Search

Personalized genome sequencing has revealed millions of genetic differences between individuals, but our understanding of their clinical relevance remains largely incomplete. To systematically decipher the effects of human genetic variants, we obtained whole-genome sequencing data for 809 individuals from 233 primate species and identified 4.3 million common protein-altering variants with orthologs in humans. We show that these variants can be inferred to have nondeleterious effects in humans based on their presence at high allele frequencies in other primate populations. We use this resource to classify 6% of all possible human protein-altering variants as likely benign and impute the pathogenicity of the remaining 94% of variants with deep learning, achieving state-of-the-art accuracy for diagnosing pathogenic variants in patients with genetic diseases., INTRODUCTION: Millions of people have received genome and exome sequencing to date, a collective effort that has illuminated for the first time the vast catalog of small genetic differences that distinguish us as individuals within our species. However, the effects of most of these genetic variants remain unknown, limiting their clinical utility and actionability. New approaches that can accurately discern disease-causing from benign mutations and interpret genetic variants on a genome-wide scale would constitute a meaningful initial step towards realizing the potential of personalized genomic medicine. RATIONALE: As a result of the short evolutionary distance between humans and nonhuman primates, our proteins share near-perfect amino acid sequence identity. Hence, the effects of a protein-altering mutation found in one species are likely to be concordant in the other species. By systematically cataloging common variants of nonhuman primates, we aimed to annotate these variants as being unlikely to cause human disease as they are tolerated by natural selection in a closely related species. Once collected, the resulting resource may be applied to infer the effects of unobserved variants across the genome using machine learning. RESULTS: Following the strategy outlined above we obtained whole-genome sequencing data for 809 individuals from 233 primate species and cataloged 4.3 million common missense variants. We confirmed that human missense variants seen in at least one nonhuman primate species were annotated as benign in the ClinVar clinical variant database in 99% of cases. By contrast, common variants from mammals and vertebrates outside the primate lineage were substantially less likely to be benign in the ClinVar database (71 to 87% benign), restricting this strategy to nonhuman primates. Overall, we reclassified more than 4 million human missense variants of previously unknown consequence as likely benign, resulting in a greater than 50-fold increase in the nu

[INTRODUCTION] Genome-wide association studies (GWASs) have identified thousands of common genetic variants that are predictive of common disease susceptibility, but these variants individually have mild effects on disease owing to the effects of natural selection. By contrast, rare genetic variants can have large effects on common disease risk, but their use in genetic risk prediction has been limited to date owing to the difficulty of distinguishing pathogenic from benign variants and estimating the magnitude of their effects., [RATIONALE] PrimateAI-3D is a three-dimensional convolutional neural network for missense variant–effect prediction, which was trained with common genetic variants from the population sequencing of 233 primate species. By applying this method to estimate the pathogenicity of rare coding variants in 454,712 UK Biobank individuals, we aimed to improve rare-variant association tests and genetic risk prediction for common diseases and complex traits., [RESULTS] We performed rare-variant burden tests for 90 well-powered, clinically relevant phenotypes in the UK Biobank exome dataset. Stratifying missense variants with PrimateAI-3D greatly improved gene discovery, revealing 73% more significant gene-phenotype associations (false discovery rate <0.05) compared with not using PrimateAI-3D. When benchmarked against prior studies, gene-phenotype pairs identified with our method were better supported by orthogonal genetic evidence from GWAS and genes from related Mendelian disorders. In addition, PrimateAI-3D scores showed the strongest correlation among existing variant interpretation algorithms for predicting the quantitative effects of rare variants on continuous clinical phenotypes. Having validated our method for finding gene-phenotype relationships, we next constructed a rare-variant polygenic risk score (PRS) model by combining the rare-variant genes for each phenotype, weighting variants by their PrimateAI-3D prediction score and the direction and effect size of each associated gene. For comparison, we constructed common-variant PRS models and evaluated the performance of the two models for genetic risk prediction in a withheld-test subset of the cohort. Although common variants better explained overall population variance, rare-variant PRSs had more power at the ends of the distribution to identify individuals at the greatest risk for disease, and thus may be more relevant for population genetic screening and risk management. By contrast to common-variant PRS models derived from European populations that show poor generalization to non-Europeans, rare-variant PRSs were substantially more portable to different cohorts and ancestry groups that were not seen during model training. Moreover, because they incorporate orthogonal information from nonoverlapping sets of variants, we combined rare- and common-variant PRS models into a unified model and observed further improvement in genetic risk prediction for common d, [CONCLUSION] Understanding the impact of rare variants in common diseases is of prime interest for both precision medicine and the discovery of drug targets. By leveraging advances in variant effect prediction, we have demonstrated major improvements in rare-variant burden testing and genetic risk prediction. Notably, we observed that nearly all individuals carried at least one rare penetrant variant for the phenotypes we examined, demonstrating the utility of personal genome sequencing for otherwise healthy individuals in the general population.

