Detecting Human Contamination in Low Coverage NGS sample

Question

[igorm]

Hi,

I have created contaminated human sequence R1, R2 fq.gz files by adding 10% random sequences from another human sample. When I align to the reference genome, the average coverage is around 1.0x.

I have tried to detect the contamination with verifybamid and also GATK GetPileupSummaries/CalcualteContamination but the the contamination is not detected.

Has anyone been successful detecting contamination on low coverage .bam files with coverage around 1.0x?

Thanks.

Answer 1

[igorm]

SOLVED: The .bam had to be duplication-marked. When using the .bam produced by Picard's MarkDuplicates with verifybamid, I get the correct contamination estimate.