Accurate proteome-wide missense variant effect prediction with AlphaMissense

2023/09/19 by Jun Cheng, Guido Novati, Joshua Pan +13 · 103 citations
Biochemistry, Genetics and Molecular Biology · #Genetic Associations and Epidemiology #Genomic variations and chromosomal abnormalities #Genomics and Rare Diseases

paper · doi:10.1126/science.adg7492

openalex publication_date 2023/09/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/31

Abstract

The vast majority of missense variants observed in the human genome are of unknown clinical significance. We present AlphaMissense, an adaptation of AlphaFold fine-tuned on human and primate variant population frequency databases to predict missense variant pathogenicity. By combining structural context and evolutionary conservation, our model achieves state-of-the-art results across a wide range of genetic and experimental benchmarks, all without explicitly training on such data. The average pathogenicity score of genes is also predictive for their cell essentiality, capable of identifying short essential genes that existing statistical approaches are underpowered to detect. As a resource to the community, we provide a database of predictions for all possible human single amino acid substitutions and classify 89% of missense variants as either likely benign or likely pathogenic.

Citations

• Rare-Variant Association Analysis: Study Designs and Statistical Tests
• REVEL: An Ensemble Method for Predicting the Pathogenicity of Rare Missense Variants
• Yehudi: An Orchestrated System for the Interoperability of Urban Data and Models.
• The mutational constraint spectrum quantified from variation in 141,456 humans
• Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences
• Highly accurate protein structure prediction with AlphaFold
• Highly accurate protein structure prediction for the human proteome
• КОМПЬЮТЕРНАЯ ТОМОГРАФИЯ – ОСНОВА ОПТИМИЗАЦИИ МИНИТОРАКОТОМНОГО ДОСТУПА
• Ensembl 2022
• Tranception: protein fitness prediction with autoregressive transformers and inference-time retrieval
• Evolutionary-scale prediction of atomic-level protein structure with a language model

