[BioC] edge-less pathways in graphite

[Paul Shannon]

(Getting this conversation back on the list)

Gabriele Sales predicts the fix for the problem reported by Hamid will be in devel before long.

  - Paul

Begin forwarded message:

> From: Gabriele Sales <gabriele.sales at unipd.it>
> Subject: Re: Fwd: [BioC] edge-less pathways in graphite
> Date: April 26, 2012 1:41:56 AM PDT
> To: Paul Shannon <pshannon at fhcrc.org>
> 
> Dear Paul,
> 
> we have checked this pathway and, in fact, we have found a problem.
> 
> Our conversion software emits the following error:
> 
> [ERROR] in pathway "path:hsa04012": a relation references element 54
> that is either undefined or belongs to an invalid type
> 
> This kind of error is severe enough that any further activity on the
> pathway topology is aborted. The results is that no edge is emitted.
> Genes (i.e. nodes) are processed in an independent step, so they still
> appear in the final graph.
> 
> Let's now turn our attention to the original cause of this problem. The
> relation blocking the conversion is the following:
> 
> <relation entry1="58" entry2="59" type="PPrel">
>   <subtype name="compound" value="54"/>
> </relation>
> 
> Node 54 is defined like this:
> 
> <entry id="54" name="path:hsa05214" type="map"
>       link="http://www.kegg.jp/dbget-bin/www_bget?hsa05214">
>  <graphics name="Glioma" fgcolor="#000000" bgcolor="#FFFFFF"
>       type="roundrectangle" x="927" y="641" width="102" height="34"/>
> </entry>
> 
> As you can see it represents a pathway, but according to the relation
> above it should be a compound.
> 
> Looking at the pathway image, the relation corresponds to the chain:
> 
> PI3K -> PIP3 -> PKB/Akt
> 
> Therefore, it seems like that the middle node (PIP3) should really be
> node 55.
> 
> <entry id="55" name="cpd:C05981" type="compound"
>       link="http://www.kegg.jp/dbget-bin/www_bget?C05981">
>  <graphics name="C05981" fgcolor="#000000" bgcolor="#FFFFFF"
>       type="circle" x="385" y="649" width="8" height="8"/>
> </entry>
> 
> This error shows that our conversion is somewhat too rigid. It should be
> able to cope with this kind of problems by discarding the (single) edge,
> rather than giving up completely. We are aware of this limitation and we
> are already working on something better for the next graphite release.
> 
> 
> Best,
> g