removing sequences identified by Chimera Slayer

[katrine]

Hi,

I would like to remove sequences identified as chimeric by Chimera
Slayer.

I ran identify_chimeric_seqs.py with Chimera Slayer on my
seqs_rep_set_aligned fasta file, as recommended in the instructions
for identify_chimeric_seqs.py.

The output from Chimera slayer is a list of numbers from the fasta
identifier in my seqs_rep_set_aligned fasta file. In my case, here is
what an example fasta identifier looks like:

>1001 143_66284 RC:1..236

and here is what part of the chimera slayer output looks like:

1001 14228 114813
1002 14096 36733
1011 31951 190309

1001 here corresponds to the fasta id above. I don't think this number
is unique - how can I use this output to remove individual chimeric
sequences, rather than all the sequences falling into a particular
OTU?

I noticed a way to count OTUs minus the chimeric sequences:

http://qiime.sourceforge.net/scripts/count_otus_minus_chimeras.html

maybe these instructions are incomplete? would this remove all the
sequences in entire OTUs?

Is it possible to set up OTU filtering? Or perhaps I should run
Chimera Slayer on the sequences at an earlier stage, before picking
OTUs? Perhaps I should pick OTUs with 100% identity if the
infrastructure is set up to work on files after OTU picking...

thanks!
Katrine

[Jose Carlos Clemente]

Hi Katrine,

probably Jens can answer the rest of your questions, but
concerning this one:

> Or perhaps I should run Chimera Slayer on the sequences at
> an earlier stage, before picking OTUs?

I am working on testing whether there are significant differences
when checking chimeras before and after picking OTUs. In principle,
we believe the clustering process should help remove some of the
chimeras. Also, checking on the OTUs rather than on the original
sequences should be much faster. So most data right now is being
checked after OTU picking.

What PCR conditions did you use? In general, high number of cycles
(~50) or high concentration of the initial template (~10^7 molecules/microliter)
generate significantly more chimeras, so if your conditions are not like
this I suspect the differences before/after OTU picking are not significant.

Jose

[katrine]

Thank you Jose,

Actually, we did use a large number of PCR cycles (2 rounds), so I am
concerned about chimeras...it might be worth running chimera slayer
before picking OTUs.

either way, I would like to learn how to remove the sequences
identified as chimeric.

thank you!
Katrine

On Dec 6, 4:05 pm, Jose Carlos Clemente <jose.cleme...@gmail.com>
wrote:

[Jens Reeder]

Hi Katrine

There is a script called filter_fast.py that does the job.

Have a look at the chimera checking tutorial to see how we do it usually:
http://qiime.sourceforge.net/tutorials/chimera_checking.html

Jens