Proactive Variant Effect Mapping to Accelerate Genetic Diagnosis for Pediatric Cardiac Arrest

While genetic testing is becoming standard of care for patients with potentially inherited cardiovascular disease, the prevalence of uncertain results severely limits its utility. One promising approach is to generate variant effect maps that report the function of all possible variants in a gene prospectively. The proactive clinical application of these maps is nascent, and requires careful integration with current American College of Medical Genetics guidelines for variant interpretation. Here, we describe three pediatric cases of cardiac arrest or sudden cardiac death with variants of uncertain significance in calmodulin genes. We demonstrate the prospective clinical utility of a calmodulin variant effect map to inform variant interpretation, and therefore diagnosis and family care, in each case. This study was approved by the Stanford University and Vanderbilt University Medical Center IRBs. Consent was waived based on low risk of de-identified retrospective data collection per the IRB.

Effect sizes of causal variants for gene expression and complex traits differ between populations

Despite the growing number of genome-wide association studies (GWAS) for complex traits, it remains unclear whether effect sizes of causal genetic variants differ between populations. In principle, effect sizes of causal variants could differ between populations due to gene-by-gene or gene-by-environment interactions. However, comparing causal variant effect sizes is challenging: it is difficult to know which variants are causal, and comparisons of variant effect sizes are confounded by differences in linkage disequilibrium (LD) structure between ancestries. Here, we develop a method to assess causal variant effect size differences that overcomes these limitations. Specifically, we leverage the fact that segments of European ancestry shared between European-American and admixed African-American individuals have similar LD structure, allowing for unbiased comparisons of variant effect sizes in European ancestry segments. We apply our method to two types of traits: gene expression and low-density lipoprotein cholesterol (LDL-C). We find that causal variant effect sizes for gene expression are significantly different between European-Americans and African-Americans; for LDL-C, we observe a similar point estimate although this is not significant, likely due to lower statistical power. Cross-population differences in variant effect sizes highlight the role of genetic interactions in trait architecture and will contribute to the poor portability of polygenic scores across populations, reinforcing the importance of conducting GWAS on individuals of diverse ancestries and environments.

MaveDB v2: a curated community database with over three million variant effects from multiplexed functional assays

A central problem in genomics is understanding the effect of individual DNA variants. Multiplexed Assays of Variant Effect (MAVEs) can help address this challenge by measuring all possible single nucleotide variant effects in a gene or regulatory sequence simultaneously. Here we describe MaveDB v2, which has become the database of record for MAVEs. MaveDB now contains a large fraction of published studies, comprising over two hundred datasets and three million variant effect measurements. We created tools and APIs to streamline data submission and access, transforming MaveDB into a hub for the analysis and dissemination of these impactful datasets.

Loss-of-function, gain-of-function and dominant-negative mutations have profoundly different effects on protein structure: implications for variant effect prediction

Most known pathogenic mutations occur in protein-coding regions of DNA and change the way proteins are made. Taking protein structure into account has therefore provided great insight into the molecular mechanisms underlying human genetic disease. While there has been much focus on how mutations can disrupt protein structure and thus cause a loss of function (LOF), alternative mechanisms, specifically dominant-negative (DN) and gain-of-function (GOF) effects, are less understood. Here, we have investigated the protein-level effects of pathogenic missense mutations associated with different molecular mechanisms. We observe striking differences between recessive vs dominant, and LOF vs non-LOF mutations, with dominant, non-LOF disease mutations having much milder effects on protein structure, and DN mutations being highly enriched at protein interfaces. We also find that nearly all computational variant effect predictors underperform on non-LOF mutations, even those based solely on sequence conservation. However, we do find that non-LOF mutations could potentially be identified by their tendency to cluster in space. Overall, our work suggests that many pathogenic mutations that act via DN and GOF mutations are likely being missed by current variant prioritisation strategies, but that there is considerable scope to improve computational predictions through consideration of molecular disease mechanisms.

