BACKGROUND: DNA sequencing is increasingly incorporated into the routine care of cancer patients, many of whom also carry inherited, moderate/high-penetrance variants associated with other diseases. Yet, the prevalence and consequence of such variants remain unclear. METHODS: We analyzed the germline genomes of 10,389 adult cancer cases in the TCGA cohort, identifying pathogenic/likely pathogenic variants in autosomal-dominant genes, autosomal-recessive genes, and 59 medically actionable genes curated by the American College of Molecular Genetics (i.e., the ACMG 59 genes). We also analyzed variant- and gene-level expression consequences in carriers. RESULTS: The affected genes exhibited varying pan-ancestry and population-specific patterns, and overall, the European population showed the highest frequency of pathogenic/likely pathogenic variants. We further identified genes showing expression consequence supporting variant functionality, including altered gene expression, allelic specific expression, and mis-splicing determined by a massively parallel splicing assay. CONCLUSIONS: Our results demonstrate that expression-altering variants are found in a substantial fraction of cases and illustrate the yield of genomic risk assessments for a wide range of diseases across diverse populations.

BACKGROUND: Epstein-Barr virus (EBV)-associated gastric carcinomas (EBVaGCs) present unique molecular signatures, but the tumorigenesis of EBVaGCs and the role EBV plays during this process remain poorly understood. METHODS: We applied whole-exome sequencing, EBV genome sequencing, and whole-genome bisulfite sequencing to multiple samples (n = 123) derived from the same patients (n = 25), which covered saliva samples and different histological stages from morphologically normal epithelial tissues to dysplasia and EBVaGCs. We compared the genomic landscape between EBVaGCs and their precursor lesions and traced the clonal evolution for each patient. We also analyzed genome sequences of EBV from samples of different histological types. Finally, the key molecular events promoting the tumor evolution were demonstrated by MTT, IC50, and colony formation assay in vitro experiments and in vivo xenograft experiments. RESULTS: Our analysis revealed increasing mutational burden and EBV load from normal tissues and low-grade dysplasia (LD) to high-grade dysplasia (HD) and EBVaGCs, and oncogenic amplifications occurred late in EBVaGCs. Interestingly, within each patient, EBVaGCs and HDs were monoclonal and harbored single-strain-originated EBV, but saliva or normal tissues/LDs had different EBV strains from that in EBVaGCs. Compared with precursor lesions, tumor cells showed incremental methylation in promotor regions, whereas EBV presented consistent hypermethylation. Dominant alterations targeting the PI3K-Akt and Wnt pathways were found in EBV-infected cells. The combinational inhibition of these two pathways in EBV-positive tumor cells confirmed their synergistic function. CONCLUSIONS: We portrayed the (epi) genomic evolution process of EBVaGCs, revealed the extensive genomic diversity of EBV between tumors and normal tissue sites, and demonstrated the synergistic activation of the PI3K and Wnt pathways in EBVaGCs, offering a new potential treatment strategy for this disease.

We developed a sensitive sequencing approach that simultaneously profiles microsatellite instability, chromosomal instability, and subclonal structure in cancer. We assessed diverse repeat motifs across 225 microsatellites on colorectal carcinomas. Our study identified elevated alterations at both selected tetranucleotide and conventional mononucleotide repeats. Many colorectal carcinomas had a mix of genomic instability states that are normally considered exclusive. An MSH3 mutation may have contributed to the mixed states. Increased copy number of chromosome arm 8q was most prevalent among tumors with microsatellite instability, including a case of translocation involving 8q. Subclonal analysis identified co-occurring driver mutations previously known to be exclusive.

