As said, the paper reported interactions of about 1,000 proteins most of which were new. These interactions are useful for people who work on the proteins involved. In fact, I guess that our data will keep researchers busy for years. We also developed a new way to do two-hybrid screens that efficiently identifies false positives, which used to be a major problem of this technique.

Good question. The idea to do "proteomics" gradually developed while I was working on my Ph.D. at the EMBL in Heidelberg. I tried to find protein interactions of a single protein—a vertebrate formin, and had a hard time trying to find a handful of candidates. I figured that there must be a more efficient way to identify protein interactions, especially given the risk that not every protein works with a particular assay. When I heard that Stan Fields planned to screen all yeast proteins for interactions in a systematic way I thought: "That’s the way to go—even if 9 out of 10 proteins won’t work there are still 600 that should work!"

We generated a record number of protein interactions and developed a new, automated method to detect protein-protein interactions on a large scale. Such protein interactions are as important as the interactions of the parts in an engine: if the parts don’t work together properly, the whole system breaks down. That’s why protein interactions are crucial for biological systems.

Dr. Peter Uetz, group leader at the Institute of Toxicology and Genetics, Research Center Karlsruhe.