search

qbic-pipelines / vcftomaf

MIT License
9 stars 5 forks source link

FreeBayes is currently not supported #2

Open FriederikeHanssen opened 10 months ago

FriederikeHanssen commented 10 months ago

Description of feature

I noticed a couple of things when testing with tumor-only samples:

1) The used --vcf-tumor-id is not correctly resolved:

Sample sheet has: QFEND268A3,QFEND268A3, as sample,tumor_id . For some reason this is what is in the .command.sh:

    --tumor-id QFEND286A1   --vcf-tumor-id  QFEND286A1  \
    --input-vcf QFEND268A3.vcf \
    --output-maf QFEND268A3.maf

but the actual --vcf-tumor-id QFEND286A1 would be NGS02QFEND268A3_NGS02QFEND268A3 , so a) it is not correctly resolved from the given information and b) somehow the last digit is changed. On further inspection it has turned out that in general the ids are completey swapped occasionally. I guess some channel magic doesn't work somewhere, i.e. here:

    --tumor-id QFEND282A3   --vcf-tumor-id  QFEND282A3  \
    --input-vcf QFEND270AE.vcf \
    --output-maf QFEND270AE.maf

2) Freebayes vcfs did not work for me, but resulted in: Command error: [E::vcf_parse_format] Invalid character '.' in 'GQ' FORMAT field at chr1:10146 Error: VCF parse error 3) There is an unused bcftools stats module

FriederikeHanssen commented 6 months ago
  1. and 3. have been addressed by PR #3