BACKGROUND: The incidence of Gram-negative bloodstream infections (BSIs), predominantly caused by Escherichia coli and Klebsiella species, continues to increase; however, the causes of this are unclear and effective interventions are therefore hard to design. METHODS: In this study, we sequenced 3468 unselected isolates over a decade in Oxfordshire (UK) and linked this data to routinely collected electronic healthcare records and mandatory surveillance reports. We annotated genomes for clinically relevant genes, contrasting the distribution of these within and between species, and compared incidence trends over time using stacked negative binomial regression. RESULTS: We demonstrate that the observed increases in E. coli incidence were not driven by the success of one or more sequence types (STs); instead, four STs continue to dominate a stable population structure, with no evidence of adaptation to hospital/community settings. Conversely in Klebsiella spp., most infections are caused by sporadic STs with the exception of a local drug-resistant outbreak strain (ST490). Virulence elements are highly structured by ST in E. coli but not Klebsiella spp. where they occur in a diverse spectrum of STs and equally across healthcare and community settings. Most clinically hypervirulent (i.e. community-onset) Klebsiella BSIs have no known acquired virulence loci. Finally, we demonstrate a diverse but largely genus-restricted mobilome with close associations between antimicrobial resistance (AMR) genes and insertion sequences but not typically specific plasmid replicon types, consistent with the dissemination of AMR genes being highly contingent on smaller mobile genetic elements (MGEs). CONCLUSIONS: Our large genomic study highlights distinct differences in the molecular epidemiology of E. coli and Klebsiella BSIs and suggests that no single specific pathogen genetic factors (e.g. AMR/virulence genes/sequence type) are likely contributing to the increasing incidence of BSI overall, that association with AMR genes in E. coli is a contributor to the increasing number of E. coli BSIs, and that more attention should be given to AMR gene associations with non-plasmid MGEs to try and understand horizontal gene transfer networks.

Deep learning is a subdiscipline of artificial intelligence that uses a machine learning technique called artificial neural networks to extract patterns and make predictions from large data sets. The increasing adoption of deep learning across healthcare domains together with the availability of highly characterised cancer datasets has accelerated research into the utility of deep learning in the analysis of the complex biology of cancer. While early results are promising, this is a rapidly evolving field with new knowledge emerging in both cancer biology and deep learning. In this review, we provide an overview of emerging deep learning techniques and how they are being applied to oncology. We focus on the deep learning applications for omics data types, including genomic, methylation and transcriptomic data, as well as histopathology-based genomic inference, and provide perspectives on how the different data types can be integrated to develop decision support tools. We provide specific examples of how deep learning may be applied in cancer diagnosis, prognosis and treatment management. We also assess the current limitations and challenges for the application of deep learning in precision oncology, including the lack of phenotypically rich data and the need for more explainable deep learning models. Finally, we conclude with a discussion of how current obstacles can be overcome to enable future clinical utilisation of deep learning.

BACKGROUND: Rapid advances in the past decade have shown that dysbiosis of the gut microbiome is a key hallmark of rheumatoid arthritis (RA). Yet, the relationship between the gut microbiome and clinical improvement in RA disease activity remains unclear. In this study, we explored the gut microbiome of patients with RA to identify features that are associated with, as well as predictive of, minimum clinically important improvement (MCII) in disease activity. METHODS: We conducted a retrospective, observational cohort study on patients diagnosed with RA between 1988 and 2014. Whole metagenome shotgun sequencing was performed on 64 stool samples, which were collected from 32 patients with RA at two separate time-points approximately 6-12 months apart. The Clinical Disease Activity Index (CDAI) of each patient was measured at both time-points to assess achievement of MCII; depending on this clinical status, patients were distinguished into two groups: MCII+ (who achieved MCII; n = 12) and MCII- (who did not achieve MCII; n = 20). Multiple linear regression models were used to identify microbial taxa and biochemical pathways associated with MCII while controlling for potentially confounding factors. Lastly, a deep-learning neural network was trained upon gut microbiome, clinical, and demographic data at baseline to classify patients according to MCII status, thereby enabling the prediction of whether a patient will achieve MCII at follow-up. RESULTS: We found age to be the largest determinant of the overall compositional variance in the gut microbiome (R2 = 7.7%, P = 0.001, PERMANOVA). Interestingly, the next factor identified to explain the most variance in the gut microbiome was MCII status (R2 = 3.8%, P = 0.005). Additionally, by looking at patients' baseline gut microbiome profiles, we observed significantly different microbiome traits between patients who eventually showed MCII and those who did not. Taxonomic features include alpha- and beta-diversity measures, as well as several microbial taxa, such as Coprococcus, Bilophila sp. 4_1_30, and Eubacterium sp. 3_1_31. Notably, patients who achieved clinical improvement had higher alpha-diversity in their gut microbiomes at both baseline and follow-up visits. Functional profiling identified fifteen biochemical pathways, most of which were involved in the biosynthesis of L-arginine, L-methionine, and tetrahydrofolate, to be differentially abundant between the MCII patient groups. Moreover, MCII+ and MCII- groups showed significantly different fold-changes (from baseline to follow-up) in eight microbial taxa and in seven biochemical pathways. These results could suggest that, depending on the clinical course, gut microbiomes not only start at different ecological states, but also are on separate trajectories. Finally, the neural network proved to be highly effective in predicting which patients will achieve MCII (balanced accuracy = 90.0%, leave-one-out cross-validation), demonstrating potential clinical utility of gut microbiome profiles. CONCLUSIONS: Our findings confirm the presence of taxonomic and functional signatures of the gut microbiome associated with MCII in RA patients. Ultimately, modifying the gut microbiome to enhance clinical outcome may hold promise as a future treatment for RA.

