Ross Stewart

Obtaining a precise genetic tuberous sclerosis complex diagnosis is a challenge as many missense TSC2 variants are variants of uncertain significance. Variants of uncertain significance in TSC2 have been resolved by one-at-a-time functional assays, but these assays cannot scale to the 3634 TSC2 missense variants of uncertain significance observed so far. To address this challenge, we use massively parallel sequencing to measure the steady-state abundance of almost 9000 TSC2 missense variants and develop an mTOR pathway activity assay using genome editing and cell sorting to generate activity scores for 391 missense variants. We observe that 1256 of 8864 (14.17%) missense variants assayed have altered TSC2 abundance, and 69 of 391 (17.65%) missense variants assayed have altered mTOR pathway activity. Calibration and integration of these data into classification of variants identified in a clinical cohort putatively reclassifies 212 of 276 (76.8%) TSC2 missense variants of uncertain significance. These datasets will lead to improved genetic diagnosis of tuberous sclerosis complex with potential positive impacts on the clinical management of patients and their families.

Germline pathogenic variants in RAD51D and XRCC2 , which encode RAD51 paralogs that form a heterodimer within the BCDX2 complex, confer increased cancer risk and homologous recombination deficiency. However, most RAD51D and XRCC2 variants in ClinVar are classified as variants of uncertain significance (VUS), limiting clinical utility. Here, we applied saturation genome editing (SGE) to measure the effects of 5,412 RAD51D and 3,743 XRCC2 variants on cellular fitness and additionally assessed 2,876 and 2,069 variants in these genes for effects on RNA expression. Fitness scores discriminated pathogenic from benign variants with near-perfect accuracy (AUC=0.994 for RAD51D ; AUC=1.000 for XRCC2 ). Integration of RNA expression data revealed RAD51D , but not XRCC2 , is exceptionally sensitive to splice-altering variation, with 24% of RAD51D loss-of-function missense variants acting through RNA-mediated mechanisms compared to only 5% in XRCC2 . These SGE datasets provide strong, splice-resolved functional evidence to support variant classification across both genes.

The utility of clinical genetic testing is limited because around 90% of missense variants in ClinVar remain of uncertain clinical significance. Variant effect predictors (VEPs) can score any missense variant, potentially empowering variant classification. Realizing this potential requires calibration to translate predictor scores into evidence. However, genome-wide calibration ignores predictor heterogeneity across genes, causing evidence misassignment. We developed an automated, data-adaptive framework that optimizes two complementary approaches: gene-specific calibration for genes with enough variants for calibration, and domain-aggregate calibration for other disease-associated genes, which groups variants from protein domains with similar predictor score distributions for calibration. Applied to three predictors across 2,769 genes, this framework assigned evidence to 10.6% more variants on average while generally improving evidence accuracy compared to genome-wide calibration. These calibrations and the resulting calibrated computational evidence are available through the PredictMD portal. Our framework substantially increases the clinical utility of VEPs for variant classification.

The utility of clinical genetic testing is limited because around 90% of missense variants in ClinVar remain of uncertain clinical significance. Variant effect predictors (VEPs) can score any missense variant, potentially empowering variant classification. Realizing this potential requires calibration to translate predictor scores into evidence. However, genome-wide calibration ignores predictor heterogeneity across genes, causing evidence misassignment. We developed an automated, data-adaptive framework that optimizes two complementary approaches: gene-specific calibration for genes with enough variants for calibration, and domain-aggregate calibration for other disease-associated genes, which groups variants from protein domains with similar predictor score distributions for calibration. Applied to three predictors across 2,769 genes, this framework assigned evidence to 10.6% more variants on average while generally improving evidence accuracy compared to genome-wide calibration. These calibrations and the resulting calibrated computational evidence are available through the PredictMD portal. Our framework substantially increases the clinical utility of VEPs for variant classification.

Over 90% of missense variants across ∼4,000 disease-associated genes are variants of uncertain significance (VUS). Experimental variant effect measurements provide critical evidence about pathogenicity and inform disease biology, but most variants lack data and clinical translation has been limited. The Impact of Genomic Variation on Function Consortium generated experimental data for 62,215 variants across ten genes using multiplexed assays and 1,407 variants across 163 genes using arrayed assays, curated 193,139 additional community-generated variant effect measurements across 30 additional genes, and developed automated calibration methods for translating experimental data and variant effect predictions into clinical evidence. To reduce current VUS, we developed a scalable workflow using only experimental and predictive evidence, enabling reclassification of 75% of the 16,115 VUS in these genes as pathogenic or benign with 90,000 unobserved variants; 62% had enough evidence to be “preclassified” as pathogenic or benign. We validated our data, evidence and classifications using All of Us and created interactive resources to enable clinical use of the calibrated data. Thus, for 40 genes, representing 1% of the clinical genome, we resolve most existing VUS and future variants, illustrating how systematic use of scalable evidence can empower genomic medicine.

Disruption of protein–protein interactions (PPIs) is a major mechanism of a variant’s deleterious effect. Computational tools are needed to assess such variants at scale, yet existing predictors rarely consider loss of specific interactions, particularly when variants perturb binding interfaces without significantly affecting protein stability. To address this problem, we present MutPred-PPI, a graph attention network that predicts interaction-specific (edgetic) effects of missense variants by operating on AlphaFold 3-based protein complex contact graphs with protein language model embeddings imposed upon nodes. We systematically evaluated our model with stringent group cross-validation as well as benchmark data recently collected within the IGVF Consortium. MutPred-PPI outperformed all baseline methods across all evaluation criteria, achieving an AUC of 0.85 on seen proteins and 0.72 on previously unseen proteins in cross-validation, demonstrating strong generalizability despite scarce training data. To demonstrate biomedical relevance, we applied MutPred-PPI to variants from ClinVar, HGMD, COSMIC, gnomAD, and two de novo neurodevelopmental disorder-linked datasets. Disease-associated variants from Clin-Var and HGMD showed strong enrichment for both quasi-null and edgetic effects, whereas population variants from gnomAD increasingly preserved interactions with higher allele frequencies. Notably, we observed a strong edgetic disruption signature in highly recurrent cancer variants from both the full COSMIC dataset and a subset of variants from oncogenes. Recurrent tumor suppressor gene variants and autism spectrum disorder-associated variants exhibited moderate quasi-null enrichment, whilst neurodevelopmental disorder-linked variants showed a weak edgetic disruption signature. These results indicate distinct PPI perturbation mechanisms across disease types and show that MutPred-PPI captures functionally relevant molecular effects of pathogenic variants.