Selecting Protein Protein Interaction Databases

Question

[Sudeep]

Hi all,
I am looking for protein protein (or gene gene) interaction databases where the interactions (edges) between the interactors are well defined, I mean the edge should be either a pathway or known molecular reaction and not just the type of assay or pmid references for the interaction . Based on this I have narrowed my selection to KEGG, Reactome and I am currently looking through humancyc. Do you know about any other interaction database where I can get this information ? I have already gone through usual String, IntAct, MIP ... databases and I couldn't find this information on most them. I am looking for human or mouse interaction data.

Thank you.

[Alex Paciorkowski]

This question gets asked quite a bit. Here is a search of the site:

http://www.biostars.org/search/?q=protein+protein+interaction+databases

Beware the false positive, unvalidated interaction. Also, I'm not certain how frequently the STRING dev team updates their dataset either. There are known interactions that just aren't reflected in STRING. These tools have a long way to go, and should be used very cautiously.

Answer 1

[VS]

I think STRING is the best resource available, it is only a question of taking out what you need from it. You can get the flat files and parse out only those interactions supported by sources you want like 'Kegg_Pathways' . You could use the evidence annotation in STRING to filter out interactions from text-mining etc. and retain only those supported by experimental evidence. They have several evidence type available --

Existing evidence channels: neighborhood, fusion, cooccurence, array, experimental, database, textmining.

STRING even scores the strength of each interaction observed based on probabilities combined from different sources of evidences.

Alternatively, you can try ConsensusPathDb , it combines several resources you prefer for human, mouse and yeast.

[Sudeep]

After posting this question, I looked into ConsensusPathDb, I like their diagram with type of interactions listed for the databases.