CAPPI is an inference framework designed to integrate protein-protein interaction data across species and indentify highly probable interactions. We currently provide predicted versions of seven eukaryotic interactomes.

CAPPI is a statistical inference framework based on a duplication and divergence model of network evolution. It uses gene phylogenies as a guide to aggregates information from all homologous proteins, relative to their proximity in the gene tree. Intuitively, the closer a given pair of proteins is to another pair, the more impact the evidence for one pair has on predicting the interaction of the other pair.

Our Bayesian model naturally takes into account the inherent reliability and coverage of each input dataset on which the inference is based. The amount and reliability of supporting evidence, as well as the evolutionary proximity of the observed interactions to the pair of proteins in question, determines the posterior probability of interaction computed by our framework.

CAPPI framework along with its input data and results is decribed in our BMC Bioinformatics paper:
Janusz Dutkowski and Jerzy Tiuryn (2009), Phylogeny-guided interaction mapping in seven eukaryotes.