Extracting Curated Protein-Protein Interactions From Reactome

Question

[User 8848]

Dear all, I'm working on a project where I need a gold standard for protein-protein interaction. I got a recommendation to use all physical interactions from Reactome pathways (and also KEGG) but it's not clear to me how to retrieve the interactions from the pathways. There is a number of downloads on the Reactome website (http://www.reactome.org/download/index.html) but none of them seem to fit the bill. I tried to inspect the SBML and BioPAX files but I had trouble finding software which would handle such large files. I also looked at the database schema at http://www.reactome.org/cgi-bin/classbrowser?DB=gk_current but again I can't find the relevant information.

I would greatly appreciate any hints where the interaction data are so I can write a script to extract them.

Answer 1

[B. Arman Aksoy]

I think if it is only protein-protein interactions that you are interested in, then you should stick to another database provider than Reactome, such as iRef (http://wodaklab.org/iRefWeb/), BIND, BioGrid, MINT or etc... iRef pools data across providers and you can filter the interactions based on your preferences and just download the filtered ones. It is pretty easy to consume this as it is simply a CSV file.

Reactome data is useful when you are interested in transports, complex formations, phosphroylations, etc... And if you want to access this database programmatically, I would go with the BioPAX format and Paxtools library (http://www.biopax.org/paxtools.php).

Or alternatively you can use PathwayCommons.org (another data source which pools data across several data providers [Reactome included]) and use their batch download service (http://www.pathwaycommons.org/pc-snapshot/). There, you will find both BioPAX and SIF (Simple Interaction Format) dumps of the whole database which you can make use of for your own project quite easily.

Answer 2

[Robin Haw]

If you can send an email to [e-mail] with more information about the type of interaction data you are looking for, we would like to try and help you.

[User 8848]

I'll try that, thanks!

Answer 3

[Giovanni M Dall'Olio]

EDIT

I don't think you can get the data you want from Reactome. In Reactome, when they annotate the pathways, they don't make the distinction between protein-protein interactions and the other types. The same applies for KEGG.

The only thing that you may use from Reactome is the GO annotations. Each reaction in Reactome is associated to a GO term: so, you can go to GO, select all the terms that are associated with protein-protein interactions, and then retrieve all the Reactome reactions that are associated with those terms. However, I can not give you any suggestions on which terms to select.

Another approach is to use a database of protein complexes, such as the ones described in this post.

---

Old Post:

Go to the 'Download' Reactome page:

You will find many datasets there, including a set of Protein-Protein interactions in human in a tab-delimited format.

[User 8848]

I looked at that download but it states that these are not curated interactions or experimental data.