FINDOR

Functionally Informed Novel Discovery of Risk Loci

Description

This tool is designed to improve GWAS power for polygenic traits. For details of methodology please see Kichaev et al. (bioRxiv).

Required data

The core of the methodology relies on stratifying SNPs int bins of predicted chi square statistics. FINDOR uses the BaselineLD model from Gazal et al (2016 Nat Genet) for prediction. Download LDscores for the 75 annotation model of BaselineLD model here

Running FINDOR

1. Run LD-score regression with the BaselineLD model on your GWAS data to get annotation effect size estimates. See LD-score regression github

2. Run FINDOR on the entire GWAS to get re-weighted pvalues. Three inputs required:

   A. Full GWAS data set. Requires N, SNP, Z columns.

   B. BaselineLD model LD scores. Can be gzipped.

   C. .results file from an application of LD score regression with the BaselineLD model on GWAS data.

To access details on usage flags: python FINDOR.py --help

An example execution comand would look like:

python FINDOR.py --ref-ld-chr "$PATH_TO_LDSCORES"/baselineLD. --gwas-data "$PATH_TO_GWAS_DATA"/gwas.data --regression-results "$PATH_TO_GWAS_DATA"/gwas.data.results --out "$PATH_TO_GWAS_DATA"/gwas.data.reweighted