Creating user-friendly databases of protein–protein interactions is a tough challenge. Pronet — a curated database of published protein–protein interactions produced by Myriad Genetics — is a brave attempt to address the problem.

Searching the database couldn't be simpler: you just type in a keyword and hit ‘go’. This returns a list of proteins whose database entries contain that word. Clicking on any entry in this list will take you to that protein's homepage, which contains a graphical representation of the protein's domain structure, links to other relevant databases, and sequence information. But Pronet's killer application is the addictive ‘view interactions graphically’ feature, which takes you to a page containing your protein in a box. Clicking on the box creates a spider diagram showing all the proteins it interacts with, and you can expand the network as far as you like. A ‘mouse mode’ menu allows you to delete proteins, move them, or squeeze the network to make room for more interactions. What's more, if you come across a protein you're unfamiliar with, choosing ‘info’ from the mouse mode menu will link you back to that protein's data entry page.

Pronet does have limitations: it contains only human sequences, although there are plans to include other species. It also records only interactions found using the yeast two-hybrid system, which creates some idiosyncrasies. For example, the c-Src page states that there are ‘no recorded interactions for this protein’. Links to papers describing the interactions would also be useful. That said, as more data are added, Pronet will evolve into an invaluable tool for anyone wanting to track protein–protein interactions.