Upgrading Isoforms to proteins?

[pablopareja]

I need some insight on how we are going to do this? @rtobes @eparejatobes

[pablopareja]

Here's an example of how they're specified:

<comment type="alternative products">
<event type="alternative splicing"/>
<isoform>
<id>Q13362-1</id>
<name>Gamma-3</name>
<sequence type="displayed"/>
</isoform>
<isoform>
<id>Q13362-2</id>
<name>Gamma-1</name>
<sequence type="described" ref="VSP_005112"/>
</isoform>
<isoform>
<id>Q13362-3</id>
<name>Gamma-2</name>
<sequence type="described" ref="VSP_005113"/>
</isoform>
<isoform>
<id>Q13362-4</id>
<name>4</name>
<sequence type="described" ref="VSP_043645 VSP_005113"/>
<note>No experimental confirmation available.</note>
</isoform>
<isoform>
<id>Q13362-5</id>
<name>5</name>
<sequence type="described" ref="VSP_046768"/>
<note>
Ref.3 (BAG63760) sequence is in conflict in position: 3:N->K. No experimental confirmation available.
</note>
</isoform>
</comment>

Here's the link for the entry XML file http://www.uniprot.org/uniprot/Q13362.xml

[pablopareja]

hello? @eparejatobes @rtobes :cactus:

[eparejatobes]

The general principles we agreed on (I think) were

• isoforms become proteins
• there should be a relation (isoform or something similar) between them, making proteins a union of codiscrete graphs

What we need to decide on is whether we want to have different rel types for the different "events" or we will make this a property.

[rtobes]

I think that event is not so important and it's better as property.

The most important thing for isoforms is their sequence and the fact that functions and annotations in general can be assigned to one specific isoform and not the the others.

@epareja knows more than me about isoforms

[pablopareja]

Would it be interesting to retrieve for instance all isoforms that were generated by let's say frameshift?

[rtobes]

In your example frameshift is an event or a feature?

In any case yes, it would be interesting

[pablopareja]

I meant an event

[pablopareja]

So first-level Isoform vertices exist now in Bio4j linked to Protein entities via ProteinIsoform edges. The event that generates the isoform is modelled through IsoformEventGenerator edges going from Isoform vertices to AlternativeProduct vertices. Closing this...