A database server for predicting protein-protein interactions

yLarge-scale protein interactions are known for several species due to the recent improvements in experimental methods for detecting protein interactions. However, direct determination of all the interactions between the human proteins is difficult even with current high-throughput methods. This paper describes a database server called HPID (http://www.hpid.org) that (1) provides structural interactions between human proteins precomputed from existing structural and experimental data and (2) predicts structural interactions between proteins submitted by users. The structural interactions were obtained by finding known structural interactions of PDB in SCOP domains and then by finding homologs of the domains in target proteins. Based on the structural interactions, we constructed two protein interaction maps, one for human and another for yeast. We believe this is the first attempt to map a whole human interactome at the superfamily level and to compare a human protein interaction map with other species' interaction map.