# How likely is it to find primers sequences in already-trimmed reads?

So, I am analysing 18S amplicon data (fastq files) to be able to eventually investigate the taxa composition.

After removing primers (using cutadapt) from both R1 and R2 (assuming primers are found at the beginning of my sequences), I am still able to detect, apparently at random, a few reads that contain my primers (i.e. a part of the sequence matching the primer) in, again apparently random, positions.

I am able to detect almost all my primers (16 out of 24, both forward and reverse) in at least one sequence and I am wondering how common (or how weird) this is. I guess it is important to consider that my primers contains a few, i.e. from 1 to 3, wild-card bases that may match multiple other bases (apparently called whobble bases), which makes it easier to match a random sequence (there are multiple possibilities for an actual match).

Problem is that the presence of primers in the sequences may affect downstream analyses (e.g. with DADA2 they are very clear about the fact that primers have to be removed) in several ways and I was wondering if these sequences I am finding may be interfering somehow of if I can live with it.

• The primers will be at the 3'-end of the reads. Make sure you check for them there. Aug 23 at 9:43
• @DevonRyan yes, that I know and that I did so no primers at 3'. I guess it just happens to find them inside the reads, as random matches, but I wasn't expecting that many actual matches. I guess it's statistics, then.
– gabt
Aug 23 at 10:54
• Interesting, have you tried to map a read with a sequence matching a primer to the reference, or even simply blasting it? Does the reference contain the "primer sequence"? ("" as we don't know if it's a homologous sequence or not) Aug 30 at 14:25
• @KamilSJaron apparently this is something that should not be happening and I'm trying to understand why it does.
– gabt
Sep 1 at 8:03