Computational modeling of malarial parasite protein interactions reveals function on a genome-wide scale

Chris Stoeckert

Department of Genetics, Center for Bioinformatics, University of Pennsylvania

Over half of the proteins encoded by the genome of Plasmodium falciparum, the major causative agent of malaria in humans, are annotated as "hypothetical." The functions of these proteins are largely unknown due to lack of sufficient sequence similarity to characterized proteins or protein domains. Functions may be inferred from interactions of hypothetical proteins with other proteins. While experiments such as yeast 2-hybrid studies provide evidence for physical interactions, computational approaches based on joint evolutionary conservation, the fusion of proteins as seen in some genomes, and correlation of expression profiles extend exploration of the interaction space by taking into account functional interactions between proteins. Recently we applied a Bayesian framework integrating conservation, protein fusion, and expression to model the protein interactome for Plasmodium falciparum (Date & Stoeckert, Genome Res. 2006). The resultant network provides functional inferences for over 2000 hypothetical proteins. Ongoing work is extending this approach to other genomes such as related parasites, Plasmodium vivax and Toxoplasma gondii, as well as to the human genome.