Speos: an ensemble graph representation learning framework to predict core gene candidates for complex diseases.

Understanding phenotype-to-genotype relationships is a grand challenge of 21st century biology with translational implications. The recently proposed "omnigenic" model postulates that effects of genetic variation on traits are mediated by core-genes and -proteins whose activities mechanistically influence the phenotype, whereas peripheral genes encode a regulatory network that indirectly affects phenotypes via core gene products. Here, we develop a positive-unlabeled graph representation-learning ensemble-approach based on a nested cross-validation to predict core-like genes for diverse diseases using Mendelian disorder genes for training. Employing mouse knockout phenotypes for external validations, we demonstrate that core-like genes display several key properties of core genes: Mouse knockouts of genes corresponding to our most confident predictions give rise to relevant mouse phenotypes at rates on par with the Mendelian disorder genes, and all candidates exhibit core gene properties like transcriptional deregulation in disease and loss-of-function intolerance. Moreover, as predicted for core genes, our candidates are enriched for drug targets and druggable proteins. In contrast to Mendelian disorder genes the new core-like genes are enriched for druggable yet untargeted gene products, which are therefore attractive targets for drug development. Interpretation of the underlying deep learning model suggests plausible explanations for our core gene predictions in form of molecular mechanisms and physical interactions. Our results demonstrate the potential of graph representation learning for the interpretation of biological complexity and pave the way for studying core gene properties and future drug development.

The Oncology Biomarker Discovery framework reveals cetuximab and bevacizumab response patterns in metastatic colorectal cancer.

Precision medicine has revolutionised cancer treatments; however, actionable biomarkers remain scarce. To address this, we develop the Oncology Biomarker Discovery (OncoBird) framework for analysing the molecular and biomarker landscape of randomised controlled clinical trials. OncoBird identifies biomarkers based on single genes or mutually exclusive genetic alterations in isolation or in the context of tumour subtypes, and finally, assesses predictive components by their treatment interactions. Here, we utilise the open-label, randomised phase III trial (FIRE-3, AIO KRK-0306) in metastatic colorectal carcinoma patients, who received either cetuximab or bevacizumab in combination with 5-fluorouracil, folinic acid and irinotecan (FOLFIRI). We systematically identify five biomarkers with predictive components, e.g., patients with tumours that carry chr20q amplifications or lack mutually exclusive ERK signalling mutations benefited from cetuximab compared to bevacizumab. In summary, OncoBird characterises the molecular landscape and outlines actionable biomarkers, which generalises to any molecularly characterised randomised controlled trial.

An arrhythmogenic metabolite in atrial fibrillation.

BACKGROUND: Long-chain acyl-carnitines (ACs) are potential arrhythmogenic metabolites. Their role in atrial fibrillation (AF) remains incompletely understood. Using a systems medicine approach, we assessed the contribution of C18:1AC to AF by analysing its in vitro effects on cardiac electrophysiology and metabolism, and translated our findings into the human setting. METHODS AND RESULTS: Human iPSC-derived engineered heart tissue was exposed to C18:1AC. A biphasic effect on contractile force was observed: short exposure enhanced contractile force, but elicited spontaneous contractions and impaired Ca2+ handling. Continuous exposure provoked an impairment of contractile force. In human atrial mitochondria from AF individuals, C18:1AC inhibited respiration. In a population-based cohort as well as a cohort of patients, high C18:1AC serum concentrations were associated with the incidence and prevalence of AF. CONCLUSION: Our data provide evidence for an arrhythmogenic potential of the metabolite C18:1AC. The metabolite interferes with mitochondrial metabolism, thereby contributing to contractile dysfunction and shows predictive potential as novel circulating biomarker for risk of AF.

Darier's disease exhibits a unique cutaneous microbial dysbiosis associated with inflammation and body malodour.

