GATK IndelRealigner error, badly formed genome loc: unknown contig chrM.

[gumbos]

The original fasta file used to generate the alignments has the correct label of >ChrM. When I check the header of the input BAM I find the chrM @SQ line.

What could cause this?

[shyam_la]

The original fasta file used to generate the alignments has the correct label of >ChrM. When I check the header of the input BAM I find the chrM @SQ line.

What could cause this?

You really need to provide more info..

[gumbos]

What more info can I provide? I am trying to run the GATK pipeline following the best practices page. I have generated a SAM using the newest bowtie2 beta. I convert that SAM to a BAM, sort it, index it, then run RealignerTargetCreator, which finishes successfully. I use the intervals file and the same BAM to run IndelRealigner, which runs fine until the very end when it hits chrM, and reports that error.

INFO 15:50:36,054 TraversalEngine - chr9:55145149 4.13e+07 69.1 m 100.3 s 99.8% 69.2 m 9.1 s
......
##### ERROR MESSAGE: Badly formed genome loc: Parameters to GenomeLocParser are incorrect:Unknown contig chrM

So I went back and made sure that the fasta used for the alignment had the correct label, which it does. I checked the SAM file and the @SQ header has chrM, and I checked for some alignments and found chrM in the 3rd position. I checked for the presence of chrMt, as ensembl sometimes does it, anywhere in the SAM and did not find it.

[shyam_la]

Are you using the same fasta that you used for alignment for the indel realigner? I know it kind of sounds stupid to ask, but everything else seems fine..
By the way, ensembl "always" does MT. You must be using UCSC references..

[gumbos]

I am fairly sure I am using the correct reference fasta because it was all done via a shell script. As a test, I went back and took chrM out of the fasta, and it worked. However, I am now stuck on the CountCovariates step. Before I make another post, maybe you know:

My reference VCF comes from converting the ensembl GVF to VCF with gvf2vcf. I then get the error "The provided VCF file is malformed at approximately line number 4: Empty alleles are not permitted in VCF records". I noticed referring to the table at http://vcftools.sourceforge.net/VCF-poster.pdf that my VCF does not contain the format, sample1 or sample2 columns. Are these required? If so, do you know of where I can get a proper Zebrafish dbSNP VCF? UCSC doesn't have one as far as I can tell, and the dbSNP ftp files do not come in VCF format as far as I can tell.

[alyza]

I got the same message when I ran the IndelRealigner but my reference file does not include chrM. I then ran the UnifiedGenotyper anyways. Is my vcf file OK even though it gave an error that my file is truncated or corrupt?