Introduction

LDBIRD is a command-line program for computing LD metrics between all pairs of variants in its input files.

In working on the paper described below, I needed a way to quickly assess the distribution of between-chromosome LD in P.falciparum populations - as well as finding individual pairs of variants with outlying LD. Thus LDBIRD was born. The key features LDBIRD has are:

• It computes a histogram of LD values across pairs of variants as well as identifying individual pairs with strong LD.
• It handles haploid genotypes (as well as diploid genotypes, and haploid genotypes encoded as if they were diploid).
• It accepts a variety of different genotype file formats (a subset of those accepted by QCTOOL) and has helpful options for filtering samples and variants.
• Its output format (in a sqlite db) is reasonably compact and readibly post-processable.
• It's reasonably fast (given what it does).

For an introduction to running LDBIRD, please see the getting started page.

How fast is LDBIRD?

LDBIRD is likely not the fastest way you can compute LD. It limits itself to loading variants one (or rather two) at a time, and thus its algorithm is quadratic in the number of variants. If LDBIRD doesn't meet your needs, other ways to compute LD are to use plink or LDStore - or possibly tskit.

Obtaining LDBIRD

LDBIRD is currently included as part of the QCTOOL package, which is currently hosted at code.enkre.net. See the download page for full details on obtaining and installing LDBIRD.

References

LDBIRD was originally developed for a study of human and malaria parasite variation in samples collected for MalariaGEN Consortial Project 1. If you use LDBIRD in published work, please cite the following preprint (or its published equivalent):

Contact

For more information or questions, please contact me at gavin [dot] band (at) well.ox.ac.uk.