BACKGROUND: Darier's disease (DD) is a genodermatosis caused by mutations of the ATP2A2 gene leading to disrupted keratinocyte adhesion. Recurrent episodes of skin inflammation and infections with a typical malodour in DD indicate a role for microbial dysbiosis. Here, for the first time, we investigated the DD skin microbiome using a metabarcoding approach of 115 skin swabs from 14 patients and 14 healthy volunteers. Furthermore, we analyzed its changes in the context of DD malodour and the cutaneous DD transcriptome. RESULTS: We identified a disease-specific cutaneous microbiome with a loss of microbial diversity and of potentially beneficial commensals. Expansion of inflammation-associated microbes such as Staphylococcus aureus and Staphylococcus warneri strongly correlated with disease severity. DD dysbiosis was further characterized by abundant species belonging to Corynebacteria, Staphylococci and Streptococci groups displaying strong associations with malodour intensity. Transcriptome analyses showed marked upregulation of epidermal repair, inflammatory and immune defence pathways reflecting epithelial and immune response mechanisms to DD dysbiotic microbiome. In contrast, barrier genes including claudin-4 and cadherin-4 were downregulated. CONCLUSIONS: These findings allow a better understanding of Darier exacerbations, highlighting the role of cutaneous dysbiosis in DD inflammation and associated malodour. Our data also suggest potential biomarkers and targets of intervention for DD. Video Abstract.

Identification of genetic variants that impact gene co-expression relationships using large-scale single-cell data.

BACKGROUND: Expression quantitative trait loci (eQTL) studies show how genetic variants affect downstream gene expression. Single-cell data allows reconstruction of personalized co-expression networks and therefore the identification of SNPs altering co-expression patterns (co-expression QTLs, co-eQTLs) and the affected upstream regulatory processes using a limited number of individuals. RESULTS: We conduct a co-eQTL meta-analysis across four scRNA-seq peripheral blood mononuclear cell datasets using a novel filtering strategy followed by a permutation-based multiple testing approach. Before the analysis, we evaluate the co-expression patterns required for co-eQTL identification using different external resources. We identify a robust set of cell-type-specific co-eQTLs for 72 independent SNPs affecting 946 gene pairs. These co-eQTLs are replicated in a large bulk cohort and provide novel insights into how disease-associated variants alter regulatory networks. One co-eQTL SNP, rs1131017, that is associated with several autoimmune diseases, affects the co-expression of RPS26 with other ribosomal genes. Interestingly, specifically in T cells, the SNP additionally affects co-expression of RPS26 and a group of genes associated with T cell activation and autoimmune disease. Among these genes, we identify enrichment for targets of five T-cell-activation-related transcription factors whose binding sites harbor rs1131017. This reveals a previously overlooked process and pinpoints potential regulators that could explain the association of rs1131017 with autoimmune diseases. CONCLUSION: Our co-eQTL results highlight the importance of studying context-specific gene regulation to understand the biological implications of genetic variation. With the expected growth of sc-eQTL datasets, our strategy and technical guidelines will facilitate future co-eQTL identification, further elucidating unknown disease mechanisms.

Machine learning reveals STAT motifs as predictors for GR-mediated gene repression.

Glucocorticoids are potent immunosuppressive drugs, but long-term treatment leads to severe side-effects. While there is a commonly accepted model for GR-mediated gene activation, the mechanism behind repression remains elusive. Understanding the molecular action of the glucocorticoid receptor (GR) mediated gene repression is the first step towards developing novel therapies. We devised an approach that combines multiple epigenetic assays with 3D chromatin data to find sequence patterns predicting gene expression change. We systematically tested> 100 models to evaluate the best way to integrate the data types and found that GR-bound regions hold most of the information needed to predict the polarity of Dex-induced transcriptional changes. We confirmed NF-κB motif family members as predictors for gene repression and identified STAT motifs as additional negative predictors.

Predictive model of transcriptional elongation control identifies trans regulatory factors from chromatin signatures.

