Removing Chimeric sequences from Metagenomics samples

[RickBioinf]

I'm making 2 Metagenomics pipelines (one for WGS and one for 16S) and I just found out about Chimeras and I think I understand what they are.

Chimeric sequences are products when 2 halves of a PCR product "link" together to form one sequence. This is a problem because the sequence can then be assigned as a different (wrong) sequence.

But I don't (at all) see this as a problem for 16S, because this sequence is so short you'll allow a very small mismatch ratio (0-1) so you won't see anything.
The people I work with have never looked for Chimeric sequences and say that it isn't really a problem here even for WGS. Are they wrong? Why do I want to really search for chimeric sequences? What kind of tool do you advise to use? And am I wrong about the 16S approach?

Thanks for reading and if you reply, thank you even more.

[cliffbeall]

The chimera frequency in WGS will be very low and shouldn't be a problem. The frequency in 16S experiments is much higher because there are short conserved sequences that can internally prime during the PCR.

They are a problem in 16S because they chimeras can get treated as novel OTUs and inflate diversity estimates. I'm not sure what your point about mismatch ratio is.

[RickBioinf]

Hi cliffbeal,

Thank you for replying.
I see that I haven't been complete. I'm talking about the V4 region in the 16S.
With mismatch ratio (maybe I'm using the wrong term here) I mean that when you align your sequence you allow a certain amount of mismatches to the sequence, because of possible errors in your PCR.
But the problem with this is that if you have just a small amount of "mismatches" or mutations in your sequence it could be considered as the wrong organism.
So in that order you would allow a small amount of errors with the alignment and chimeras shouldn't be a problem then.

Regards,

Rick

[cliffbeall]

You often see chimeras between parent sequences with <90% identity. That can generate things that are not 97% identical to either parent if you're using that threshold so they get classified as a novel OTU.

This reference is pretty good on the formation of chimeras, though I think most people use uchime rather than chimera slayer these days.