# AlphaMissenseR: an integrated framework for investigating missense mutations in human protein-coding genes

## TL;DR

AlphaMissenseR is an R package that provides tools to analyze and visualize predictions of missense mutation pathogenicity from the AlphaMissense AI model.

## Contribution

AlphaMissenseR introduces an integrated framework for analyzing, validating, and benchmarking missense mutation predictions in human genes.

## Key findings

- AlphaMissenseR enables reproducible access to AlphaMissense predictions for all possible missense mutations in the human proteome.
- The package supports multi-level visualization of variant pathogenicity in genome browsers and on 3D protein structures.
- It integrates with clinical and experimental databases for benchmarking variant effect prediction methods.

## Abstract

AlphaMissense is an AI model from Google DeepMind that predicts the pathogenicity of every possible missense mutation in the human proteome. We present AlphaMissenseR, an R/Bioconductor package that facilitates performant and reproducible access to these predictions and that provides functionality for analysis, visualization, validation, and benchmarking. AlphaMissenseR integrates with Bioconductor facilities for genomic region analysis, and provides multi-level visualization and interactive exploration of variant pathogenicity in a genome browser and on 3D protein structures. In addition, AlphaMissenseR integrates with major clinical and experimental variant databases for contrasting predicted and clinically derived pathogenicity scores, and for systematic benchmarking of existing and new variant effect prediction methods across a large collection of deep mutational scanning assays.

AlphaMissense data resources are distributed under the CC-BY 4.0 license and the AlphaMissenseR package is available from Bioconductor (https://bioconductor.org/packages/AlphaMissenseR) under the Artistic 2.0 license.