TruSeq Adaptors reported by FastQC are true adaptors?

[Jiafen]

I asked a question about using fastx_clipper to get rid of the adaptor sequence last time and it did not work.

I googled TruSeq adaptor sequences (http://www.omicsoft.com/downloads/ng...on_list/v1.txt) and compared them with our reported TruSeq Adaptors. Because our sequences have length between 49-52, all reported adaptor have length 50. I choose the first 50 nt from TruSeq adaptor to compare with ours, the results are as follows:
4 reported TrueSeq Adaptors are exactly the same as the first 50nt of TruSeq adaptor;
3 reported TrueSeq Adaptors have almost the same sequence as the first 50 nt except the 42nd nt, they all replace A by C;
2 reported TrueSeq Adaptors, their 2nd-50 nt are the same as TruSeq Adaptor 1-49nt

I decided to try to align my reads without get rid of adaptor sequence by novoalign (using default setting) to see whether those sequences reported as TruSeq Adaptors are in the result of alignment. Unfortunately, they are. But if I specify the reported adaptor sequence as adaptor in novoalign, these sequence will be removed.

So my questions now is whether those sequences reported as TruSeq Adaptor by FastQC are true Adaptor sequences or not?

[pbseq]

If Ii remind correctly fastQC allows for some mismatches , you may also wish to compare identified adapters to fastQC database contaminants within fastQC folder
Pbseq

[simonandrews]

FastQC allows some flexibility in its matches, it also doesn't require a match to exist over the whole length of the sequence. The summary of the match will tell you how good a match it actually found.

Many of the illumina adapters are very similar to each other, differing by only a few bases so FastQC often finds a multitude of possible hits, so it just picks the first of the best set of hits to report.

Given that the program only does these searches for sequences which occur at very high levels in a library it's pretty unusual to get a complete false positive for the presence of an adapter sequence, although the identification of the exact adapter used may well not be correct.

[Jiafen]

Thank you very much, Simon. That makes more sense to me.

I want to correct the observation I mentioned earlier. I double checked manuscript of novoalign, the reported adaptor sequences are not aligned to any region, I am assuming they are true adaptors.

FastQC allows some flexibility in its matches, it also doesn't require a match to exist over the whole length of the sequence. The summary of the match will tell you how good a match it actually found.

Many of the illumina adapters are very similar to each other, differing by only a few bases so FastQC often finds a multitude of possible hits, so it just picks the first of the best set of hits to report.

Given that the program only does these searches for sequences which occur at very high levels in a library it's pretty unusual to get a complete false positive for the presence of an adapter sequence, although the identification of the exact adapter used may well not be correct.