Calculate Depth with and without Soft-Clipping

Hi guys,

I'm trying to use GATK3 (since v4 doesn't have the DepthOfCoverage tool yet) to calculate the depth of coverage across some WGS BAM files, with and with out soft clipping.

Currently, I'm using the following the following

#calculate depth, exclude soft-clipped bases
java -Xmx${mem}g -jar $jar -mbq 0 -mmq 0 -L ${bed} -o $out/${bamName}.doc -I $bam  -T DepthOfCoverage -R $ref

#calculate depth, include soft-clipped bases
java  -Xmx${mem}g -jar $jar -T ClipReads -l INFO -I $bam -o $bam.noclip.bam -R $ref -CR REVERT_SOFTCLIPPED_BASES -os clip.stats 2>> err.sc >> out.sc
java  -Xmx${mem}g -jar $jar -mbq 0 -mmq 0 -L $bed -o $out/${bamName}.noclip.bam.doc -I $bam.noclip.bam  -T DepthOfCoverage -R $ref

Basically, what I'm asking GATK to do is: 1) traverse the original BAM file and calculate the depth; 2) convert this original BAM file into a new BAM file with no soft clipping; 3) traverse this new BAM file and calculate the depth.

This is a computationally intensive process; it takes both large amounts of time and memory. And I'm wondering if there's a more efficient way to do it? If there's another tool, like samtools depth, that could accomplish this more efficiently I'm happy to switch to that or anything else.

Thanks!

bam gatk sequencing alignment

• 1,975 views • link

updated 8.2 years ago by Pierre Lindenbaum

• written 8.2 years ago by ccnn •

1 answer

I would say that whatever you're using, it will give no difference because most (all ?) tools don't use the clipped bases.

• 1 views • link

written 8.2 years ago by Pierre Lindenbaum

I understand that I may have to use GATK -T ClipReads -CR REVERT_SOFTCLIPPED_BASES to create a BAM with no soft-clipping from my original BAM, but won't some ways of calculating depth across this original and modified BAM be more efficient than others?

Alternatively, is there no tool that can, processing a BAM just once, tell me how what the depth w and w/o soft-clipping is at each spot? Or just the depth and the amount of soft-clipping and I'll add these numbers together? Thanks for your help.

• 1 views • link

written 8.2 years ago by ccnn •

but won't some ways of calculating depth across this original and modified BAM be more efficient than others?

it sounds like microoptimization and you'll consume some time and storage to run GATK

Alternatively, is there no tool that can, processing a BAM just once, tell me how what the depth w and w/o soft-clipping is at each spot?

what would be the logic in this ? if a part of the read is mapped on another region of the genome, why would you need to consider this ?

• 1 views • link

written 8.2 years ago by Pierre Lindenbaum