No optitype output from tumor/normal calling

[aadamk]

I ran tumor variant calling using an ensemble caller with vep annotation + hla calling for normal. I successfully obtained an annotated ensemble vcf, but the expected .csv containing optitype output was not generated. My aim was to get the hla genotype for normal tissue. No issues are reported in the log files, but there is no record of optitype having run (though there are references to hla typing in the log file - e.g. 'Timing: HLA typing' & 'Multiprocessing: call_hla'). I have a snapshot of the sample.yaml file I created. Is there an error in my preparation?

details:

• algorithm: aligner: false align_split_size: 5000000 nomap_split_targets: 100 mark_duplicates: false recalibrate: false realign: false remove_lcr: false platform: illumina quality_format: standard variantcaller: [vardict, mutect2] indelcaller: mutect2 hlacaller: optitype bam_clean: fixrg ensemble: numpass: 2 analysis: variant2 description: 98-PBL_control files: /project/hanslab/lyn_Mel_Case/SN0119031/98-PBL_control.bam genome_build: GRCh37 metadata: batch: mel_case22 phenotype: normal
• algorithm: aligner: false align_split_size: 5000000 nomap_split_targets: 100 mark_duplicates: false recalibrate: false realign: false remove_lcr: false platform: illumina bam_clean: fixrg quality_format: standard variantcaller: [vardict, mutect2] indelcaller: mutect2 ensemble: numpass: 2 effects: vep tools_off: [gemini] analysis: variant2 description: D14-617_C4 files: /project/hanslab/lyn_Mel_Case/SN0119031/D14-617_C4.bam genome_build: GRCh37 metadata: batch: mel_case22 phenotype: tumor fc_date: '7.5.2017' fc_name: mel_case22 upload: dir: ../final

[schelhorn]

I believe you would have to align to hg38 - at least that is what people are mostly using to include the ALTs.

[chapmanb]

Sven-Eric is exactly right, unfortunately HLA typing with OptiType only works with hg38. We use hg38 HLA alt alleles to extract HLA reads prior to feeding into OptiType. Apologies about the confusion, I added a check so bcbio will provide a useful error message instead of silently failing to call HLAs. Hopefully swapping to hg38 works for your analysis.

[aadamk]

Thanks all.
In my case, I do not have the original fastq's, only the bam file which was aligned to hg19. Any workaround for this? Does bcbio support bamtofastq conversion so that I can realign?

[chapmanb]

Thanks for following up. bcbio will convert BAMs to fastq and realign automatically for you, so a configuration like this (based on what you posted previously) will align to hg38 and HLA call only:

details:
- algorithm:
    aligner: bwa
    hlacaller: optitype
  analysis: variant2
  description: 98-PBL_control
  files: /project/hanslab/lyn_Mel_Case/SN0119031/98-PBL_control.bam
  genome_build: hg38
fc_date: '7.6.2017'
fc_name: mel_normal_hla
upload:
  dir: ../final

Hope this gets it running for you.

[aadamk]

Thanks, this worked!