Problem with Mitochondrial short variant discovery; RevertSam and MergeBamAlignment steps
Hi,
I am experiencing problems with the GATK mtDNA pipeline for short variant discovery, specifically in the RevertSam/MergeBamAlignment steps.
My input file for the pipeline is a Hg38-aligned BAM file from Illumina NovaSeq whole genome sequencing.
I then proceed with filtering the chrM reads from this Hg38-aligned BAM. In the RevertSam step, all alignment information is removed from the file. But when I try to proceed with MergeBamAlignment I get an error that my headers do not match. It indicats my chrM.bam has more header information than my reference file. This seems correct as in the filter chrM reads step I only filter the reads and not the additional header information. Whenever I try to remove the header information of all other chromosomes (except chrM) I end up with corrupted files or files that give me ‘Invalid alignment start’ errors in the RevertSam step.
Has anyone ever faced a similar (header) issue that caused you to be unable to RevertSam your file and proceed with the MergeBamAlignment? Or is there an example of what the correct headers are supposed to look like?
Kind regards,
Floor
-
RevertSam step is used to convert chrM reads to their unmapped state in order to be converted back to fastq and feed to BWA to realign to the mitochondrial genome alone. You cannot use your chrM aligned reads filtered from the original input bam directly for this analysis.
Can you try mapping your reads to mitochondrial genome and merge later?
Regards.
-
Dear Gökalp Çelik,
Thank you for your message. If I understand you correctly that is an additional step you would add between the RevertSam and MergeBamAlignment step, which is currently not indicated on the webpage. I was specifically interested in performing the pipeline as described on the website to compare it to my self-made pipeline.
Kind regards,
Floor
-
Hi again.
You may wish to refer back to the wdl that we publish within our GATK releases. There is a bwa mem mapping stage in between those two steps actually twice (there is only a single implementation called twice by the master wdl.). Once for the regular chrM and the other is for the rotated chrM reference.
https://github.com/broadinstitute/gatk/blob/master/scripts/mitochondria_m2_wdl/AlignmentPipeline.wdl
I hope this helps.
Regards.
Please sign in to leave a comment.
3 comments