Issues with ChimeraSlayer

19 views
Skip to first unread message

Kate Blackwell

unread,
Apr 14, 2016, 1:59:26 AM4/14/16
to Qiime 1 Forum
I ran a single sample through my lab's bioinformatics pipeline, once with ChimeraSlayer and once without, however, I obtained some strange results.  The # of sequences per sample for the pipeline without ChimeraSlayer was 44183 and with ChimeraSlayer was 48591.  

I would think that checking for and removing chimeras would reduce the number of sequences, not increase it...

And advice as to explain this observation?  I gone back over the commands and files used and I cannot find any mistakes.  I've attached the commands I used to generate my results.
Bioinformatics pipeline for demultiplexed MiSeq Illumina sequences.pdf

jonsan

unread,
Apr 14, 2016, 2:22:23 PM4/14/16
to Qiime 1 Forum
Hi Kate,

Hard to say exactly what's going on with your sample, but looking at your pipeline my guess is that this discrepancy is coming from making a new OTU table rather than just filtering down the one already generated by the pick_open_reference_otus.py workflow. Typically, if I do chimera detection and removal on an open-reference pipeline starting with the pick_open_reference_otus.py workflow, I will use the rep_set.fna as my input to a reference-based chimera detection program like ChimeraSlayer, and then use filter_otus_from_otu_table.py to filter out the sequences identified as chimeric directly from the generated OTU table itself.

By instead going through all the steps of alignment etc. and then making a new OTU table separately, you are potentially introducing other parameter differences that make the ultimate comparison not apples-to-apples.

Cheers,
-jon

Kate Blackwell

unread,
Apr 14, 2016, 5:56:11 PM4/14/16
to qiime...@googlegroups.com
I was following the steps listed here:

http://qiime.org/tutorials/chimera_checking.html

It advises removing the chimeric sequences from the OTU table by making a new one, but I will try filtering and see what happens as your explanation makes sense.


Edit:  After trying the filtering command instead, the sequence and OTU numbers make more sense.  Thanks!  I needed to use the final OTU table from the open reference command, wherein taxonomy had been assigned.

jonsan

unread,
Apr 15, 2016, 12:35:09 PM4/15/16
to Qiime 1 Forum
Great! Thanks for the update, and glad it worked for you.

-jon
Reply all
Reply to author
Forward
0 new messages