Author information: (1)Department of Medical Genetics, University of British Columbia and Children's & Women's Hospital, Vancouver, BC, V6H3N1, Canada. indhu.babu@bcchr.ca. (2)Department of Medical and Molecular Genetics, King's College London, Strand, London, WC2R 2LS, UK. indhu.babu@bcchr.ca. (3)Department of Medical Genetics, University of British Columbia and Children's & Women's Hospital, Vancouver, BC, V6H3N1, Canada. (4)Canada's Michael Smith Genome Sciences Centre, BC Cancer Agency, Vancouver, BC, V5Z4S6, Canada. (5)Bioinformatics Graduate Program, University of British Columbia, Vancouver, BC, V6T1Z4, Canada. (6)Illumina Inc., San Diego, CA, 92121, USA.

BACKGROUND: Klebsiella spp. are opportunistic pathogens which can cause severe infections, are often multi-drug resistant and are a common cause of hospital-acquired infections. Multiple new Klebsiella species have recently been described, yet their clinical impact and antibiotic resistance profiles are largely unknown. We aimed to explore Klebsiella group- and species-specific clinical impact, antimicrobial resistance (AMR) and virulence. METHODS: We analysed whole-genome sequence data of a diverse selection of Klebsiella spp. isolates and identified resistance and virulence factors. Using the genomes of 3594 Klebsiella isolates, we predicted the masses of 56 ribosomal subunit proteins and identified species-specific marker masses. We then re-analysed over 22,000 Matrix-Assisted Laser Desorption Ionization - Time Of Flight (MALDI-TOF) mass spectra routinely acquired at eight healthcare institutions in four countries looking for these species-specific markers. Analyses of clinical and microbiological endpoints from a subset of 957 patients with infections from Klebsiella species were performed using generalized linear mixed-effects models. RESULTS: Our comparative genomic analysis shows group- and species-specific trends in accessory genome composition. With the identified species-specific marker masses, eight Klebsiella species can be distinguished using MALDI-TOF MS. We identified K. pneumoniae (71.2%; n = 12,523), K. quasipneumoniae (3.3%; n = 575), K. variicola (9.8%; n = 1717), "K. quasivariicola" (0.3%; n = 52), K. oxytoca (8.2%; n = 1445), K. michiganensis (4.8%; n = 836), K. grimontii (2.4%; n = 425) and K. huaxensis (0.1%; n = 12). Isolates belonging to the K. oxytoca group, which includes the species K. oxytoca, K. michiganensis and K. grimontii, were less often resistant to 4th-generation cephalosporins than isolates of the K. pneumoniae group, which includes the species K. pneumoniae, K. quasipneumoniae, K. variicola and "K. quasivariicola" (odds ratio = 0.17, p < 0.001, 95% confidence interval [0.09,0.28]). Within the K. pneumoniae group, isolates identified as K. pneumoniae were more often resistant to 4th-generation cephalosporins than K. variicola isolates (odds ratio = 2.61, p = 0.003, 95% confidence interval [1.38,5.06]). K. oxytoca group isolates were found to be more likely associated with invasive infection to primary sterile sites than K. pneumoniae group isolates (odds ratio = 2.39, p = 0.0044, 95% confidence interval [1.05,5.53]). CONCLUSIONS: Currently misdiagnosed Klebsiella spp. can be distinguished using a ribosomal marker-based approach for MALDI-TOF MS. Klebsiella groups and species differed in AMR profiles, and in their association with invasive infection, highlighting the importance for species identification to enable effective treatment options.

