samtools pileup coverage

[guavajuice]

I have a BAM with duplicates marked by picard. When I run pileup with concensus calling on the BAM, pileup reports coverages without the duplicates. I'm pretty certain about that because i did a "samtools view bam chrX:xxx-xxx | wc -l " to count the number of reads in a BAM and it's much more than what pileup reported

I couldn't find any documentation that says samtools pileup excludes duplicated marked reads in its calculation of coverage. Is there anyway to make pileup include duplicates marked reads in the coverage?

Thanks.

[nilshomer]

I have a BAM with duplicates marked by picard. When I run pileup with concensus calling on the BAM, pileup reports coverages without the duplicates. I'm pretty certain about that because i did a "samtools view bam chrX:xxx-xxx | wc -l " to count the number of reads in a BAM and it's much more than what pileup reported

I couldn't find any documentation that says samtools pileup excludes duplicated marked reads in its calculation of coverage. Is there anyway to make pileup include duplicates marked reads in the coverage?

Thanks.

See the "-m" option that remove reads with certain bits set in the flag (like PCR duplicates). Modify however you want.

[guavajuice]

samtools pileup process reads by pushing them in a pileup buffer. one of the conditions it goes through before pushing is to check the the read flag variable "flag_mask" in the pileup buffer structure

The flag_mask is defined when
1) mate is unmapped
2) not primary alignment
3) QC failture
4) optical or PCR duplicate

so it bascially won't include duplicates in a pileup unless you rewrite that condition and compile samtools yourself.

For my case, if i want duplicates read to be reported in samtools, i'll have to run pileup on the BAM before marking it with duplicates.

[bair]

After apply MarkDuplicate on bam file from bwa alignment, samtools flagstat can detect the number of duplicates, how to setup the QC flag for samtools to detect? In my case samtools always detects 0 QC failure.

Thanks

[lh3]

use "-m 772"