Protein–protein interactions literally hold cellular processes together, but keeping up with the ever-expanding literature on what might bind what is enough to make you fall apart, and it's likely to get worse as whole-genome scans for protein–protein interactions churn them out in bulk (see the review by Albertha J. M. Walthout and Marc Vidal on page 55 of our January issue). PreBIND — a new adjunct to the Biomolecular Interaction Network Database ( BIND) built by researchers at the Samuel Lunenfeld Research Institute in Toronto — helps you to search the literature for possible interactions, and then submit bona fide interactions to BIND.

The version now available is a prototype that searches for protein–protein interactions in the yeast Saccharomyces cerevisiae, using proteins described in the NCBI's RefSeq database, but there are plans to add more organisms. PreBIND is simple to use, and the help page is a model of clarity. You can search PreBIND by entering a protein name or RefSeq accession number. PreBIND then returns a list of papers that discuss other molecules that interact with your protein of interest. The power behind PreBIND is a trained alogorithm that allows it to recognize papers that discuss interactions. For that reason, it searches much more thoroughly than you could doing a simple PubMed search. For each paper that it finds, PreBIND lists the proteins discussed in the paper and provides a score of the likelihood that the paper contains interaction information. You can then review the paper, decide whether or not it discusses a true interaction, and submit your response to BIND. This will then be refereed by a moderator before being added to BIND. Over time, PreBIND and BIND should help us to gain valuable and reliable insights into molecular interactions.