This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!
Why is HaplotypeCaller dropping half of my reads?
I have been trying HaplotypeCaller to find SNPs and INDELS in viral read data (haploid) but am finding that it throws away around half of my reads and I don't understand why. A small proportion (8%) are filtered out duplicates and 0.05% fail on mapping quality but I can't account for the majority of lost reads. I appreciate that GATK wasn't built for viral sequences but would you have an idea of what could be causing this?
I use the following command after marking duplicates and realigning around indels:
java -jar GenomeAnalysisTK.jar -T HaplotypeCaller -R Ref.fasta -I realigned_reads.bam --genotyping_mode DISCOVERY -ploidy 1 -bamout reassembled.bam -o rawvariants.vcf
I have also tried the same file with UnifiedGenotype and I get the result I expect i.e. most of my reads are retained and I have SNP calls that agree with a VCF constructed in a different program so I assume the reads are lost as part of the local realignment?