AlphaMissense

What AlphaMissense Is

AlphaMissense is a pathogenicity predictor developed by Google DeepMind, built on the AlphaFold protein structure prediction framework. It leverages deep knowledge of protein three-dimensional structure to predict whether a missense variant is likely to cause disease. By understanding how proteins fold and where specific amino acids sit within that fold, AlphaMissense can assess whether a substitution would disrupt the protein's structure or function.

The model was published in Science (Cheng et al., 2023) and has been shown to outperform most other missense predictors in independent benchmarks against ClinVar validation datasets.

Score Interpretation

AlphaMissense scores range from 0 to 1, with higher scores indicating a greater likelihood of pathogenicity. Variants are classified into three categories:

Strengths and Limitations

AlphaMissense's primary strength is that it incorporates protein three-dimensional structural information. While sequence-based tools like SIFT can only assess conservation at a position, AlphaMissense understands whether the amino acid sits in the protein core, on the surface, at an interaction interface, or near a catalytic site. It was trained on human population data and primate conservation data, giving it a human-specific perspective on pathogenicity.

The main limitation is scope: AlphaMissense only predicts impact for missense variants. It does not assess in-frame indels, splice variants, nonsense variants, or any non-coding variants. Its "Ambiguous" category covers a meaningful fraction of all possible missense variants where the model lacks confidence.