Very small quantity of adapter contamination, real or not?

Question

[Ismaelrymy]

Hi there.

I have 3RAD data for studying SNPs. I'm quality-checking the reads, and found a very small quantity of adapter contamination, mainly Illumina universal adapter (FastQc says <0.1% (not even a warning), Process_radtags outputs 0.77% with 2 possible missmatches).

As I understand, I should not have, in theory, adapter contamiantion, as insert size is much longer than read length (around 550 bp vs 150bp). Should I still remove this small quantity of adapter sequences, assuming they are stranded small inserts that somehow passed size selection or should I just consider them false positives? What would be your usual approach to this situation?

Cheers.

[GenoMax]

Use bbduk.sh (GUIDE) to scan and trim. It will remove adapter contamination down to a single base since it has a overlap mode for paired-end reads.

While insert size is larger than read length there is generally a normal distribution in your library so you will have some short inserts in your library (unless something was done to size select).

[Ismaelrymy]

Thanks for the quick answer!

I was going to exclude them directly with process_radtags (stacks), as I cannot trim them, I need same-length reads for posterior analysis, but I think you are right about the small quantity of small inserts, most of the adapter contamiantion occurs in the last parts of the reads.

[GenoMax]

If you are going to use stacks then that should work.