Cited by

• CATVariant: a web server for integrated protein variant interpretation across sequence, structure, population, and clinical evidence
• FASTK post-transcriptional regulators – a ‘FAST-tracK’ in mitochondrial gene expression
• Generalizable and scalable protein stability prediction with rewired protein generative models
• Review of protein structure-based analyses that illuminate plant stress mechanisms
• Genotype–phenotype correlations and clinical outcomes of genetic TRPC6 podocytopathies
• Reassessing the role of the p.(Arg304Gln) missense AIP variant in pituitary tumorigenesis
• Variant effect predictor correlation with functional assays is reflective of clinical classification performance
• Plants tolerate substantial rates of plastid mistranslation via regulated proteostasis
• Analysis of 470,000 exome-sequenced UK biobank participants identifies genes containing rare variants which confer dementia risk
• ARID2-related disorder: further delineation of the clinical phenotype of 27 novel individuals and description of an epigenetic signature
• HapScoreDB: a database of protein language model functional scores for haplotype-resolved protein sequences
• Guiding questions to avoid data leakage in biological machine learning applications
• Analysis of whole genome sequence data shows association of Alzheimer’s disease with rare coding variants in ABCA7 , PSEN1 , SORL1 and TREM2
• Host control of persistent Epstein–Barr virus infection
• Functional diversification of the cephalopod proteome by RNA-editing
• Predicting the functional impact of single nucleotide variants in Drosophila melanogaster with FlyCADD
• acmgscaler : an R package and Colab for standardized gene-level variant effect score calibration within the ACMG/AMP framework
• Functional residuomics – analyzing how missense mutations impact cellular systems
• AlphaGenome: advancing regulatory variant effect prediction with a unified DNA sequence model
• The proteome of the late Middle Pleistocene Harbin individual
• Analysis of more than 400,000 women provides case-control evidence for BRCA1 and BRCA2 variant classification
• SSEmb: A joint embedding of protein sequence and structure enables robust variant effect predictions
• Functional Screen of Wilson Disease ATP7B Variants Reveals Residual Transport Activities
• CGBench: Benchmarking Language Model Scientific Reasoning for Clinical Genetics Research
• Protenix-Mini+: efficient structure prediction model with scalable pairformer
• Exome-Wide Analysis Identifies a Rare EXD3 Missense Variant Associated With Diabetic Kidney Disease
• Proteome-wide model for human disease genetics
• De novo protein-coding gene variants in developmental stuttering
• Characterization of prevalent genetic variants in the Estonian Biobank body-mass index GWAS
• Reconstructing biological molecules with help from video gamers
• VN-EGNN: E(3)- and SE(3)-Equivariant Graph Neural Networks with Virtual Nodes Enhance Protein Binding Site Identification
• Mismatch repair deficiency drives malignant progression and alters the tumor immune microenvironment in glioblastoma models
• Proteomics—The State of the Field: The Definition and Analysis of Proteomes Should Be Based in Reality, Not Convenience
• RTN4IP1 mutation and endocrine failure: clinical features and possible benefits of coenzyme Q10
• Rare variants in BMAL1 are associated with a neurodevelopmental syndrome
• From data to discovery: leveraging big data in plant natural products biosynthesis research
• Fine-tuning protein language models to understand the functional impact of missense variants
• Making sense of missense: challenges and opportunities in variant pathogenicity prediction
• PRKN-linked familial Parkinson’s disease: cellular and molecular mechanisms of disease-linked variants
• Alternate RNA decoding results in stable and abundant proteins in mammals
• Computing the Human Interactome
• Evaluating Agentic Bioinformatics through Function, Evidence, and Validation
• Genomic variation drives plant flavor diversification
• Structures of the human adult muscle-type nicotinic receptor in resting and desensitized states
• Genome modeling and design across all domains of life with Evo 2
• From Likelihood to Fitness: Improving Variant Effect Prediction in Protein and Genome Language Models
• In vivo adenine base editing ameliorates Dravet syndrome phenotypes in a mouse model
• TRPtracker: a community database for monitoring praziquantel sensitivity at TRPMPZQ variants
• Structural modeling and functional characterization of a novel gain-of-function TLR8 variant causing severe inflammatory syndrome
• Hotspots of human mutation point to clonal expansions in spermatogonia
• PLM-interact: extending protein language models to predict protein-protein interactions
• Recessive, pathogenic AARS1 variants display variable loss-of-function and dominant-negative effects
• Massively parallel quantification of mutational impact on IAPP amyloid formation
• DiffInvex identifies evolutionary shifts in driver gene repertoires during tumorigenesis and chemotherapy
• A Perspective on Artificial Intelligence for Molecular Pathologists
• Colocalization of eQTLs With Type 2 Diabetes and Glycemic Traits Using Whole-Genome Sequences in Diverse Populations From the NHLBI Trans-Omics in Precision Medicine (TOPMed) Program
