Best method to get functional annotation ?

[patouch74]

Hi

I have metagenomic contigs from an assembly. I'm looking for a pipeline to get functional annotation.
I've used the soft FragGeneScan to get ORF that I mapped with Interproscan with Pfam and Tiger database.

But it takes too long (I have 70 000 sequences out from FragGeneScan) so I'm asking if there is a better way to get automatic functional annotation ?

I need GO numbers for my output .
Thanks for your helps.

[dpryan]

Perhaps blast2go is faster?

[patouch74]

I will try

[patouch74]

Unfortunately blast2go is not free.

So Ive downloaded command line HMMER with database Uniprot, the output is a blast like with a list of hits and the domain name.

eg:

E-value score bias E-value score bias exp N Sequence Description
------- ------ ----- ------- ------ ----- ---- -- -------- -----------
+ 6.6e-70 238.7 5.4 7.7e-70 238.5 5.4 1.0 1 sp|A1JMT1|MATP_YERE8 Macrodomain Ter protein OS=Yersinia en
+ 3.8e-69 236.3 4.1 4.2e-69 236.1 4.1 1.0 1 sp|Q7N603|MATP_PHOLL Macrodomain Ter protein OS=Photorhabdu

But when I use HMMER on the website, I have access to accession number: do you know how can I get it with the command line HMMER ?

[WhatsOEver]

Hi patouch74,

I can't help you with your HMMER problem, but I'd like to add something to B2G.

Though there is a "pro"-version of Blast2go, the main features (blast, go mapping, annotation, interpro scan integration) are also available in the free version. Just go to http://www.blast2go.com/start-blast2go and start the JAVA web-start with at least 4Gb RAM (the more, the better).
With your dataset I would also suggest to separate your fasta file into 4-6 smaller ones and run them independently in Blast2GO (I have had some problems with the program when analyzing bigger datasets). The separated runs can be easily combined with a build-in command afterwards.

btw: You mentioned that your previous method took you "too long". What exactly is to long for you? The Blast-step in Blast2GO could in your case easily take up to 1 week to finish...

[patouch74]

Hi WhatsOEver,
Yes, I week is too long for me

If someone know something else, I'll be glad to hear it.

[juan.isaza]

Hi WhatsOEver,
Yes, I week is too long for me

If someone know something else, I'll be glad to hear it.

you could do a local blast in xml format and then upload the xml file to Blast2GO and start with the annotation process there.

There is a web server to metagenome analysis MG-RAST http://metagenomics.anl.gov , but I don't know if that server retrieve GO terms.