janxkoci/popgen.awk

Collection of awk scripts for population & evolutionary genomics. Maybe a library later.

popgen.awk's Issues

new tool: eigselect

Since there is no tool to select samples from a dataset in eigenstrat format, it may be useful to write such tool. Currently, the only option, aside from converting to other format and back, is to use convertf with the poplistname option, which allows selecting populations - modified ind file can be used to select on the individual, rather than population level.

The new tool needs to properly subset samples from both ind and geno files. By default, it could use the ind file for marking which samples to select, or even a custom popfile (similar to derivedsfs tool), while it could optionally select random samples, given a number.

Possible options:

• list of samples or populations from the ind file (-l)
• a custom popfile (-p)
• a random set of samples (-n, or perhaps -r?)

add options for more genotype formats

The current vcfGTcount.gawk script can be expanded to report not just the basic GT summaries, but also e.g. translated genotypes (TGT in the terminology of bcftools) or even IUPAC version (or IUPACGT in bcftools). For example:

• -t = count translated genotypes
• -i = count IUPAC-formated genotypes
• -g = count numeric-style genotypes (default)

This would be handled by a function that gets called after extracting a genotype, using some if checking.

function translate(gt, ref, alt, iupac)
{
    gsub(/0/, ref, gt)
    gsub(/1/, alt, gt)
    if (iupac == 1)
        gt = iupacdict[gt] # needs a dict of iupac codes
    return gt
}

It can be handled by a single function, but maybe more efficient would be to have two functions, so that the if (iupac == 1) is called once rather than on every genotype.

eigenstrat2vcf parse string as prefix

Don't make users type the whole variable assignment syntax. Just ask for a string and put it in the variable yourself. Explain it in usage function (see gawk manual for "awksed" example).

• add usage function
• rewrite the prefix assignment code

tweak derived.gawk to make it portable

The script currently uses only one gawk-specific feature - arrays of arrays. I may be able to rewrite it to use regular "fake" awk multidimensional array instead.

It would allow using faster mawk (even though the script takes only a few minutes to process 4.5M sites for 50 samples) or bioawk for compressed data. Not a high priority though (it's already fast enough and input can be piped from zcat).

Links:

https://stackoverflow.com/questions/3060600/awk-array-iteration-for-multi-dimensional-arrays