Elucidating Network Topology with Sequence: A Highly Specific Method for the Prediction of Physical Interactions in Non-Model Species

We describe a new technique, Elucidation of Network Topology with Sequence (ENTS), for the computational prediction of protein-protein interactions based only on primary sequence data. The ENTS classifier uses a random forest approach to efficiently predict interactions on a whole-genome scale for any eukaryotic organism, utilizing pairwise combinations of conserved domains and predicted subcellular localization of proteins. We found that ENTS can maintain at least 99.95% specificity while predicting between 22% and 64% of experimentally-verified protein interactions in a variety of model and non-model organisms. The predicted interactions are biologically meaningful, as indicated by shared functional (GO) annotations, co-occurrence in KEGG pathways, and coexpression of putatively interacting proteins in public microarray datasets. Furthermore, we demonstrate some of the biological insights that can be gained from these interaction networks, using examples of disease-disease associations in humans, clusters of metabolic pathways in Populus trichocarpa, and the evolutionary dynamics of duplicated genes in paleopolyploid plants. The ENTS classifier and will be a valuable tool for the de novo annotation of genome sequences, providing initial clues about regulatory and metabolic network topology, and revealing relationships that are not immediately obvious from traditional homology-based annotations.