UCHIME - file preparation

[Lawrence Davies]

Hi QIIMERS,

I am analysing fungal sequences based on the ITS region. There are no
aligned ITS fungal databases for use in ChimeraSlayer that I am aware.
I have decided to use a de novo chimera removal step using UCHIME.

The UCHIME documentation states that the input file must list the OTUs
with sequence abundances using the format e.g.

>FQ56TRV12;size=14

Do you know of any method so that the output from pick_rep_set.py also
displays the sequence abundance? Or, do you know how much of a
difference it would make if the abundance data is not given?

Many thanks,

Lawrence

[Jose Carlos Clemente]

Hi Lawrence,

uchime uses abundance information to break ties among potential
chimeras. For instance, if sequence C is a chimera formed from A and
B, uchime will find the triplet (A,B,C) and then use abundance to tag
C as chimeric (since its abundance will be lower than A and B).

So if you don't provide abundance, uchime won't have a way to tell
which is which, and I think results will be wrong 2/3 of the time as
the ties will get randomly resolved.

The script pick_rep_sets.py does not provide that information, so what
you'd like to do instead is use pick_otus.py following the
instructions detailed here:

http://www.qiime.org/svn_documentation/tutorials/otupipe.html

Jose

[Lawrence Davies]

Hi Jose,

 

The de novo chimera removal has worked really well. I compared it to ChimeraSlayer and the number of chimeras detected was within a few percent for both techniques. I have tried it on some sequences without a reference database and I had around 50% chimeras at 97% similarity threshhold. This seems like a lot (although not unheard of) but I wanted to check a few myself to make sure they are actually chimeras.

 

I was going to use a feature in Arb to do this - Arb takes into account the secondary structure of RNA so by comparing the parent sequences to the 'chimeric sequence' I might see a number of potentially poorly called nucleotides at either side of  a helix in the chimera that are not in the parent which would give us some indication that it probably really is chimeric. It's not definitive but I can't think how else to do it.

 

So my question is - is there a function in uchime that tells you which sequences have been highlighted as the parent sequences of the chimera? or do you know of any other way to test if a potenitally chimeric sequence is in fact a chimera?

 

Cheers,

 

Lawrence

[Jose Carlos Clemente]

Lawrence,

I believe the uchime_de_novo_chimera_filtering.log file will give you
the high abundance (parent) sequences, but that is about it. If you
ran OTUpipe/usearch outside of QIIME there might be a way to get more
information, but that would be out of the scope of this forum. You
might want to try that or ask directly Robert Edgar if such
information is available somewhere.

As for ways to test whether chimeras are chimeras, I guess there is
not much more available than what you have already tried: compare with
other tools and align to structure to find things that seem
suspicious.

Jose

On Wed, May 30, 2012 at 12:29 PM, Lawrence Davies

[Lawrence Davies]

Hi Jose,

 

OK thanks for your help. I guess it is a bit of a grey area really.

 

Cheers,

 

Lawrence