Overrepresented sequences from FastQC report

[morning latte]

I have a question regarding overrepresented sequences from FastQC report. I really appreciate for your replies in advance. Please see attached.

First, I am going to remove Illumina Single End PCR Primer 1 using cutadapt. After this, there are three TruSeq Adapters left. In fact, they share same sequences but just different length (69, 67, 65bp). In this case, can I just put the longest one among three as input then the rest two shorter ones are removed at the same time? Or should I do it the opposite way?

[westerman]

I suggest longest to shortest. That way if the read matches the longer adapter then the adapter will be removed completely. If the read matches a short adapter then it would be caught later in the process.

[morning latte]

Just want to make sure, do you mean I can just put the longest sequence as input, right? Thanks for your reply.

[westerman]

I would put in all three sequences starting from the longest to the shortest. Note that I don't use cutadapt very much and instead use trimmomatic (mainly because it worked properly with paired-end sequences which is the bulk of my work) however the times I do use cutadapt I tend to throw all possible vectors at it just to make sure I get clean results.

[morning latte]

I get it. Thanks a LOT.

[simonandrews]

A more flexible approach is to not use a specific adapter sequence at all. Pretty much all Illumina adapters start with a common seed sequence (I think some of the small RNA ones do something different and strange, but I may be mis-remembering), and then diverge into the more specific barcodes and application specific variants.

In trim galore (which is our wrapper around cutadapt which makes it a bit easier to work with, and adds some extra options), we default to using this seed sequence (AGATCGGAAGAGC) to trim with. This way it doesn't matter which variant you're using as they'll all get trimmed and you don't need to worry which barcodes you used in a particular run.

[westerman]

Simon has a good point. You can just use the seed sequence since any read that contained an entire adapter beyond the seed sequence would be a very poor read indeed. I was focusing too much on your initial comment of having 3 adapters to get rid of instead of focusing on the best solution.

[morning latte]

Thanks all. This really helped.