I previously asked this question here: Reading vcf file using python gives UnicodeDecodeError
and the answer was to use vcftools instead of bcftools while merging files (the file I was trying to read is a merged file). At the moment this fixed my issue and I moved forward.
I revisited the files to know why one of them worked and the other did not work!
I found that; when using bcftools to merge the three files it gave this warning:
[W::bcf_hdr_merge] Trying to combine "PS" tag definitions of different types .
As a result the file contains strange characters
GT:GQ:DP:GL:PS:ADALL:AD 0|1:415:.:-57.1927,0,-41.4646:<8D>^L:.:. GT:GQ:DP:GL:PS:ADALL:AD 1|0:346:.:-56.3461,0,-34.6188:<8D>^L:.:. ./.:.:.:.:^A:.:.
As you can see <8D>^L and ^A should not be in the output.
When I visited the same locations but when using vcftools for merging :
GT:GQ:DP:GL:PS:ADALL:AD 0|1:.:.:.:-57.192706988686034,0.0,-41.46462464424806:415:68749
Is there a way to change this behavior in bcftools? or it is an issue in implementation?
Thanks
The file resulted from using bcftools:
##FORMAT=<ID=PS,Number=1,Type=Integer,Description="Phase set identifier">.
The file resulted from vcftools :
FORMAT=<ID=PS,Number=1,Type=String,Description="Phase set in which this variant falls">
__Update__
The answer is as suggested by Pierre Lindenbaum is to use the below command before merging:
sed 's/ID=PS,Number=1,Type=Integer,Descri/ID=PS,Number=1,Type=String,Descri/'
so the definitions are not the same. I would used sed to convert the first file:
sed 's/ID=PS,Number=1,Type=Integer,Descri/ID=PS,Number=1,Type=String,Descri/'
What I wrote is the header line in the merged file resulted from both tools, and as you can see the file resulted from bcftools is corrupted. using sed here will not change the corrupted results.
In case if you were asking for the files that were merged:
one of the files contains:
##FORMAT=<ID=PS,Number=1,Type=Integer,Description="Phase set identifier"> .
the other two files does not contains this field.
Log in to answer this question.
what is the definition of BOTH vcf header files for
##FORMAT=<ID=PS.....