Why do my coverage maps look like complexity traces

[The_Roads]

Hi,

I am working with illumina 2x 50bp reads from pipeline 1.4. when i assemble the reads using either CLCGWB or Ngen i get coverage maps that look almost identical to complexity traces of my ref seqs. Does anyone have any idea what might be causing this?

The_Roads

[jkbonfield]

I know nothing of how CLCGWB or Ngen work as aligners. How do they handle repeats? Low complexity sequence is often repetitive, so if you use an aligner that doesn't support random-placement of repeated sequences you'll get a lack of data in some regions. I'd expect it to be a bit more binary though.

It's possible there's also some library preparation issue, but that's not really my area.

[The_Roads]

Not sure about Ngen but CLC has an option to place them randomly whcih is how all these runs are assembled. we did some more work and it looks like ELAND is giving the same results. we do have very high coverage, peaking at 30k in our full assemblies but if we take only 1/10 of the reads we still get the same complexity-like coverage.

[The_Roads]

btw when i say complexity trace i mean the CLCGWB local complexity plot, that is apparently based on SEG from Wooton and Ferderhen '93/'96 if that is any help.