Convert Demo Pathways from Pathway Commons about pathways2go HOT 11 CLOSED

Starting on 'Adenylate cyclase activating pathway' from Reactome. PC URL http://www.pathwaycommons.org/pc/record2.do?id=485728

from pathways2go.

Only modifications needed to convert the PC commons version of the Reactome pathway was to change the expectation for the 'db' field in the XREFs that connect things to uniprot, chebi, and GO. Reactome uses strings like UniProt, CheBi, GENE ONTOLOGY while PC uses strings like 'uniprot database', chebi, gene ontology. So here we have a good example of why it would preferable to use URIs for naming things over otherwise unspecified strings...

from pathways2go.

Looking at WikiPathways ACE Inhibitor Pathway http://apps.pathwaycommons.org/view?uri=http%3A%2F%2Fidentifiers.org%2Fwikipathways%2FWP554 .

Its going to be a lot more tuning to make this work.

For a start, gene entities are typed as DNA in BioPAX - and have both NCBI gene ids and UniProt ids. And some of them don't have any ids. Also looks like they are using a lot of different process process representations, none of which are our trusty 'biochemical reaction'.

from pathways2go.

Need to do something with protein A interacts with protein B. (direct instantiations of the Interaction class in biopax.)

Plan is to create a generic molecular function node, make it a part of the biological process for the pathway, add the proteins to it as participants, and then link them to each other with 'interacts with'.

Without that structure, I think it will be hard to connect the interaction event to the pathway in the GO-CAM model.

from pathways2go.

Have a working conversion of WikiPathways via PC now. Need to come back to defining victory here as 'working' currently means that the resulting model is OWL consistent, loads in Noctua, and basically seems right. Something more formal would be useful.

Note that the current pattern looks pretty atrocious in Noctua for both the reasons elaborated on before (complexes) and because it results in a proliferation of un-typed molecular function nodes with no causal flow.

from pathways2go.

The ACE inhibitor pathway from WikiPathways functional. http://apps.pathwaycommons.org/view?uri=http%3A%2F%2Fidentifiers.org%2Fwikipathways%2FWP554 checking the tickbox..

from pathways2go.

Working on kegg http://apps.pathwaycommons.org/view?uri=http%3A%2F%2Fidentifiers.org%2Fkegg.pathway%2Fhsa00780

First run, found a reversible reaction. Need to add a pattern for that..

from pathways2go.

Making the assumption that reversible reactions run in a left to right manner. (Which is at least half right.) If GO develops suitable patterns for representing reversibility, this can be revisited.

from pathways2go.

Next fun condition for the kegg pathway is that they add the Controller entities as parts of the pathway so they were showing up as blank mf nodes. Added constraint that mf nodes not be Control entities. (The information we want regarding the controller entity enabling the target reaction is still coming through.)

from pathways2go.

Kegg biotin pathway conversion is functional. Right now it has very little GO annotation (I think only the top level process). Investigating use of provided EC numbers to map to MFs.

from pathways2go.

Added EC mapper. Ends up with multiple MF types per node as there are multiple ECs on the reaction and (I suspect) one EC could quite likely map to more than one GO term. I think this is okay. Declaring this done.

from pathways2go.