I am working with publicly available data sets of VCF files. I accessed VCF files broken out by patient and by chromosome with just the 0/0 calls, and unfortunately the ALT column includes a value <non_ref> on every line. I also have VCF files per patient with 1/1 and 0/1 calls across the entire genome, those do have values in the ALT column, such as A, G, or CATGTT, for instance.
I merged all files by patient, but then when I try to use bcftools merge across patients, the problem is the single merged vcf file (with 5000 patients) now thinks of <non_ref> literally as one of the potential ALT calls.
Sadly, I cannot go back upstream in this public data set and re-run these files with GATK.
I am wondering if anyone has any ideas on how to get vcftools, bcftools or gatk vcf merge functions to ignore the <non_ref> value in the ALT column on some lines in each file?
P.S. I tried a recode the files manually with perl -pe "s/<non_ref>/./g" but bcftools is throwing flags, as in missing value in ALT column.
Jim
I will answer my own question:
I started with single sample VCF files and some had lines with 0/0 calls with alt designated by NON_REF
"bcftools merge" created problems for me because it treated NON_REF as a literal ALT allele, which sometimes made other 0/1 calls turn into 0/2 calls.
However, when I reverted to the older software vcftools and ran: "vcf-merge" it always put NON_REF at the end of the list of potential ALT alleles, and did not cause and problems.
So, bcftools thinks of NON_REF as a literal ALT allele call, but vcftools does not seem to think that NON_REF is a literal allele
--and it works.
Just to clarify, the lines in the file that include a 1/1 or 0/1 call, do include an ALT value as in G,C, or GATC the lines in the VCF file that are 0/0 calls only have ALT values that read <non_ref>
You should edit your original post to clarify this, afterwards you can delete this answer. Adding an answer should only be used to provide a solution to the issue, not for clarifications.
Log in to answer this question.
There is no need to SHOUT. I have adapted your title and simultaneously made it more specific.