possible bug in parseKGML?
1
0
Entering edit mode
@martin-pirkl-11782
Last seen 2.5 years ago
Switzerland, Basel, Eth Zürich

Hi,

how is the "binding/association" tag handled, when a xml is parsed to a graph object in R? E.g. for the entry

 <relation entry1="47" entry2="53" type="PPrel">
<subtype name="binding/association" value="---"/>
</relation>

I get an arrow from 47 to 53. However, shouldn't it be bidirectional, because the direction is not known/clear?

Pathway example: http://www.kegg.jp/pathway/hsa04660

When I parse this pathway, I get a path from ZAP70 to LCK via binding/association, but I don't get one from LAT to NFAT. Am I missing something?

Here is some example code:

    res <- retrieveKGML(pathwayid=04660, organism="hsa", destfile="temp.xml", method="internal", quiet=TRUE)
keggpw <- parseKGML("temp.xml")
kgraph <- KEGGpathway2Graph(keggpw)


Best,

Martin

kegggraph • 727 views
1
Entering edit mode
@zhang-jitao-david-5737
Last seen 2.6 years ago
Switzerland

Dear Martin,

Do I get you correctly that you raised two questions?

The first question is about the bidirectionality. Currently it is not supported, namely only the relation from node 47 to node 53 is parsed. The reason is that I believe (though I may be wrong!) when I looked it a few years ago, bidirectional interactions are explicitly modelled in KGML files, namely with two relation items.

The second question is about the interaction between LAT and NFAT. I cannot find the relation in the diagram on http://www.kegg.jp/pathway/hsa04660 as I could do with ZAP70->LCK. Maybe I missed your question?

Best wishes,
David

0
Entering edit mode

Dear David,

Thank you for the fast reply. Do I understand correctly, that you assume a direction for the "binding/association" for nodes 47 and 53, even though the diagram does not show a direction?

In the diagram there is an undirected edge from LAT to  PLC-gamma1, which activates CAN and in turn NFAT. That's what I mean with path from LAT to NFAT.

The general issue I have is, apart from the fact that 47 is entry1 and 53 is entry2 in the xml file, there is no evidence for the edge being directed the way it is (47 -> 53) in the kgraph object. If the intention of the xml was to encode entry1 and entry2 as a directed edge no matter what specific relation ("activation", "inhibition" or "binding"), why does the diagram show undirected edges?

I hope this is not too confusing and thank you for your time spent on this.

Best,

Martin