Functional Assessment Of Genes

Question

[Abhi]

We have identified a set of genes to be differentially expressed between treatment and control. The current gene annotation doesnt have functional information about the gene.

What we would like to do is asses the biological functions of these genes( may be based on GO and then cluster them). I was wondering if there is something already out there to do so in semi automated way.

I am thinking of blasting them against pfam or be myhits dbase but wanted to take your opinion.

PS : cross posted on BioConductor mailing list

Thanks! -Abhi

[Larry_Parnell]

The approach you take and the usefulness of BLAST, GO, or other tools in relation to even other datasets to which you can compare depends on which organism you're studying.

[Sean Davis]

What are the IDs for your genes? Are they Entrez Gene IDs, unigene, or what?

Answer 1

[Hamish]

If the proteins corresponding to these genes are already in the public databases then you could try looking at InterPro Matches. For each release of the UniProtKB and UniParc protein sequence databases InterPro run InterProScan on all the sequences and produce a set of XML files which detail the signature matches of the InterPro member databases (PFam, ProDom, ProSite, SMART, etc.) and the corresponding InterPro identification. The InterPro Matches databases are available from the InterPro FTP site, InterPro web site, InterPro BioMart, dbfetch, WSDbfetch and SRS@EBI. If you don't have UniProtKB or UniParc identifiers for the proteins the PICR and UniProt ID Mapping services provide a means of using the identifiers you have and mapping them to UniProtKB and/or UniParc, or using the protein sequence to find the identifiers of identical sequences described in the databases.

If the sequences are unknown then you could try feeding them through InterProScan, to get InterPro and optionally GO annotations. You can either download the software and install the software locally or use the InterProScan REST or SOAP web services provided by EMBL-EBI to process the data.

Another option is Blast2GO which provides a number of annotation options, including an InterProScan based process which uses the EMBL-EBI web services.

[Abhi]

Thanks for detailed reply Hamish ..I think InterProScan is the way to go...

Answer 2

[Alex Paciorkowski]

One tool I often use for this task is GEDI. You can batch search for multiple genes at once, and it pulls multiple annotations for you (i.e. from GO and other sources). You can also export your list from GEDI to network algorithms such as STRING. Good luck!

[Abhi]

Thanks Alex..unfortunately my data is not human. I am not sure if GEDI which looks like a wonderful tool will work for Chlamydomonas ..

[Alex Paciorkowski]

Ah. Yes, well then I would say this resource would not be of much assistance. Sorry about that. Good luck nevertheless!

Answer 3

[Larry_Parnell]

The approach you take and the usefulness of BLAST, GO, or other tools in relation to even other datasets to which you can compare depends on which organism you're studying. For example, do not neglect genetics to provide information on function. If the gene has been knocked out or over-expressed in your organism or similar (say mouse KO for human functional annotation), that can reveal much about function of the encoded protein.

[Abhi]

Larry : thanks for your reply. We are studying Chlamydomonas and for now just want to see if the genes have some known domains inside them based on which we could atleast approximate some function.

[Larry_Parnell]

Then, follow Hamish's suggestion, but also mine by looking for KO data to confirm or refute those predictions in terms of extending functional domains to biological response.