Stand Alone Otu Generation Program (Ngs)

Question

[Daniel]

I'm looking for a program for generating OTUs (as the title suggests!) which I will be able to incorporate into a pipeline I'm working on. I know RDP/Mothur is the holy sacrament for diversity work but I just want to extract the OTU generation portion.

Really, what I'm looking for is an output saying:

@1 PID: x OTUs  
@2 PID: y OTUs
@3 PID: z OTUs etc

and then have a list with a representative sequence for each OTU at each PID for megablasting/databasing.

jMOTU covers all the bases, but its a java application so I cant integrate it. ( http://www.nematodes.org/bioinformatics/jMOTU/index.shtml )

All recommendations appreciated!

[Pierre Lindenbaum]

why can't you integrate it "because it's a java app" ?

[Daniel]

Maybe I'm mistaken, but the only way I can use jMOTU is through the GUI and there's no command line interface to put into a script. Anything I am going to write is going to be perl though, as I have no java knowledge. Am I missing a trick?

[Pierre Lindenbaum]

ah ok, it's a gui problem, not a java problem.

Answer 1

[Louis Letourneau]

Mothur will do this. http://www.mothur.org/wiki/Main_Page

In the end you'll get for different cluster distances (OTUs pariwise ditance of 0.01, 0.05, 0.1, etc), the number of otus found and the reads that fall into those otus.

pick one of the reads, or use the gapped read alignment to build your consensus sequence.

There are a lot of good examples on the mothur wiki to generate the OTUs.

And it's 'command-lineable'.

Answer 2

[Andreas]

UCLUST could be an option

It's also used by QIIME for OTU picking. Have a look at the manual.

Andreas

[Science_Robot]

Don't forget to mention that MG-RAST uses it now too!

...but it's closed source

[Daniel]

I think this sounds like the most appropriate. Im only just looking into qiime now but like the look of it, and if this is what its OTU generation is based on it sounds solid to me. cheers.

Answer 3

[Science_Robot]

You can use CD-HIT to get representative sequences at a specified level of similarity

For example, 98% for "species" with 16S sequences.