• Monoallelic POLR3A Variants Cause Early‐Onset Peripheral Neuropathy
• Pathogenic variants in the cohesin loader subunit MAU2 underlie a distinct Cornelia de Lange Syndrome subtype
• Dynamicasome: a molecular dynamics-guided and AI-driven pathogenicity prediction catalogue for all genetic mutations
• Reinforcement learning-based dynamic ensemble for missense variant effect prediction and tiered prioritization of VUS
• Talin-1 variants associated with spontaneous coronary artery dissection (SCAD) highlight how even subtle changes in multi-functional scaffold proteins can manifest in disease
• Fe( ii ) and 2-oxoglutarate-dependent dioxygenases for natural product synthesis: molecular insights into reaction diversity
• A Systematic Review on the Generative AI Applications in Human Medical Genomics
• Supervised learning of protein variant effects across large‐scale mutagenesis datasets
• Structural and functional brain asymmetry in relation to heterogeneous causes of situs inversus totalis
• Deep learning applications advance plant genomics research
• Molecular representation learning: cross-domain foundations and future Frontiers
• Large-scale genomic investigation of pediatric cholestasis reveals a novel hepatorenal ciliopathy caused by PSKH1 mutations
• Genomic heterogeneity inflates the performance of variant pathogenicity predictions
• Learning the language of protein-protein interactions
• MAVISp: A Modular Structure-Based Framework for Protein Variant Effects
• AlphaRING-X: accurate interpretation of missense variant deleteriousness based on protein structural stability
• Calibrated Variant Effect Prediction at the Residue Level Using Conditional Score Distributions
• Nucleotide dependency analysis of genomic language models detects functional elements
• PubMind: Literature-Based Genetic Variant Extraction and Functional Annotation Using Large Language Models
• The structural context of mutations in proteins predicts their effect on antibiotic resistance
• Domain-specific mechanisms of YAP1 variants in ocular coloboma revealed by in-vitro and organoid studies
• Genetic analysis of neurodegenerative diseases
• Mutations within the predicted fragment-binding region of FAM83G/SACK1G abolish its interaction with the Ser/Thr kinase CK1α
• Advancing regulatory variant effect prediction with AlphaGenome
• The neuropathy-linked protein TECPR2 is a Rab5 effector that regulates cargo recycling from early endosomes
• Protenix-Mini: Efficient Structure Predictor via Compact Architecture, Few-Step Diffusion and Switchable pLM
• The mutagenic forces shaping the genomes of lung cancer in never smokers
• Why variant effect predictors and multiplexed assays agree and disagree
• Predicting the functional impact of single nucleotide variants in Drosophila melanogaster with FlyCADD
• Accurate variant effect estimation in FACS-based deep mutational scanning data with Lilace
• The Reactome Pathway Knowledgebase 2024
• Guidelines for releasing a variant effect predictor
• acmgscaler: An R package and Colab for standardised gene-level variant effect score calibration within the ACMG/AMP framework
• Variant Set Distillation
• Geographic and age variations in mutational processes in colorectal cancer
• BMFM-DNA: A SNP-aware DNA foundation model to capture variant effects
• Novel Variants in VARS2 Demonstrate the Phenotypic Variability of a Rare Mitochondriopathy That Responds to Valine Supplementation
• Recent Advances in Genome Editing and Bioinformatics: Addressing Challenges in Genome Editing Implementation and Genome Sequencing
• Assessing variant effect predictors and disease mechanisms in intrinsically disordered proteins
• Rare Variant Analyses in Ancestrally Diverse Cohorts Reveal Novel ADHD Risk Genes
• Accurate Identification and Mechanistic Evaluation of Pathogenic Missense Variants with Rhapsody-2
• SafeGenes: Evaluating the Adversarial Robustness of Genomic Foundation Models
• Bonds and bytes: The odyssey of structural biology
• Benchmarking AlphaMissense pathogenicity predictions against cystic fibrosis variants
• G2P Explorer: A Native iOS Framework for Residue-Level Genomics to Proteomics Visualization and Structural Variant Interpretation
• DeepBD: A Grounded Agentic Workflow for Variant Prioritization and Diagnosis of Genetic Birth Defects
• AnnotateMissense: a genome-wide annotation and benchmarking framework for missense pathogenicity prediction
• Classification of SARS-CoV-2 Variants through The Epistatical Circos Plots with Convolutional Neural Networks
• Variant Classification Using Proteomics‐Informed Large Language Models Increases Power of Rare Variant Association Studies and Enhances Target Discovery
• Deep Learning-Driven Protein Structure Prediction and Design: Key Model Developments by Nobel Laureates and Multi-Domain Applications

Related

• Making sense of missense: challenges and opportunities in variant pathogenicity prediction
• AnnotateMissense: a genome-wide annotation and benchmarking framework for missense pathogenicity prediction
• Genome-wide prediction of disease variant effects with a deep protein language model
• Proteome-wide model for human disease genetics
• AlphaRING-X: accurate interpretation of missense variant deleteriousness based on protein structural stability