Promoter-proximal Polymerase II (Pol II) pausing is a key rate-limiting step for gene expression. DNA and RNA-binding trans-acting factors regulating the extent of pausing have been identified. However, we lack a quantitative model of how interactions of these factors determine pausing, therefore the relative importance of implicated factors is unknown. Moreover, previously unknown regulators might exist. Here we address this gap with a machine learning model that accurately predicts the extent of promoter-proximal Pol II pausing from large-scale genome and transcriptome binding maps and gene annotation and sequence composition features. We demonstrate high accuracy and generalizability of the model by validation on an independent cell line which reveals the model's cell line agnostic character. Model interpretation in light of prior knowledge about molecular functions of regulatory factors confirms the interconnection of pausing with other RNA processing steps. Harnessing underlying feature contributions, we assess the relative importance of each factor, quantify their predictive effects and systematically identify previously unknown regulators of pausing. We additionally identify 16 previously unknown 7SK ncRNA interacting RNA-binding proteins predictive of pausing. Our work provides a framework to further our understanding of the regulation of the critical early steps in transcriptional elongation.

A20 and the noncanonical NF-κB pathway are key regulators of neutrophil recruitment during fetal ontogeny.

Newborns are at high risk of developing neonatal sepsis, particularly if born prematurely. This has been linked to divergent requirements the immune system has to fulfill during intrauterine compared with extrauterine life. By transcriptomic analysis of fetal and adult neutrophils, we shed new light on the molecular mechanisms of neutrophil maturation and functional adaption during fetal ontogeny. We identified an accumulation of differentially regulated genes within the noncanonical NF-κB signaling pathway accompanied by constitutive nuclear localization of RelB and increased surface expression of TNF receptor type II in fetal neutrophils, as well as elevated levels of lymphotoxin α in fetal serum. Furthermore, we found strong upregulation of the negative inflammatory regulator A20 (Tnfaip3) in fetal neutrophils, which was accompanied by pronounced downregulation of the canonical NF-κB pathway. Functionally, overexpressing A20 in Hoxb8 cells led to reduced adhesion of these neutrophil-like cells in a flow chamber system. Conversely, mice with a neutrophil-specific A20 deletion displayed increased inflammation in vivo. Taken together, we have uncovered constitutive activation of the noncanonical NF-κB pathway with concomitant upregulation of A20 in fetal neutrophils. This offers perfect adaption of neutrophil function during intrauterine fetal life but also restricts appropriate immune responses particularly in prematurely born infants.

DeepSom: a CNN-based approach to somatic variant calling in WGS samples without a matched normal.

MOTIVATION: Somatic mutations are usually called by analyzing the DNA sequence of a tumor sample in conjunction with a matched normal. However, a matched normal is not always available, for instance, in retrospective analysis or diagnostic settings. For such cases, tumor-only somatic variant calling tools need to be designed. Previously proposed approaches demonstrate inferior performance on whole-genome sequencing (WGS) samples. RESULTS: We present the convolutional neural network-based approach called DeepSom for detecting somatic single nucleotide polymorphism and short insertion and deletion variants in tumor WGS samples without a matched normal. We validate DeepSom by reporting its performance on five different cancer datasets. We also demonstrate that on WGS samples DeepSom outperforms previously proposed methods for tumor-only somatic variant calling. AVAILABILITY AND IMPLEMENTATION: DeepSom is available as a GitHub repository at https://github.com/heiniglab/DeepSom. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

A proteome-scale map of the SARS-CoV-2-human contactome.

Understanding the mechanisms of coronavirus disease 2019 (COVID-19) disease severity to efficiently design therapies for emerging virus variants remains an urgent challenge of the ongoing pandemic. Infection and immune reactions are mediated by direct contacts between viral molecules and the host proteome, and the vast majority of these virus-host contacts (the 'contactome') have not been identified. Here, we present a systematic contactome map of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) with the human host encompassing more than 200 binary virus-host and intraviral protein-protein interactions. We find that host proteins genetically associated with comorbidities of severe illness and long COVID are enriched in SARS-CoV-2 targeted network communities. Evaluating contactome-derived hypotheses, we demonstrate that viral NSP14 activates nuclear factor κB (NF-κB)-dependent transcription, even in the presence of cytokine signaling. Moreover, for several tested host proteins, genetic knock-down substantially reduces viral replication. Additionally, we show for USP25 that this effect is phenocopied by the small-molecule inhibitor AZ1. Our results connect viral proteins to human genetic architecture for COVID-19 severity and offer potential therapeutic targets.

