It's not that clear what you are trying to achieve. For inferring protein-protein interactions, I would use STRING. For inferring protein domain-domain interactions I would use DOMINE although you would need to annotate your sequences with Pfam first.

Go through this paper and their software: RCDP , I think its part of DOMINE:

Co-evolutionary analysis of domains in interacting proteins reveals insights into domain-domain interactions mediating protein-protein interactions

InterPare; Paper: A protein domain interaction interface database

InterDom is a database of putative interacting protein domains derived from multiple sources, ranging from domain fusions (Rosetta Stone), protein interactions (DIP and BIND), protein complexes (PDB), to scientific literature (MEDLINE).

I think this article can also be useful :

Martin Weigta, Robert A. Whitea, Hendrik Szurmantc, James A. Hochc, and Terence Hwaa. Identification of direct residue contacts in protein–protein interaction by message passing

http://www.pnas.org/content/early/2008/12/30/0805923106.short