The AlphaMissense publication outlines how a variant of AlphaFold / DeepMind was used to predict missense variant pathogenicity. Supporting data on Zenodo include, for instance 70+M variants across hg19 and hg38 genome builds. The AlphaMissense package allows ready access to the data, downloading individual files to DuckDB databases for ready exploration and integration into R and Bioconductor worksflows.
Installation
Install the package from Bioconductor or GitHub, ensuring correct Bioconductor dependencies.
When the package is available on Bioconductor, use
if (!"BiocManager" %in% rownames(installed.packages()))
install.packages("BiocManager", repos = "https://cloud.R-project.org")
if (BiocManager::version() >= "3.19") {
BiocManager::install("AlphaMissenseR")
} else {
stop(
"'AlphaMissenseR' requires Bioconductor version 3.19 or later, ",
"install from GitHub?"
)
}
Use the pre-release or devel version with
if (!"remotes" %in% rownames(installed.packages()))
install.packages("remotes", repos = "https://cloud.R-project.org")
remotes::install_github(
"mtmorgan/AlphaMissenseR",
repos = BiocManager::repositories()
)
Load the library.
Next steps
Visit the introductory article to learn more about using this package. The visualization article shows how missense effects can be plotted on (AlphaFold or other) protein structures.