Marking Dups On Pcr Free Genomes - Is It Necessary?

Question

[richardc.gsc]

Hi all, In our PCR free human genome libraries we routinely generate 100X coverage with about 3% duplicates (marked by Picard). Since dup marking is relatively slow and less paralellized in our pipeline than other processes we are wondering if any had any thoughts on how terrible it might be if we simply skipped the dup marking for the PCR free genomes.

Any thoughts?

Answer 1

[Gabriel R.]

My 2 cents:

You should not do it because you end removing sequences that are not duplicates but really two "independent" observations. However, my personal experience also tells me that you have duplicated clusters that are labeled as a single one. Perhaps you want to measure how endemic that effect is prior to doing something about it.