Issue runing Hatchet2 with hg38 reference_version

[wchukwu]

I get a No snps found in normal! error when I attempt to run Hatchet2 using reference_version=hg38. Reference fasta files are hg38. Interestingly, when I run the relevant bcftools commands from the genotype_snps function locally, I obtain non-empty snp files so it only fails as part of the hatchet run command. Also, if I erroneously set reference_version=hg19, the hatchet run generates non-empty snp files.
I have included my hatchet.ini file here.
hatchet_gitissue.txt

[balabanmetin]

Hi,

did you try chr_notation = False?

If reference chromosomes have the the prefix "chr", set this to True. Otherwise False.

[balabanmetin]

There might be another reason for this issue and I think this is more likely. Make sure your SNP panel has the same chromosome notation (chr prefix exists or not) as the reference. The screenshot shows my setup for grch38. See "chrnotation" in the SNP filename

[wchukwu]

Hi Metin,

Unfortunately, this did not fix the issue. To troubleshoot, I ran multiple instances of hatchet to exhaust the combinations of the reference_version and chr_notation:

Reference version	chr_notation	ini_file	stout_error
""	True	norefT_git.txt	norefT.log
""	False	norefF_git.txt	norefF.log
hg19	True	hg19T_git.txt	hg19T.log
hg19	False	hg19F_git.txt	hg19F.log
hg38	True	hg38T_git.txt	hg38T.log
hg38	False	hg38F_git.txt	hg38F.log

Both the hg38T and hg38F instances fail without calling any snps. Of all, hg19T gets the farthest and fails on phasing but again, it is not the reference version I need.
I also don't provide a snp file myself but rely on this #If unspecified, HATCHet selects a list of known germline SNPs based on <run.reference_version> and <run.chr_notation>
Thanks for your help so far!

[balabanmetin]

Can you share

1. the bam header for the matched normal
2. the bam header for the tumor
3. reference genome fasta record ids (grep ">" from the reference you are using)

[wchukwu]

1. normalbam_header.txt
2. tumorbam_header.txt
3. reffasta_ids.txt