[INTRODUCTION] Millions of people have received genome and exome sequencing to date, a collective effort that has illuminated for the first time the vast catalog of small genetic differences that distinguish us as individuals within our species. However, the effects of most of these genetic variants remain unknown, limiting their clinical utility and actionability. New approaches that can accurately discern disease-causing from benign mutations and interpret genetic variants on a genome-wide scale would constitute a meaningful initial step towards realizing the potential of personalized genomic medicine., [RATIONALE] As a result of the short evolutionary distance between humans and nonhuman primates, our proteins share near-perfect amino acid sequence identity. Hence, the effects of a protein-altering mutation found in one species are likely to be concordant in the other species. By systematically cataloging common variants of nonhuman primates, we aimed to annotate these variants as being unlikely to cause human disease as they are tolerated by natural selection in a closely related species. Once collected, the resulting resource may be applied to infer the effects of unobserved variants across the genome using machine learning., [RESULTS] Following the strategy outlined above we obtained whole-genome sequencing data for 809 individuals from 233 primate species and cataloged 4.3 million common missense variants. We confirmed that human missense variants seen in at least one nonhuman primate species were annotated as benign in the ClinVar clinical variant database in 99% of cases. By contrast, common variants from mammals and vertebrates outside the primate lineage were substantially less likely to be benign in the ClinVar database (71 to 87% benign), restricting this strategy to nonhuman primates. Overall, we reclassified more than 4 million human missense variants of previously unknown consequence as likely benign, resulting in a greater than 50-fold increase in the number of annotated missense variants compared to existing clinical databases. To infer the pathogenicity of the remaining missense variants in the human genome, we constructed PrimateAI-3D, a semisupervised 3D-convolutional neural network that operates on voxelized protein structures. We trained PrimateAI-3D to separate common primate variants from matched control variants in 3D space as a semisupervised learning task. We evaluated the trained PrimateAI-3D model alongside 15 other published machine learning methods on their ability to distinguish between benign and pathogenic variants in six different clinical benchmarks and demonstrated that PrimateAI-3D outperformed all other classifiers in each of the tasks., [CONCLUSION] Our study addresses one of the key challenges in the variant interpretation field, namely, the lack of sufficient labeled data to effectively train large machine learning models. By generating the most comprehensive primate sequencing dataset to date and pairing this resource with a deep learning architecture that leverages 3D protein structures, we were able to achieve meaningful improvements in variant effect prediction across multiple clinical benchmarks.

Understanding life on the molecular level requires knowledge about protein-protein interactions (PPIs), but experimental data is limited and complex, rendering computational methods indispensable. We first analyzed PPI interfaces in all known PPI structures and found surprising differences in binding modes. Then we developed a method based on artificial neural networks that predicts from sequence which amino acids interact in a PPI. Another new state-of-the-art method determines whether two proteins interact. Third, addressing the problem of missing protein annotations in network-based analysis, we developed a function predictor that relies only on sequential similarity to proteins with known annotations., Um Leben auf molekularer Ebene zu verstehen brauchen wir Wissen über Protein-Protein Interaktionen (PPIs). Experimentelle Daten sind jedoch rar und kompliziert und machen computergestützte Methoden unverzichtbar. Diese Arbeit analysierte Proteinberührungsflachen in allen bekannt PPI Strukturen und fanden überraschende Unregelmäßigkeiten im Bindungsverhalten. Dann trainierten wir künstliche Neuronale Netze die nur anhand der Proteinsequenzen vorhersagen welche Aminosäuren in einer PPI interagieren. Eine weitere Methode bestimmt ob zwei Proteine überhaupt interagieren. Zuletzt adressierten wir fehlende Annotationen bei der netzwerkbasierten Analyse von Proteinfunktionen und fanden eine neue Methode zur Proteinfunktionsvorhersage, die nur auf Sequenzähnlichkeit zu bereits annotierten Proteinen beruht.

Understanding life on the molecular level requires knowledge about protein-protein interactions (PPIs), but experimental data is limited and complex, rendering computational methods indispensable. We first analyzed PPI interfaces in all known PPI structures and found surprising differences in binding modes. Then we developed a method based on artificial neural networks that predicts from sequence which amino acids interact in a PPI. Another new state-of-the-art method determines whether two proteins interact. Third, addressing the problem of missing protein annotations in network-based analysis, we developed a function predictor that relies only on sequential similarity to proteins with known annotations., Um Leben auf molekularer Ebene zu verstehen brauchen wir Wissen über Protein-Protein Interaktionen (PPIs). Experimentelle Daten sind jedoch rar und kompliziert und machen computergestützte Methoden unverzichtbar. Diese Arbeit analysierte Proteinberührungsflachen in allen bekannt PPI Strukturen und fanden überraschende Unregelmäßigkeiten im Bindungsverhalten. Dann trainierten wir künstliche Neuronale Netze die nur anhand der Proteinsequenzen vorhersagen welche Aminosäuren in einer PPI interagieren. Eine weitere Methode bestimmt ob zwei Proteine überhaupt interagieren. Zuletzt adressierten wir fehlende Annotationen bei der netzwerkbasierten Analyse von Proteinfunktionen und fanden eine neue Methode zur Proteinfunktionsvorhersage, die nur auf Sequenzähnlichkeit zu bereits annotierten Proteinen beruht.