Skip to content

JARVIS: a comprehensive deep learning framework to prioritise non-coding variants in whole genomes

License

Notifications You must be signed in to change notification settings

astrazeneca-cgr-publications/JARVIS

Repository files navigation

JARVIS

Overview

  • JARVIS ("Junk" Annotation genome-wide Residual Variation Intolerance Score): a comprehensive deep learning framework to prioritise non-coding variants in whole genomes, using human-lineage purifying selection features and primary sequence context.

  • gwRVIS (genome-wide Residual Variation Intolerance Score): genome-wide intolerance to variation score

Publication:
Prioritizing non-coding regions based on human genomic constraint and sequence context with deep learning.
Vitsios et al., Nature Communications, March 8, 2021 https://doi.org/10.1038/s41467-021-21790-4


Installation instructions for JARVIS/gwRVIS modules

  • Python dependencies
conda create -n jarvis python=3.7 r r-devtools r-tidyverse 
conda config --add channels bioconda
conda config --add channels conda-forge
conda activate jarvis  


conda install --file requirements.txt  
  • R dependencies

install.packages(c("glm2", "glmnet", "lmridge", "plotmo", "pRoc", "ggplot2", "ggridges", "RColorBrewer")) devtools::install_github("thomasp85/patchwork")



Run

  • Instructions to generate the JARVIS and gwRVIS scores are available in the README.md file within modules.
  • Subsequent sub-folders may also contain their own README files with instructions to run them independently or for ad-hoc analyses.
  • Other folders and their sub-folders (such as ensembl/, gnomad/ and other_datasets/) are accompanied with README files and scripts to download and pre-process any other required datafiles that are not available in the JARVIS GitHub repositoy.



Data availability

JARVIS and gwRVIS scores, across the whole genome, are publicly available at the following location: http://jarvis.public.cgr.astrazeneca.com

All scores have been generated based on the hg19 human assembly version.