ReadBackedPhasing didn't merge consecutive phased sites into MNP records ?
Dear GATK team,
I have been using GATK for years on variant calling (thank you first!), but this is the first time I tried ReadBackedPhasing to emit MNPs, and it didn't output any MNPs. Could you please help me to find out the reason?
I was calling variants on whole exome data, and, following the GATK's best practices, basically the procedure was:
HaplotypeCaller with --emitRefConfidence GVCF -->
ReadBackedPhasing with -enableMergeToMNP
Every step went successfully without errors.
In the output VCF file of ReadBackedPhasing, tag
HP, which was not in the output of previous steps, was added to the genotype section, but no consecutive phased sites were merged into MNP records. I have checked those variants, and many of them should have been merged into MNP records. For example, the alignment of one of them is shown in the attached figure.
The version of GATK that I tried are nightly-2015-08-16-g3087cd7 and nightly-2015-10-29-gc252559. The complete command I used for ReadBackedPhasing is shown below, and I also used options
--phaseQualityThresh but still no MNPs were emitted.
java -Xmx9g -jar GenomeAnalysisTK.jar \ -T ReadBackedPhasing \ -R human_g1k_v37_decoy.fasta \ -enableMergeToMNP \ --variant VQSR_out.vcf.gz \ -L VQSR_out.vcf.gz \ -o out.vcf \ -I 2.bam \ -I 3.bam \ -I 4.bam \ -I 5.bam \ -et NO_ET -K key_file