Network reconstruction for trans acting genetic loci using multi-omics data and prior information.

BACKGROUND: Molecular measurements of the genome, the transcriptome, and the epigenome, often termed multi-omics data, provide an in-depth view on biological systems and their integration is crucial for gaining insights in complex regulatory processes. These data can be used to explain disease related genetic variants by linking them to intermediate molecular traits (quantitative trait loci, QTL). Molecular networks regulating cellular processes leave footprints in QTL results as so-called trans-QTL hotspots. Reconstructing these networks is a complex endeavor and use of biological prior information can improve network inference. However, previous efforts were limited in the types of priors used or have only been applied to model systems. In this study, we reconstruct the regulatory networks underlying trans-QTL hotspots using human cohort data and data-driven prior information. METHODS: We devised a new strategy to integrate QTL with human population scale multi-omics data. State-of-the art network inference methods including BDgraph and glasso were applied to these data. Comprehensive prior information to guide network inference was manually curated from large-scale biological databases. The inference approach was extensively benchmarked using simulated data and cross-cohort replication analyses. Best performing methods were subsequently applied to real-world human cohort data. RESULTS: Our benchmarks showed that prior-based strategies outperform methods without prior information in simulated data and show better replication across datasets. Application of our approach to human cohort data highlighted two novel regulatory networks related to schizophrenia and lean body mass for which we generated novel functional hypotheses. CONCLUSIONS: We demonstrate that existing biological knowledge can improve the integrative analysis of networks underlying trans associations and generate novel hypotheses about regulatory mechanisms.

Probing cell identity hierarchies by fate titration and collision during direct reprogramming.

Despite the therapeutic promise of direct reprogramming, basic principles concerning fate erasure and the mechanisms to resolve cell identity conflicts remain unclear. To tackle these fundamental questions, we established a single-cell protocol for the simultaneous analysis of multiple cell fate conversion events based on combinatorial and traceable reprogramming factor expression: Collide-seq. Collide-seq revealed the lack of a common mechanism through which fibroblast-specific gene expression loss is initiated. Moreover, we found that the transcriptome of converting cells abruptly changes when a critical level of each reprogramming factor is attained, with higher or lower levels not contributing to major changes. By simultaneously inducing multiple competing reprogramming factors, we also found a deterministic system, in which titration of fates against each other yields dominant or colliding fates. By investigating one collision in detail, we show that reprogramming factors can disturb cell identity programs independent of their ability to bind their target genes. Taken together, Collide-seq has shed light on several fundamental principles of fate conversion that may aid in improving current reprogramming paradigms.

Tissue-specific multi-omics analysis of atrial fibrillation.

Genome-wide association studies (GWAS) for atrial fibrillation (AF) have uncovered numerous disease-associated variants. Their underlying molecular mechanisms, especially consequences for mRNA and protein expression remain largely elusive. Thus, refined multi-omics approaches are needed for deciphering the underlying molecular networks. Here, we integrate genomics, transcriptomics, and proteomics of human atrial tissue in a cross-sectional study to identify widespread effects of genetic variants on both transcript (cis-eQTL) and protein (cis-pQTL) abundance. We further establish a novel targeted trans-QTL approach based on polygenic risk scores to determine candidates for AF core genes. Using this approach, we identify two trans-eQTLs and five trans-pQTLs for AF GWAS hits, and elucidate the role of the transcription factor NKX2-5 as a link between the GWAS SNP rs9481842 and AF. Altogether, we present an integrative multi-omics method to uncover trans-acting networks in small datasets and provide a rich resource of atrial tissue-specific regulatory variants for transcript and protein levels for cardiovascular disease gene prioritization.

