE detection due to the complexity of Chinese semantics, despite ˃10 million cases of drug-related adverse events occurring annually in China. To address these challenges, we propose DKADE, a deep learning and knowledge graph-based framework for identifying ADEs. DKADE infers missing drug entities and evaluates their correlations with ADEs by combining medication orders and existing drug knowledge. Moreover, DKADE can automatically screen for new adverse drug reactions. Experimental results show that DKADE achieves an overall F1-score value of 91.13%. Furthermore, the adaptability of DKADE is validated using real-world external clinical data. In summary, DKADE is a powerful tool for studying drug safety and automating adverse event monitoring.

Computational analysis of RNA sequences constitutes a crucial step in the field of RNA biology. As in other domains of the life sciences, the incorporation of artificial intelligence and machine learning techniques into RNA sequence analysis has gained significant traction in recent years. Historically, thermodynamics-based methods were widely employed for the prediction of RNA secondary structures; however, machine learning-based approaches have demonstrated remarkable advancements in recent years, enabling more accurate predictions. Consequently, the precision of sequence analysis pertaining to RNA secondary structures, such as RNA-protein interactions, has also been enhanced, making a substantial contribution to the field of RNA biology. Additionally, artificial intelligence and machine learning are also introducing technical innovations in the analysis of RNA-small molecule interactions for RNA-targeted drug discovery and in the design of RNA aptamers, where RNA serves as its own ligand. This review will highlight recent trends in the prediction of RNA secondary structure, RNA aptamers and RNA drug discovery using machine learning, deep learning and related technologies, and will also discuss potential future avenues in the field of RNA informatics.

Protein-deoxyribonucleic acid (DNA) interactions are important in a variety of biological processes. Accurately predicting protein-DNA binding affinity has been one of the most attractive and challenging issues in computational biology. However, the existing approaches still have much room for improvement. In this work, we propose an ensemble model for Protein-DNA Binding Affinity prediction (emPDBA), which combines six base models with one meta-model. The complexes are classified into four types based on the DNA structure (double-stranded or other forms) and the percentage of interface residues. For each type, emPDBA is trained with the sequence-based, structure-based and energy features from binding partners and complex structures. Through feature selection by the sequential forward selection method, it is found that there do exist considerable differences in the key factors contributing to intermolecular binding affinity. The complex classification is beneficial for the important feature extraction for binding affinity prediction. The performance comparison of our method with other peer ones on the independent testing dataset shows that emPDBA outperforms the state-of-the-art methods with the Pearson correlation coefficient of 0.53 and the mean absolute error of 1.11 kcal/mol. The comprehensive results demonstrate that our method has a good performance for protein-DNA binding affinity prediction. Availability and implementation: The source code is available at https://github.com/ChunhuaLiLab/emPDBA/.

Ribonucleic acids (RNAs) play crucial roles in living organisms and some of them, such as bacterial ribosomes and precursor messenger RNA, are targets of small molecule drugs, whereas others, e.g. bacterial riboswitches or viral RNA motifs are considered as potential therapeutic targets. Thus, the continuous discovery of new functional RNA increases the demand for developing compounds targeting them and for methods for analyzing RNA-small molecule interactions. We recently developed fingeRNAt-a software for detecting non-covalent bonds formed within complexes of nucleic acids with different types of ligands. The program detects several non-covalent interactions and encodes them as structural interaction fingerprint (SIFt). Here, we present the application of SIFts accompanied by machine learning methods for binding prediction of small molecules to RNA. We show that SIFt-based models outperform the classic, general-purpose scoring functions in virtual screening. We also employed Explainable Artificial Intelligence (XAI)-the SHapley Additive exPlanations, Local Interpretable Model-agnostic Explanations and other methods to help understand the decision-making process behind the predictive models. We conducted a case study in which we applied XAI on a predictive model of ligand binding to human immunodeficiency virus type 1 trans-activation response element RNA to distinguish between residues and interaction types important for binding. We also used XAI to indicate whether an interaction has a positive or negative effect on binding prediction and to quantify its impact. Our results obtained using all XAI methods were consistent with the literature data, demonstrating the utility and importance of XAI in medicinal chemistry and bioinformatics.

