FastQC report about WGBS.

Question

[hxlei613]

Hi~ I'm working on some WGBS data now.

After quality and adapter trimming, Sequence Duplication Levels and Per sequence GC content still cannot pass . In Per sequence GC content, the read peak is higher than theoretical distibution. Is this ok ?

Thank you very much if you can provide some help !

Here are some pictures from FastQC after trimming.

There is a small fluctuation at the first few bases.Should I trim it ? At the end, the sharp decrease of A at the last position is a result of removing the adapter sequence very stringently, i.e. even a single trailing A at the end is removed.

Should I deduplicate sequence during quality control ( before mapping ) or filtering reads after alignments using deduplicate_bismark ?

Answer 1

[igor]

Don't worry too much about FastQC reports. They are very conservative. It's nearly impossible to have them all pass. You should manually look at the report and see if each section makes sense.

See previous discussion here about FastQC: FastQ quality check : what can we correct ?

And you can use deduplicate_bismark to remove duplicates, which is convenient if you are using Bismark for everything else.

[hxlei613]

Thank you very much ! Do you mean duplicates can be kept before mapping ?

[Devon Ryan]

Correct, there's no reason to bother deduplicating before alignment.

[igor]

Yes. Usually you identify duplicates if two reads (or read pairs) align to the same exact spot in the genome.

[hxlei613]

For pair-end alignments, does bismark consider a duplicate if both partner reads start and end at the exact same position ? Or if only one of the partner reads ?

[hxlei613]

Oh, I figure it out. A duplicate is which both partner reads start and end at the exact same position. Thank you very much.