BACKGROUND: Colorectal cancer (CRC) is a major cancer type whose mechanism of metastasis remains elusive. METHODS: In this study, we characterised the evolutionary pattern of metastatic CRC (mCRC) by analysing bulk and single-cell exome sequencing data of primary and metastatic tumours from 7 CRC patients with liver metastases. Here, 7 CRC patients were analysed by bulk whole-exome sequencing (WES); 4 of these were also analysed using single-cell sequencing. RESULTS: Despite low genomic divergence between paired primary and metastatic cancers in the bulk data, single-cell WES (scWES) data revealed rare mutations and defined two separate cell populations, indicative of the diverse evolutionary trajectories between primary and metastatic tumour cells. We further identified 24 metastatic cell-specific-mutated genes and validated their functions in cell migration capacity. CONCLUSIONS: In summary, scWES revealed rare mutations that failed to be detected by bulk WES. These rare mutations better define the distinct genomic profiles of primary and metastatic tumour cell clones.

BACKGROUND: Colorectal cancer is the 2nd leading cause of cancer-related deaths with few patients benefiting from biomarker-guided therapy. Mutation expression is essential for accurate interpretation of mutations as biomarkers, but surprisingly, little has been done to analyze somatic cancer mutations on the expression level. We report a large-scale analysis of allele-specific mutation expression. METHODS: Whole-exome and total RNA sequencing was performed on 137 samples from 121 microsatellite stable colorectal cancers, including multiregional samples of primary and metastatic tumors from 4 patients. Data were integrated with allele-specific resolution. Results were validated in an independent set of 241 colon cancers. Therapeutic associations were explored by pharmacogenomic profiling of 15 cell lines or patient-derived organoids. RESULTS: The median proportion of expressed mutations per tumor was 34%. Cancer-critical mutations had the highest expression frequency (gene-wise mean of 58%), independent of frequent allelic imbalance. Systematic deviation from the general pattern of expression levels according to allelic frequencies was detected, including preferential expression of mutated alleles dependent on the mutation type and target gene. Translational relevance was suggested by correlations of KRAS/NRAS or TP53 mutation expression levels with downstream oncogenic signatures (p < 0.03), overall survival among patients with stage II and III cancer (KRAS/NRAS: hazard ratio 6.1, p = 0.0070), and targeted drug sensitivity. The latter was demonstrated for EGFR and MDM2 inhibition in pre-clinical models. CONCLUSIONS: Only a subset of mutations in microsatellite stable colorectal cancers were expressed, and the "expressed mutation dose" may provide an opportunity for more fine-tuned biomarker interpretations.

BACKGROUND: Genome-wide association studies (GWAS) have identified over 56 susceptibility loci associated with Alzheimer's disease (AD), but the genes responsible for these associations remain largely unknown. METHODS: We performed a large transcriptome-wide association study (TWAS) leveraging modified UTMOST (Unified Test for MOlecular SignaTures) prediction models of ten brain tissues that are potentially related to AD to discover novel AD genetic loci and putative target genes in 71,880 (proxy) cases and 383,378 (proxy) controls of European ancestry. RESULTS: We identified 53 genes with predicted expression associations with AD risk at Bonferroni correction threshold (P value < 3.38 × 10-6). Based on fine-mapping analyses, 21 genes at nine loci showed strong support for being causal. CONCLUSIONS: Our study provides new insights into the etiology and underlying genetic architecture of AD.

