You could do worse than check out the bioperl pages. Either at CPAN or Bioperl.

Either site should lead you to Ewan Birney's Bio::OntologyIO::simpleGOparser or to some of the other work on handling GO files. The Bioperl community is very obliging and usually pretty handy at saving you from reiventing the wheel

Hope this helps

A.A.

There is a Perl API available for working with MySQL GO databases in the go-dev CVS repository (the entire CVS tree can be downloaded in one go, although this gives you a lot of stuff you probably won't need). There is further documentation to be found at the Gene Ontology Software Group site).

There are also a couple of scripts available on the GO FTP site which work with the flat file format. They may give you some idea of how to proceed, should you need to stick with flat files for whatever reason.

Hope this helps,

Tim

Update: Turns out, the Perl API includes a module specifically for parsing GO flat files. Have fun :-)

That module link is interesting (readmore below). It shows how GO can be passed into XML format. Is this often done? Is the XML format here easy to work with? And can anyone give examples of code with which this XML format is used? I expect that most people work with the flat files. I previously used simple but long perl scripts to pick up network paths and then did the rest of the analysis in Excel.

<readmore>

eg format: go-text for storing graphs and metadata:

  !version: $Revision: 1.19 $
  !date: $Date: 2002/12/11 01:15:01 $
  !editors: Michael Ashburner (FlyBase), Midori Harris (SGD), Judy Bla
+ke (MGD)
  $Gene_Ontology ; GO:0003673
   $cellular_component ; GO:0005575
    %extracellular ; GO:0005576
     <fibrinogen ; GO:0005577
      <fibrinogen alpha chain ; GO:0005972
      <fibrinogen beta chain ; GO:0005973

this is the following file parsed with events turned directly into XML
+:

<subgraph>
 <term>
  <acc>GO:0003673</acc>
  <name>Gene_Ontology</name>
  <is_root>1</is_root>
 </term>
 <term>
  <acc>GO:0005575</acc>
  <name>cellular_component</name>
  <rel>
   <type>isa</type>
   <obj>GO:0003673</obj>
  </rel>
 </term>
 <term>
  <acc>GO:0005576</acc>
  <name>extracellular</name>
  <rel>
   <type>isa</type>
   <obj>GO:0005575</obj>
  </rel>
 </term>
 <term>
  <acc>GO:0005577</acc>
  <name>fibrinogen</name>
  <rel>
   <type>partof</type>
   <obj>GO:0005576</obj>
  </rel>
 </term>
 <term>
  <acc>GO:0005972</acc>
  <name>fibrinogen alpha chain</name>
  <rel>
   <type>partof</type>
   <obj>GO:0005577</obj>
  </rel>
 </term>
 <term>
  <acc>GO:0005973</acc>
  <name>fibrinogen beta chain</name>
  <rel>
   <type>partof</type>
   <obj>GO:0005577</obj>
  </rel>
 </term>
</subgraph>
[download]

Is this often done?

About once a month :-)

I suspect that the XML format might be easier to work with than the flat files, but I must confess ignorance. You're quite correct that most of the work on the ontologies by the GO consortium itself uses the flat files with editors/browsers such as DAG-Edit and AmiGO. At present the flat files in the GO CVS are the most up-to-date version of the ontology (I think).

Tim

Disclaimer: I'm just an interested party with little to no inside experience; I don't speak for GO :-)

Scott
Project coordinator of the Generic Model Organism Database Project