I have a clone library of appr. 500 partial 16S rRNA gene sequences (800-900bases). I have been analysing it using mothur v. 1.12.3.
I have made trim.seqs and then align.seqs using silva.bacteria.fasta as a template. Thereafter I filtered the sequences along with silva.gold.align and run chimera.slayer using the default parameters. I am pretty surprised with the result. Over 50% of my sequences were detected as chimeras.
I have proceeded similarly with another 16S clone library analysis and with one 16S 454-pyrosequecing analysis and in those cases the number of chimeras detected were “tolerable” for me.
Is there some things to consider when using chimera.slayer? Adjustment of parameters or something? Chimera.bellerophon detected even more chimeras than slayer, whereas chimera.pintail (used with silva.gold.align as reference) detected fewer.
Or do I just happen to have a chimera-rich library in my hand?
Unfortunately, the paper describing ChimeraSlayer is not yet published; however, I do understand that it has been accepted. Obviously, I can’t speak to your specific dataset, but the overall relationship between the three methods that you are observing sounds about correct. As always, I’d encourage you to take the putative chimeras and determine whether they actually look chimeric manually.
I got back to the 16S clone library data I had analysed with chimera.slayer last autumn. Now I used the newest version of Mothur 1.19.
So, with the version 1.12.3 chimera.slayer I detected over 50% of my seqs as chimeras. Using version 1.19 with exactly the same fasta and template file as previously I got zero chimeric seqs.
I did hear from my collague that getting false positives is now more unlikely than before but can it really be possible that appr. 260 seqs previously detected as chimeras really are not chimeric?
I’m afraid that it is true that the version you had been using would have a high false positive rate (http://www.mothur.org/wiki/Mothur_v.1.19.0). We are very sorry for this confusion. Also, if you didn’t receive this notice from us, that means you are not on the mailing list. I would encourage you and everyone else using mothur to please subscribe to the list at http://www.mothur.org/mailman/listinfo/mothur-announce.