jlc-christie / plink2merlin Goto Github PK

Currently, as explained in the how it works section of the readme, plink is used to prune SNPs in LD and also remove low minor allele frequency variants. However, to read the genotypes for these variants, plink is called again to include only the required SNPs, and then recode used to output in the plaintext (.ped/.map) format, which the genotypes are then read from.

It would likely be a lot faster to instead use --make-bed and read the genotypes directly, since producing and parsing the text file takes up the vast majority of computing time on the script. I chose not to do this, however, as doing this requires testing it well to ensure reading directly from the binary works as expected, and that you're not determining incorrect genotypes.

I may implement this in the future, or in a more generic python module which can perform arbitrary operations on plink binary files, but for now this is quick enough (~10 mins).

PRs referencing this welcome.