Problems with finding out indels using Unified Genotyper ? using GATK

[swapnil2188]

Problems with finding out indels using Unified Genotyper ? (GATK)

I followed the std protocol of First base reaclibarting the .bam file followed by using the realigned .bam file to call for SNPs and Indels.
It calculates snps but shows an error for Indels -

The command I used is
" java -jar GenomeAnalysisTK.jar -I realigned_sample.hg19.sorted.bam -R hg19.fa -T UnifiedGenotyper -glm INDEL -o indelcalls_110527_s_3_1_fq.hg19.vcf -stand_call_conf 50 -S SILENT ".......

I get an error saying - "SAM/BAM file
SAMFileReader{realigned_sample.fq.hg19.bam} is malformed: Adjacent I/D events in read HWUSI-EAS68R:59:1:16:12673:10747 -- cigar: 4M4I3D34M2043N9M" .....

I looked through different forums for this error and its a common problem that many people face. The solutions are unclear and confusing. Can anyone please suggest me clear solution for this ?

Thanks !

[Rocketknight]

This is a known issue with the GATK dealing with reads from Stampy. I talked with the author of Stampy about this and he suggested that the fault was with GATK, as adjacent I/D events are legal in SAM files. Nonetheless, he added a command line option, --gatkcigarworkaround which will stop Stampy outputting adjacent I/D events in CIGAR strings. It's available in the development version of Stampy, http://www.well.ox.ac.uk/~gerton/software/Stampy/dev/ .

Alternatively, if you run GATK IndelRealigner with --USE_SW instead of USE_READS or KNOWNS_ONLY when you're doing indel realignment it will generate new alignments and CIGAR strings, which may also resolve the issue.

I have no idea how exactly Stampy modifies CIGAR strings to remove adjacent I/D events (presumably it just inserts one mismatch in the middle) or how GATK's Indelrealigner deals with them, so I don't know what the "best" approach here is. Probably the safest thing to do is to use the --gatkcigarworkaround and then also run indelrealigner with --USE_SW.