Download Interaction Networks

Question

[Markus Krupp]

Hi BioStar Community,

for my new project I have to consider directed interaction networks:

e.g.

---

Pathway1:

• A --> B
• B --| C

Pathway 2

• X --| Y

---

...and the ideal way to represent those relation for my purpose would be:

---

Pathway1 - A - B - activates

Pathway1 - B - C - inhibits

Pathway2 - X - Y - inhibits

---

Some database support those informations in graphically way, e.g. BioCarta, KEGG (needs a licenes, now /-: ), Reactome, PathwayCommons, NCI PathwayInteractionDatabase... So far, all databases I had a look at did not support a download file just supporting this format (...ok this was not really surprising (; ) ...most of them supported the BioPax level formats, but for my purpose this format is overpowered and it would need a tremendous amount of time to parse those data out of the BioPax files.

Do you know an interaction database which support gen-gen or protein-protein interaction networks which have a "low standard" download file which would be much easier to parse? ...the database should contain interaction data on humans, it would be great if they further support information to murine systems (rat, mouse)!

Thanks for your help!

Answer 1

[Giovanni M Dall'Olio]

The format you posted is similar to a SIF format. You just need to convert the BioPax files to it; there are a lot of tools for doing it, for example Cytoscape:

• Get Cytoscape
• Load your BioPax file
• Export it as a SIF format file

The SIF file will look like the following:

nodeA <relationship type> nodeB
nodeC <relationship type> nodeA
nodeD <relationship type> nodeE nodeF nodeB

It seems that there are also some command-line converters for BioPax files, for example the PaxTools library

A little suggestion coming from real life experience: after the conversion, check that the direction of the interaction (e.g. which gene inhibits which) is correct.

[paulr]

The current version of PaxTools is failing on the latest BioCarta and Reactome pathways available through the NCI PID.

[Markus Krupp]

Thanks Giovanni, indeed using cytoscape and export those interaction would be a good point. Many thanks!

Answer 2

[Khader Shameer]

You can use STRING to generate this type of data. You can get the information about activation, inhibition, binding etc under actions tab. See an example using CR1

You can download the raw data file for your organism of interest and parse these information.

[Markus Krupp]

Hi Khader,

also thanks for your help. ...STRING is a very good source for those interaction (...by the way I was involved in generating some data for STRING ;) ) ...but as you mentioned you have to parse the data and so far I try to skip that step. The data I ask for are "just" for temporary input to test scoring function of networks. In the final state we have a cooperation with some other guys who will support network data (...hopefully ;) ) Thank you!