Go Terms In Ensembl

Question: Go Terms In Ensembl

9.9 years ago by

Pi • 520

Pi • 520 wrote:

Greetings

When using ensembl, database entries are linked to GO terms at the transcript level. According to the documentation in the Perl api, the GO terms listed are those associated with the Swiss Prot entry for the transcript. These GO terms are then linked directly to the transcript as xrefs via the appropriate tables in the database (object-xref and ontology_xref). However, as far as I can see, there is no method (using the database directly) of determining the category of a GO term(biological process/cellular component/ molecular function). There do not seem to be any fields in the database schema that capture this information

Please can you advise if there is a method of determining the GO term category.

EDIT: In response to the second answer, this code resides in the perl api for retrieving ontology terms

SELECT DISTINCT
        term.term_id,
        term.accession,
        term.name,
        term.definition,
        term.subsets,
        ontology.namespace
FROM    ontology
  JOIN  term USING (ontology_id)
  JOIN  synonym USING (term_id)
WHERE   ( term.name LIKE ? OR synonym.name LIKE ? ));

but where are these tables (term/ontology/synonym). They are not in the main databases and they are not in the ontology_mart database either on the biomart server

Thank-you for your time

gene ensembl • 3.5k views

ADD COMMENT • link •

modified 9.6 years ago by Travis • 2.8k • written 9.9 years ago by Pi • 520

9.9 years ago by

Travis • 2.8k

USA

Travis • 2.8k wrote:

If you're using the API, you need to set up an ontology adaptor:

e.g.

my $goadaptor = $registry->get_adaptor('Multi', 'Ontology', 'GOTerm');

You then need to pull the dblinks for your transcript object

i.e.

my $trandblinks = $transcript->get_all_DBLinks;
my @trandblinks = @{$trandblinks};
            while (my $trandblink = shift @trandblinks) {
                    my $tranprimid = $trandblink->primary_id;
                    my $trandispid = $trandblink->display_id;
                    my $trandb = $trandblink->dbname;

Then use the ontology adaptor to get the required details...

                     if ($trandb eq "GO") {

                            my $term = $goadaptor->fetch_by_accession($trandispid);
                            if ($term->namespace) {
                                    my $process = $term->namespace;

etc.

This will get you the process category...

Hope it helps!

ADD COMMENT • link modified 9.9 years ago by Neilfws ♦ 49k • written 9.9 years ago by Travis • 2.8k

Thanks for your reply. Do you know where this information is stored in the underlying database? I prefer to work with the db directly if possible.

ADD REPLY • link written 9.9 years ago by Pi • 520

9.8 years ago by

Travis • 2.8k

USA

Travis • 2.8k wrote:

Are you doing this work with local copies of the databases?

The ontology information is stored in a database called ensembl_ontology_6*.

The naming convention is different from the other databases as it is species independent.

ADD COMMENT • link written 9.8 years ago by Travis • 2.8k

it would seem this is the missing link i was looking for!!

ADD REPLY • link written 9.8 years ago by Pi • 520

9.9 years ago by

Neilfws ♦ 49k

Sydney, Australia

Neilfws ♦ 49k wrote:

You can go from a transcript ID to GO terms with categories using BioMart. Briefly:

Click "MARTVIEW" (top menu)
Select database (Ensembl genes 61) and dataset (e.g. Homo sapiens genes)
Click "Filters" (left menu)
Under GENE check "ID list limit" and either paste or upload Ensembl transcript IDs
Click "Attributes" (left menu)
Under EXTERNAL check "GO Term Accession (bp)", "GO Term Accession (cc)" and under go molecular function (mf), "GO Term Accession"
Click "Results" (top left menu)

This will return a table with columns for BP, CC and MF and GO IDs where they exist for the transcript.

ADD COMMENT • link written 9.9 years ago by Neilfws ♦ 49k

BioMart has updated for v62. I would like to point out the new 'GO domain' attribute which will tell you which of the three categories the GO term fits into (i.e. biological process, cellular component or molecular function).

ADD REPLY • link written 9.8 years ago by Giulietta - Ensembl Helpdesk • 1.2k

Please log in to add an answer.

Similar posts • Search »