How to: Protein-protein interaction networks for non-model organisms (marine bivalve)?

Question

[mo_vo_]

I have a set of proteins of interest (as fastas, with no NCBI Accession nr.) from a non-model organism (the beautiful marine bivalve Ruditapes philippinarum) and I would like to map possible protein-protein interaction.

• Under what conditions can I use STRING networks predicted for different species?
• Are Blast2Go sequence descriptions suitable input names or are there more appropriate/better options?
• Is there a more appropriate/better tool for my purpose?

Any help, suggestions, comments and criticism is very welcome!

Thanks

Answer 1

[Jean-Karim Heriche]

I would suggest building an interaction network by orthology. You first infer orthologs of your proteins of interest in model organisms (e.g. using TreeFam, you can put your proteins into existing phylogenetic trees) then extract corresponding protein interactions from various databases. When the orthology relation is not one to one, you could weight the interactions by something like 1 over the number of orthologs.

[mo_vo_]

Thanks Jean-Karim I will give it a go.