This is not a trivial task depending on how well studied your network is or what information you want to gather. It sounds like an interaction dataset is what you want rather than scouring the PDB for static structures (that would be such a hit and miss process I'm not sure it's worth bothering).
- Upload you list of gene codes to string-db.
- You'll see the network here, and structures are made clear of known complexes in an expandable thumbnail.
- It's a limited web-server tool, and whilst very powerful, isn't very flexible. Cytoscape might be needed for a more novel study because it is scriptable.
- From string-db download the
.SIFfile (I think it's an old-school save icon on the web-server). - Download Cytoscape here.
- Then use
File>Import>Table>From file. - This will build an interaction network and you'll see which genes/proteins interact with one another.
Catch: whilst this provides spacial information, you can't get a picture of temporal information: i.e you know what is interacting, but not when nor for how long.