Genetic variation influencing DNA methylation provides insights into molecular mechanisms regulating genomic function.

We determined the relationships between DNA sequence variation and DNA methylation using blood samples from 3,799 Europeans and 3,195 South Asians. We identify 11,165,559 SNP-CpG associations (methylation quantitative trait loci (meQTL), P < 10-14), including 467,915 meQTL that operate in trans. The meQTL are enriched for functionally relevant characteristics, including shared chromatin state, High-throuhgput chromosome conformation interaction, and association with gene expression, metabolic variation and clinical traits. We use molecular interaction and colocalization analyses to identify multiple nuclear regulatory pathways linking meQTL loci to phenotypic variation, including UBASH3B (body mass index), NFKBIE (rheumatoid arthritis), MGA (blood pressure) and COMMD7 (white cell counts). For rs6511961 , chromatin immunoprecipitation followed by sequencing (ChIP-seq) validates zinc finger protein (ZNF)333 as the likely trans acting effector protein. Finally, we used interaction analyses to identify population- and lineage-specific meQTL, including rs174548 in FADS1, with the strongest effect in CD8+ T cells, thus linking fatty acid metabolism with immune dysregulation and asthma. Our study advances understanding of the potential pathways linking genetic variation to human phenotype.

Polygenic risk for coronary artery disease in the Scottish and English population.

BACKGROUND: Epidemiological studies have repeatedly observed a markedly higher risk for coronary artery disease (CAD) in Scotland as compared to England. Up to now, it is unclear whether environmental or genetic factors might explain this phenomenon. METHODS: Using UK Biobank (UKB) data, we assessed CAD risk, based on the Framingham risk score (FRS) and common genetic variants, to explore the respective contribution to CAD prevalence in Scotland (n = 31,963) and England (n = 317,889). We calculated FRS based on sex, age, body mass index (BMI), total cholesterol (TC), high density lipoprotein cholesterol (HDL-C), systolic blood pressure (SBP), antihypertensive medication, smoking status, and diabetes. We determined the allele frequency of published genome-wide significant risk CAD alleles and a weighted genetic risk score (wGRS) for quantifying genetic CAD risk. RESULTS: Prevalence of CAD was 16% higher in Scotland as compared to England (8.98% vs. 7.68%, P < 0.001). However, the FRS only predicted a marginally higher CAD risk (less than 1%) in Scotland (12.5 ± 10.5 vs.12.6 ± 10.6, P = 0.03). Likewise, the overall number of genome-wide significant variants affecting CAD risk (157.6 ± 7.7 and 157.5 ± 7.7; P = 0.12) and a wGRS for CAD (2.49 ± 0.25 in both populations, P = 0.14) were remarkably similar in the English and Scottish population. Interestingly, we observed substantial differences in the allele frequencies of individual risk variants. Of the previously described 163 genome-wide significant variants studied here, 35 variants had higher frequencies in Scotland, whereas 37 had higher frequencies in England (P < 0.001 each). CONCLUSIONS: Neither the traditional risk factors included in the FRS nor a genetic risk score (GRS) based on established common risk alleles explained the higher CAD prevalence in Scotland. However, we observed marked differences in the distribution of individual risk alleles, which emphasizes that even geographically and ethnically closely related populations may display relevant differences in the genetic architecture of a common disease.

scPower accelerates and optimizes the design of multi-sample single cell transcriptomic studies.

Single cell RNA-seq has revolutionized transcriptomics by providing cell type resolution for differential gene expression and expression quantitative trait loci (eQTL) analyses. However, efficient power analysis methods for single cell data and inter-individual comparisons are lacking. Here, we present scPower; a statistical framework for the design and power analysis of multi-sample single cell transcriptomic experiments. We modelled the relationship between sample size, the number of cells per individual, sequencing depth, and the power of detecting differentially expressed genes within cell types. We systematically evaluated these optimal parameter combinations for several single cell profiling platforms, and generated broad recommendations. In general, shallow sequencing of high numbers of cells leads to higher overall power than deep sequencing of fewer cells. The model, including priors, is implemented as an R package and is accessible as a web tool. scPower is a highly customizable tool that experimentalists can use to quickly compare a multitude of experimental designs and optimize for a limited budget.

