Someone asked us: can the output of mergeSTR be used as input to mergeSTR? In today's episode, we set out to answer that question by writing a test!

For the test, I subsetted one of our test files, CEU_test.vcf.gz, into three smaller files, each with non-overlapping sample subsets: 10 samples, 19 samples, and 150 samples. The original file had 10+19+150 = 179 samples. Then, I wrote a test that merged the 10 sample and 19 sample files and then merged the output of that with the 150 sample file. Finally, I check that the final output is equivalent to the original file.

Unfortunately, the test failed. I haven't dug into why yet. The test is reporting that the headers aren't identical, but I think there are deeper issues too? For example, the alleles in the GT column aren't the same, either. I checked with

diff -sq \
<(bcftools query -f '%CHROM %POS[\t%GT]\n' test_output.vcf) \
<(bcftools query -f '%CHROM %POS[\t%GT]\n' original.vcf)

Anyway, I'm opening this PR in case we want to fix this. Or we can just leave it open indefinitely as a way of reporting the issue to everyone -- at least until someone has a chance to fix it.