Multiple sequence alignment is widely used for sequence analysis, such as identifying important sites and phylogenetic analysis. Traditional methods, such as progressive alignment, are time-consuming. To address this issue, we introduce StarTree, a novel method to fast construct a guide tree by combining sequence clustering and hierarchical clustering. Furthermore, we develop a new heuristic similar region detection algorithm using the FM-index and apply the k-banded dynamic program to the profile alignment. We also introduce a win-win alignment algorithm that applies the central star strategy within the clusters to fast the alignment process, then uses the progressive strategy to align the central-aligned profiles, guaranteeing the final alignment's accuracy. We present WMSA 2 based on these improvements and compare the speed and accuracy with other popular methods. The results show that the guide tree made by the StarTree clustering method can lead to better accuracy than that of PartTree while consuming less time and memory than that of UPGMA and mBed methods on datasets with thousands of sequences. During the alignment of simulated data sets, WMSA 2 can consume less time and memory while ranking at the top of Q and TC scores. The WMSA 2 is still better at the time, and memory efficiency on the real datasets and ranks at the top on the average sum of pairs score. For the alignment of 1 million SARS-CoV-2 genomes, the win-win mode of WMSA 2 significantly decreased the consumption time than the former version. The source code and data are available at https://github.com/malabz/WMSA2.

Survival analysis is critical to cancer prognosis estimation. High-throughput technologies facilitate the increase in the dimension of genic features, but the number of clinical samples in cohorts is relatively small due to various reasons, including difficulties in participant recruitment and high data-generation costs. Transcriptome is one of the most abundantly available OMIC (referring to the high-throughput data, including genomic, transcriptomic, proteomic and epigenomic) data types. This study introduced a multitask graph attention network (GAT) framework DQSurv for the survival analysis task. We first used a large dataset of healthy tissue samples to pretrain the GAT-based HealthModel for the quantitative measurement of the gene regulatory relations. The multitask survival analysis framework DQSurv used the idea of transfer learning to initiate the GAT model with the pretrained HealthModel and further fine-tuned this model using two tasks i.e. the main task of survival analysis and the auxiliary task of gene expression prediction. This refined GAT was denoted as DiseaseModel. We fused the original transcriptomic features with the difference vector between the latent features encoded by the HealthModel and DiseaseModel for the final task of survival analysis. The proposed DQSurv model stably outperformed the existing models for the survival analysis of 10 benchmark cancer types and an independent dataset. The ablation study also supported the necessity of the main modules. We released the codes and the pretrained HealthModel to facilitate the feature encodings and survival analysis of transcriptome-based future studies, especially on small datasets. The model and the code are available at http://www.healthinformaticslab.org/supp/.

RNA localization is essential for regulating spatial translation, where RNAs are trafficked to their target locations via various biological mechanisms. In this review, we discuss RNA localization in the context of molecular mechanisms, experimental techniques and machine learning-based prediction tools. Three main types of molecular mechanisms that control the localization of RNA to distinct cellular compartments are reviewed, including directed transport, protection from mRNA degradation, as well as diffusion and local entrapment. Advances in experimental methods, both image and sequence based, provide substantial data resources, which allow for the design of powerful machine learning models to predict RNA localizations. We review the publicly available predictive tools to serve as a guide for users and inspire developers to build more effective prediction models. Finally, we provide an overview of multimodal learning, which may provide a new avenue for the prediction of RNA localization.

