[BioC] anybody parsing BioPAX 2?

[Ossi Koivistoinen]

Hi Paul,

I'm preparing master's thesis related to this subject and my instructor, 
Leo Lahti, pointed me to this discussion. In my study I'm trying to 
analyze BioPAX 2 data (from Reactome) to build network of all proteins 
that are linked to pathways and also to identify interactions between 
proteins based on interactions between pathways and other classes in 
Biopax data.

I used Rredland to parse the Biopax owl file on Mac OS X 10.4 and I've had 
no problems with Rredland. Berkeley DB and other dependencies have been 
available through fink project.

When starting the work I had no experience with R but some experience with 
other programming languages. I found the data.frame returned by rredland 
mostly incomprehensible and didn't know much about R packages and the R 
community so I proceeded in writing S4 object model of BioPAX level 2 and 
converted the rdf data to graph of objects of my personal BioPAX level 2 
implementation. In hindsight, creating this own graph representation seems 
extra work and I think it might have been better to build maybe igraph or 
graphNEL model of the rdf data directly.

Anyway, we have here BioPAX level 2 S4 object model under work and means 
for constructing it from owl file. I have used this object model to select 
only some edges of the BioPAX data to construct simpler graphs relevant in 
my study. If you are interested in the project you can also continue 
discussion by email with me (ossi.koivistoinen at iki.fi) or with Leo 
(leo.lahti at iki.fi). In particular, we would like to know if there are 
others who could potentially join and contribute to the project but all 
comments and feedback are welcome.

Best Regards,
Ossi