A systematic genotype-phenotype map for missense variants in the human intellectual disability-associated gene GDI1

Next generation sequencing has become a common tool in the diagnosis of genetic diseases. However, for the vast majority of genetic variants that are discovered, a clinical interpretation is not available. Variant effect mapping allows the functional effects of large numbers of single amino acid variants to be characterized in parallel. Here, we employ a variant effect mapping framework, combining functional assays with machine learning, to assess the effects of 89% of all possible amino acid substitutions in the human intellectual disability-associated gene, GDI1. We show that the resulting variant effect map can be used to discriminate pathogenic from benign variants at levels of precision higher than those achieved by current computational prediction tools. Our variant effect map recovers known biochemical and structural features of GDI1 and reveals new structural regions which may be important for GDI1 function. We explore how this variant effect map can be used to aid in the interpretation of novel GDI1 variants as they are discovered, and to re-classify previously observed variants of unknown significance.

An enhanced variant effect predictor based on a deep generative model and the Born-Again Networks

The development of an accurate and reliable variant effect prediction tool is important for research in human genetic diseases. A large number of predictors have been developed towards this goal, yet many of these predictors suffer from the problem of data circularity. Here we present MTBAN (Mutation effect predictor using the Temporal convolutional network and the Born-Again Networks), a method for predicting the deleteriousness of variants. We apply a form of knowledge distillation technique known as the Born-Again Networks (BAN) to a previously developed deep autoregressive generative model, mutationTCN, to achieve an improved performance in variant effect prediction. As the model is fully unsupervised and trained only on the evolutionarily related sequences of a protein, it does not suffer from the problem of data circularity which is common across supervised predictors. When evaluated on a test dataset consisting of deleterious and benign human protein variants, MTBAN shows an outstanding predictive ability compared to other well-known variant effect predictors. We also offer a user-friendly web server to predict variant effects using MTBAN, freely accessible at http://mtban.kaist.ac.kr. To our knowledge, MTBAN is the first variant effect prediction tool based on a deep generative model that provides a user-friendly web server for the prediction of deleteriousness of variants.

Assessing computational variant effect predictors with a large prospective cohort

Computational predictors can help interpret pathogenicity of human genetic variants, especially for the majority of variants where no experimental data are available. However, because we lack a high-quality unbiased test set, identifying the best-performing predictors remains a challenge. To address this issue, we evaluated missense variant effect predictors using genotypes and traits from a prospective cohort. We considered 139 gene-trait combinations with rare-variant burden association based on at least one of four systematic studies using phenotypes and whole-exome sequences from ~200K UK Biobank participants. Using an evaluation set of 35,525 rare missense variants and the relevant associated traits, we assessed the correlation of participants' traits with scores derived from 20 computational variant effect predictors. We found that two predictors—VARITY and REVEL—outperformed all others according to multiple performance measures. We expect that this study will help in selecting variant effect predictors, for both research and clinical purposes, while providing an unbiased benchmarking strategy that can be applied to additional cohorts and predictors.

Predicting functional consequences of mutations using molecular interaction network features

Variant interpretation remains a central challenge for precision medicine. Missense variants are particularly difficult to understand as they change only a single amino acid in a protein sequence yet can have large and varied effects on protein activity. Numerous tools have been developed to identify missense variants with putative disease consequences from protein sequence and structure. However, biological function arises through higher order interactions among proteins and molecules within cells. We therefore sought to capture information about the potential of missense mutations to perturb protein interaction networks by integrating protein structure and interaction data. We developed 16 network-based annotations for missense mutations that provide orthogonal information to features classically used to prioritize variants. We then evaluated them in the context of a proven machine-learning framework for variant effect prediction across multiple benchmark datasets to demonstrate their potential to improve variant classification. Interestingly, network features resulted in larger performance gains for classifying somatic mutations than for germline variants, possibly due to different constraints on what mutations are tolerated at the cellular versus organismal level. Our results suggest that modeling variant potential to perturb context-specific interactome networks is a fruitful strategy to advance in silico variant effect prediction.