inconsistency between the read depth reported by HC and by samtools mpileup
Posted in Ask the GATK team | Last updated on


Comments (2)

Hi everyone,

I'm calling SNPs by using GATK HC. I'm also using samtools mpileup to give specific base type and read depth information for each called SNP position (using the same bam file, of course). However, I found that in about 10% of cases, in the positions where GATK called SNPs, samtools mpileup did not report any read coverages. That confused me, so I put the bam file into igv and tablet to visualize the reads. However, both of them confirmed that there were indeed no reads in those positions at all.

Here is the command I used for GATK:

java -jar GenomeAnalysisTK.jar -T HaplotypeCaller -R index/genome_mu50.fasta -I wt_rg.bam -o wt_raw.vcf

for samtools mpileup:

samtools mpileup -f index/genome_mu50.fasta wt_rg.bam > wt.coverage

Please see the vcf table: Please notice that a couple of SNPs were called between 1550190 and 1550201, and reasonable allele depths were reported to back up these results.

Please see the igv snip shot of this area: There is not a single read in this area!

I also tried UnifiedGenotyper, and those SNPs were not called. Is this a bug in HC? or I did something wrong? Any comments will be appreciated. Thanks!


Return to top Comment on this article in the forum