BACKGROUND: Epithelial ovarian cancer (OC) is a heterogenous disease consisting of five major histologically distinct subtypes: high-grade serous (HGSOC), low-grade serous (LGSOC), endometrioid (ENOC), clear cell (CCOC) and mucinous (MOC). Although HGSOC is the most prevalent subtype, representing 70-80% of cases, a 2013 landmark study by Domcke et al. found that the most frequently used OC cell lines are not molecularly representative of this subtype. This raises the question, if not HGSOC, from which subtype do these cell lines derive? Indeed, non-HGSOC subtypes often respond poorly to chemotherapy; therefore, representative models are imperative for developing new targeted therapeutics. METHODS: Non-negative matrix factorisation (NMF) was applied to transcriptomic data from 44 OC cell lines in the Cancer Cell Line Encyclopedia, assessing the quality of clustering into 2-10 groups. Epithelial OC subtypes were assigned to cell lines optimally clustered into five transcriptionally distinct classes, confirmed by integration with subtype-specific mutations. A transcriptional subtype classifier was then developed by trialling three machine learning algorithms using subtype-specific metagenes defined by NMF. The ability of classifiers to predict subtype was tested using RNA sequencing of a living biobank of patient-derived OC models. RESULTS: Application of NMF optimally clustered the 44 cell lines into five transcriptionally distinct groups. Close inspection of orthogonal datasets revealed this five-cluster delineation corresponds to the five major OC subtypes. This NMF-based classification validates the Domcke et al. analysis, in identifying lines most representative of HGSOC, and additionally identifies models representing the four other subtypes. However, NMF of the cell lines into two clusters did not align with the dualistic model of OC and suggests this classification is an oversimplification. Subtype designation of patient-derived models by a random forest transcriptional classifier aligned with prior diagnosis in 76% of unambiguous cases. In cases where there was disagreement, this often indicated potential alternative diagnosis, supported by a review of histological, molecular and clinical features. CONCLUSIONS: This robust classification informs the selection of the most appropriate models for all five histotypes. Following further refinement on larger training cohorts, the transcriptional classification may represent a useful tool to support the classification of new model systems of OC subtypes.

BACKGROUND: Gene expression-based subtyping has the potential to form a new paradigm for stratified treatment of colorectal cancer. However, current frameworks are based on the transcriptomic profiles of primary tumors, and metastatic heterogeneity is a challenge. Here we aimed to develop a de novo metastasis-oriented framework. METHODS: In total, 829 transcriptomic profiles from patients with colorectal cancer were analyzed, including primary tumors, liver metastases, and non-malignant liver samples. High-resolution microarray gene expression profiling was performed of 283 liver metastases from 171 patients treated by hepatic resection, including multiregional and/or multi-metastatic samples from each of 47 patients. A single randomly selected liver metastasis sample from each patient was used for unsupervised subtype discovery by nonnegative matrix factorization, and a random forest prediction model was trained to classify multi-metastatic samples, as well as liver metastases from two independent series of 308 additional patients. RESULTS: Initial comparisons with non-malignant liver samples and primary colorectal tumors showed a highly variable degree of influence from the liver microenvironment in metastases, which contributed to inter-metastatic transcriptomic heterogeneity, but did not define subtype distinctions. The de novo liver metastasis subtype (LMS) framework recapitulated the main distinction between epithelial-like and mesenchymal-like tumors, with a strong immune and stromal component only in the latter. We also identified biologically distinct epithelial-like subtypes originating from different progenitor cell types. LMS1 metastases had several transcriptomic features of cancer aggressiveness, including secretory progenitor cell origin, oncogenic addictions, and microsatellite instability in a microsatellite stable background, as well as frequent RAS/TP53 co-mutations. The poor-prognostic association of LMS1 metastases was independent of mutation status, clinicopathological variables, and current subtyping frameworks (consensus molecular subtypes and colorectal cancer intrinsic subtypes). LMS1 was also the least heterogeneous subtype in comparisons of multiple metastases per patient, and tumor heterogeneity did not confound the prognostic value of LMS1. CONCLUSIONS: We report the first large study of multi-metastatic gene expression profiling of colorectal cancer. The new metastasis-oriented subtyping framework showed potential for clinically relevant transcriptomic classification in the context of metastatic heterogeneity, and an LMS1 mini-classifier was constructed to facilitate prognostic stratification and further clinical testing.