How to extract edges from GPML file ?

[Krithika-Bhuvan]

I’m trying to figure out how to extract edges from a network using the GPML file . I’m using the example of WP1589.gpml. I think I need to extract the information from the <Interaction> tag. These tags have <Point> child tags which I'm guessing will give me information about the edges.

Question 1 - See screenshot. If you see [[9]] and [[10]] in the list, they each have two <Point> tags that could represent the an edge from the “From node” to the “To node” . But see [[11]] in the list - it has 3 <Point> tags, and one <Anchor> tag. Which would be “From node” node and “To node” ? Can you clarify this ?
Link to screenshot here: https://drive.google.com/file/d/1nl3GnqLSbnQR5asWIkPqgcJKbjCh7tsH/view?usp=sharing

Question 2 - Attached is a csv file of the nodes I extracted with the help of Martin Morgan’s code in Slack. I found some GraphIDs (example “bf654”) which are in the <Interaction> section, but not defined previously in <DataNodes> section. So don’t know if this is a gene, or metabolite or anything else. How do I get around this ?
Link to csv file here: https://drive.google.com/file/d/1maGS4Im6jzkO2Tz1_4W9LkSTzpkp_N9T/view?usp=sharing

[AlexanderPico]

Hi. Before getting to your questions, let me also point you to our documentation for RDF versions of our content. This format is already structured for interaction-based use cases as each "triple" is an interaction. Basically, we've already done the hard work of extracting interactions from XML (GPML) and putting into a more interaction-friendly format.
https://www.wikipathways.org/index.php/Help:WikiPathways_Sparql_queries#Get_all_interactions_for_a_particular_pathway.

Q1 - In order to extract interactions from GPML, you'll want to focus on the <Point> tags that contain GraphRef. These point to <DataNode> GraphId and other defined entities in the GPML. The presence and meaning of ArrowHead in the Point tags will indicate the "from" and "to". Some interactions are undirected, so there is not an inherent "from" and "to" encoded.

Q2 - Interactions can connect lots of different things, including Groups, Shapes, Labels, and Anchors on other interactions. The subset of GraphRefs that point to <DataNode> GraphId are the ones that directly connect molecules. But the one that terminate on <Group> or other interaction anchors are obvioulsy also important for understanding how molecules are connected.

These are not simple interaction networks; they are curated pathway diagrams. Biology is messy and so are the models :). Again, one goal of the RDF is to simplify access to interactions, but it is all encoded in the GPML ultimately, if not obviously.

[Krithika-Bhuvan]

Thank you so much ! This is very helpful. I was not aware of this RDF format. I was under the impression that GPML was the way to go. I would be happy to try out querying the RDF file format. In that case would be my query of interest "Get all interactions for a particular pathway." My next question would be - could you point me to the document I should look at , that would allow me to do this query in R ? Yes, we indeed biology is messy :) Thanks, Krithika

…

On Wed, Feb 12, 2020 at 2:33 PM Alexander Pico ***@***.***> wrote: Hi. Before getting to you questions, let me also point you to our documentation for RDF versions of our content. This format is already structured for interaction-based use cases as each "triple" is an interaction. Basically, we've already done the hard work of extracting interactions from XML (GPML) and putting into a more interaction-friendly format. https://www.wikipathways.org/index.php/Help:WikiPathways_Sparql_queries#Get_all_interactions_for_a_particular_pathway . Q1 - In order to extract interactions from GPML, you'll want to focus on the <Point> tags that contain GraphRef. These point to <DataNode> GraphId and other defined entities in the GPML. The presence and meaning of ArrowHead in the Point tags will indicate the "from" and "to". Some interactions are undirected, so there is not an inherent "from" and "to" encoded. Q2 - Interactions can connect lots of different things, including Groups, Shapes, Labels, and Anchors on other interactions. The subset of GraphRefs that point to <DataNode> GraphId are the ones that directly connect molecules. But the one that terminate on <Group> or other interaction anchors are obvioulsy also important for understanding how molecules are connected. These are not simple interaction networks; they are curated pathway diagrams. Biology is messy and so are the models :). Again, one goal of the RDF is to simplify access to interactions, but it is all encoded in the GPML ultimately, if not obviously. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#10?email_source=notifications&email_token=AANXTHBPY5Q2FGGWDWA5FR3RCRFIBA5CNFSM4KT6PDIKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOELSCW2Q#issuecomment-585378666>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AANXTHELBU7UGRHBCR5NXHTRCRFIBANCNFSM4KT6PDIA> .