Handling duplicates and other biases in Haloplex enriched targeted sequencing data

[ForeignMan]

Hello everyone,

we're starting to work with deep sequencing data of target regions that were enriched by Agilent's Haloplex kit (Illumina MiSeq, paired-end, often >500x mean seq. depth).
How would you handle duplicate reads after alignment? The initial DNA fragmentation by restriction enzymes seems to make it impossible to remove duplicates (from PCR mainly) just like for random fragmentation. If you kept all the duplicates, does the amplification bias or other biases like strand bias hamper mutation detection significantly?

Would be great if someone had some experience he'd like to share and discuss here or knew some literature that presents some work with Haloplex sequencing data (which seems to be quite rare).

Thanks in advance,
Chris

[fzhang03]

In our Haloplex library sequencing results, we saw 200x coverage for all probed regions, but, after removing duplicated reads, it is 6x coverage. We don't know how to handle this issue yet.

[yl01]

The advice we got from Agilent support people is that for Haloplex data it is important not to remove PCR duplicates. Haloplex data sets are expected to produce a lot of reads with exact same start / end genomic coordinates (corresponding to the restriction enzyme sites), which will be seen as duplicates, but should not be removed from the analysis.

[fzhang03]

Thanks! that's helpful.