add dust-masked output

[gpertea]

Todd Wylie requested this feature:

I’m writing to ask if it is possible to apply the “dust” style low-complexity filter, but write out (masked) sequence instead of failing the masked reads?

I think for our purposes, ideally, the -D switch would still function the same—reads would be filtered out if >=50% low complexity—but another switch could also be used in conjunction to allow any low complexity base pairs to be written as N’s in the FASTQ file for the reads that were <50% low complexity. We frequently use BWA mem, and it treats soft-masked reads the same; however, N’s count as mismatches in the alignment. We often look for viral integration breakpoint reads (which look like chimeric sequences), so retaining some masked sequence and dealing with mismatches downstream can be useful.