Skip to content

Summary Statistics File Format

Brendan Bulik-Sullivan edited this page Mar 5, 2015 · 3 revisions

This page describes all new file formats introduced for use with the --h2 and --rg flags.

NOTE chromosomes are assumed to be integers. We haven't yet implemented LD Score regression for sex chromosomes

.sumstats

For GWAS data. Whitespace-delimited text, one row per SNP with a header row. Column order does not matter.

We recommend that you convert your summary statistics to the .sumstats format using the munge_sumstats.py program included with ldsc, because munge_sumstats.py checks all the gotchas that we've run into over the course of developing this software and applying it to a lot of data.

Required Columns

  1. SNP -- SNP identifier (e.g., rs number)
  2. N -- sample size (which may vary from SNP to SNP).
  3. Z -- z-score. Sign with respect to A1 (warning, possible gotcha)
  4. A1 -- first allele (effect allele)
  5. A2-- second allele (other allele)

Note that ldsc filters out all variants that are not SNPs and strand-ambiguous SNPs.