On Retrieving Protein-Protein Interactions From String-Db

Question

[Woa]

I need to understand something about the operations in String-DB. Say I have two Human proteins, gene names being ACTB and ITGB7(Which I know to interact) and I need to get ALL the interactions of these two proteins(along with interactions WITHIN the interactors), supported by two prediction methods viz. 'Experiments' and 'Databases' and at 'high confidence' value 0f 0.7000.

When I submit the two proteins using "multiple names" option I get the initial complex of ACTB bound to ITGB7. To expand this complex ie. to get the 'first neighbors' of these two proteins, which option(s) do I choose? Do I specify an arbitrary large value say 1000 in 'additional (white) nodes' field ?

Are there any suitable way to iterate at various depth (level 1, 2,...) to retrieve data?

Thanks

[Pierre Lindenbaum]

Did you see the API for string-db? http://string-db.org/help/index.jsp?topic=/org.string-db.docs/api.html

[Woa]

Hello Piere,

I know about the string API but would like to fetch data interactively this time.

[Alex Paciorkowski]

By "interactively" -- do you mean using the graphical interface? I would use the API for this, as Pierre suggests.

[Woa]

Yes Alex, I want to use the graphical interface. I think thru' the GUI String doesn't give a network with more than 2000 nodes, so a value of 2000 in the 'additional (white) nodes' field would be more appropriate.

[Woa]

One more thing I've noticed, while using the STRINGClient in cytoscape for retrieving interaction ACTB and ITGB7 separately, ACTB was NOT reported as interaction partner for ITGB7 and vice versa.

Answer 1

[Michael Kuhn]

If you really need all interactions, you have to go through the download files, as both the API and the web interface have cutoffs for the number of interaction partners that will be shown.

For a cut-off of 0.7, you should be able to select the right channels and add 1000 white nodes as you suggest. In the end, how do you want to interpret your result?