Risk Prediction of Cardiovascular Events by Exploration of Molecular Data with Explainable Artificial Intelligence.

Cardiovascular diseases (CVD) annually take almost 18 million lives worldwide. Most lethal events occur months or years after the initial presentation. Indeed, many patients experience repeated complications or require multiple interventions (recurrent events). Apart from affecting the individual, this leads to high medical costs for society. Personalized treatment strategies aiming at prediction and prevention of recurrent events rely on early diagnosis and precise prognosis. Complementing the traditional environmental and clinical risk factors, multi-omics data provide a holistic view of the patient and disease progression, enabling studies to probe novel angles in risk stratification. Specifically, predictive molecular markers allow insights into regulatory networks, pathways, and mechanisms underlying disease. Moreover, artificial intelligence (AI) represents a powerful, yet adaptive, framework able to recognize complex patterns in large-scale clinical and molecular data with the potential to improve risk prediction. Here, we review the most recent advances in risk prediction of recurrent cardiovascular events, and discuss the value of molecular data and biomarkers for understanding patient risk in a systems biology context. Finally, we introduce explainable AI which may improve clinical decision systems by making predictions transparent to the medical practitioner.

A trans locus causes a ribosomopathy in hypertrophic hearts that affects mRNA translation in a protein length-dependent fashion.

BACKGROUND: Little is known about the impact of trans-acting genetic variation on the rates with which proteins are synthesized by ribosomes. Here, we investigate the influence of such distant genetic loci on the efficiency of mRNA translation and define their contribution to the development of complex disease phenotypes within a panel of rat recombinant inbred lines. RESULTS: We identify several tissue-specific master regulatory hotspots that each control the translation rates of multiple proteins. One of these loci is restricted to hypertrophic hearts, where it drives a translatome-wide and protein length-dependent change in translational efficiency, altering the stoichiometric translation rates of sarcomere proteins. Mechanistic dissection of this locus across multiple congenic lines points to a translation machinery defect, characterized by marked differences in polysome profiles and misregulation of the small nucleolar RNA SNORA48. Strikingly, from yeast to humans, we observe reproducible protein length-dependent shifts in translational efficiency as a conserved hallmark of translation machinery mutants, including those that cause ribosomopathies. Depending on the factor mutated, a pre-existing negative correlation between protein length and translation rates could either be enhanced or reduced, which we propose to result from mRNA-specific imbalances in canonical translation initiation and reinitiation rates. CONCLUSIONS: We show that distant genetic control of mRNA translation is abundant in mammalian tissues, exemplified by a single genomic locus that triggers a translation-driven molecular mechanism. Our work illustrates the complexity through which genetic variation can drive phenotypic variability between individuals and thereby contribute to complex disease.

Asc-1 regulates white versus beige adipocyte fate in a subcutaneous stromal cell population.

Adipose tissue expansion, as seen in obesity, is often metabolically detrimental causing insulin resistance and the metabolic syndrome. However, white adipose tissue expansion at early ages is essential to establish a functional metabolism. To understand the differences between adolescent and adult adipose tissue expansion, we studied the cellular composition of the stromal vascular fraction of subcutaneous adipose tissue of two and eight weeks old mice using single cell RNA sequencing. We identified a subset of adolescent preadipocytes expressing the mature white adipocyte marker Asc-1 that showed a low ability to differentiate into beige adipocytes compared to Asc-1 negative cells in vitro. Loss of Asc-1 in subcutaneous preadipocytes resulted in spontaneous differentiation of beige adipocytes in vitro and in vivo. Mechanistically, this was mediated by a function of the amino acid transporter ASC-1 specifically in proliferating preadipocytes involving the intracellular accumulation of the ASC-1 cargo D-serine.