Running VARPP-Rule: Modelling and Prediction for Pathogenic Variant Prediction in Rare Diseases

Overview

Content

The DataLad data set contains the follwing structure:

+---genome-data
+---pre-trained-models
+---patientVcf
+---liftover
+---scratch
+---README.org

Models

The DataLad data set can be seen as a software package, that comes with the necessary data and code pre-configured. It contains two docker images:

1. gatk image: to run the picard LiftoverVcf tool, which allows to create ‘lift’ hg19 annotation of a patient .vcf file to h38. This is necessary, as the model data is annotated with the hg38 data.
2. varpp-predict-utils image: this docker image contains the scripts we need for running VARPP-RuleFit, predict or a combination of the two for a combined project workflow.

Quick start: Set up the project

In order to run the model for a project, run the following in your desired folder location on the command line:

conda activate datalad # in case you have datalad installed on the system without conda, you do not need this step
git clone https://github.com/Hobbeist/varpp-project-datalad.git
cd varpp-project-datalad