reason for untrimmed PE reads for assembly

[Xiaodong]

Dear Jon and others,

I have a question about why dDOcent requires paried-end (PE) reads for de novo assembly to be untrimmed ( demultiplex using process_radtags without -c -q, no adapter removal). 

One potential problem that I have in mind is that the reads for assembly may be filtered for bad bases in the end and thus result in a contig shorter than the length of PE pair. What other potential problems might it cause?

Thank you for explaining this.

Best wishes,

Ivan