Metabolism refers to a series of orderly chemical reactions used to maintain life activities in organisms. In healthy individuals, metabolism remains within a normal range. However, specific diseases can lead to abnormalities in the levels of certain metabolites, causing them to either increase or decrease. Detecting these deviations in metabolite levels can aid in diagnosing a disease. Traditional biological experiments often rely on a lot of manpower to do repeated experiments, which is time consuming and labor intensive. To address this issue, we develop a deep learning model based on the auto-encoder and non-negative matrix factorization named as MDA-AENMF to predict the potential associations between metabolites and diseases. We integrate a variety of similarity networks and then acquire the characteristics of both metabolites and diseases through three specific modules. First, we get the disease characteristics from the five-layer auto-encoder module. Later, in the non-negative matrix factorization module, we extract both the metabolite and disease characteristics. Furthermore, the graph attention auto-encoder module helps us obtain metabolite characteristics. After obtaining the features from three modules, these characteristics are merged into a single, comprehensive feature vector for each metabolite-disease pair. Finally, we send the corresponding feature vector and label to the multi-layer perceptron for training. The experiment demonstrates our area under the receiver operating characteristic curve of 0.975 and area under the precision-recall curve of 0.973 in 5-fold cross-validation, which are superior to those of existing state-of-the-art predictive methods. Through case studies, most of the new associations obtained by MDA-AENMF have been verified, further highlighting the reliability of MDA-AENMF in predicting the potential relationships between metabolites and diseases.

Accurately identifying phage-host relationships from their genome sequences is still challenging, especially for those phages and hosts with less homologous sequences. In this work, focusing on identifying the phage-host relationships at the species and genus level, we propose a contrastive learning based approach to learn whole-genome sequence embeddings that can take account of phage-host interactions (PHIs). Contrastive learning is used to make phages infecting the same hosts close to each other in the new representation space. Specifically, we rephrase whole-genome sequences with frequency chaos game representation (FCGR) and learn latent embeddings that 'encapsulate' phages and host relationships through contrastive learning. The contrastive learning method works well on the imbalanced dataset. Based on the learned embeddings, a proposed pipeline named CL4PHI can predict known hosts and unseen hosts in training. We compare our method with two recently proposed state-of-the-art learning-based methods on their benchmark datasets. The experiment results demonstrate that the proposed method using contrastive learning improves the prediction accuracy on known hosts and demonstrates a zero-shot prediction capability on unseen hosts. In terms of potential applications, the rapid pace of genome sequencing across different species has resulted in a vast amount of whole-genome sequencing data that require efficient computational methods for identifying phage-host interactions. The proposed approach is expected to address this need by efficiently processing whole-genome sequences of phages and prokaryotic hosts and capturing features related to phage-host relationships for genome sequence representation. This approach can be used to accelerate the discovery of phage-host interactions and aid in the development of phage-based therapies for infectious diseases.

MOTIVATION: Recent advances in spatial transcriptomics technologies have enabled gene expression profiles while preserving spatial context. Accurately identifying spatial domains is crucial for downstream analysis and it requires the effective integration of gene expression profiles and spatial information. While increasingly computational methods have been developed for spatial domain detection, most of them cannot adaptively learn the complex relationship between gene expression and spatial information, leading to sub-optimal performance. RESULTS: To overcome these challenges, we propose a novel deep learning method named Spatial-MGCN for identifying spatial domains, which is a Multi-view Graph Convolutional Network (GCN) with attention mechanism. We first construct two neighbor graphs using gene expression profiles and spatial information, respectively. Then, a multi-view GCN encoder is designed to extract unique embeddings from both the feature and spatial graphs, as well as their shared embeddings by combining both graphs. Finally, a zero-inflated negative binomial decoder is used to reconstruct the original expression matrix by capturing the global probability distribution of gene expression profiles. Moreover, Spatial-MGCN incorporates a spatial regularization constraint into the features learning to preserve spatial neighbor information in an end-to-end manner. The experimental results show that Spatial-MGCN outperforms state-of-the-art methods consistently in several tasks, including spatial clustering and trajectory inference.