1. Predicting domain-domain interactions from protein-protein interaction network. ? ? We have postulated that protein-protein interactions evolved in a parsimonious way. Consistently with the parsimony postulate, we identified the interacting domain as the smallest (weighted) set of pairs that is necessary to explain the protein interaction network. This method has proven to outperform previous approaches to prediction of domain-domain interactions (ref 1).? ? 2. Graph theoretical approaches to study stability of phylogentic characters. ? ? Previously, we studied the properties of the so called character overlap graph. The nodes of such graph are characters (where a character is any attribute that can by used in inferring phylogeny) and there is an edge between two characters if they occur together in some organism. Understanding of properties of such graph provides insights into possible modes of evolution of the corresponding characters. Our previous studies focused on studding the overlap graph of protein domains. In the last year we extended this approach to general study of evolutional stability of phylogenetic characters. Based on our graph-theoretical results, we postulated that some introns are unstable and therefore should not be used in inferring phylogeny (ref 2).? ? 3. Directed studies of intron conservation.? ? Using intuitions developed trough graph-theoretical studies of introns, we conducted a rigorous analysis of intron conservation. We demonstrated that the pattern of intron conservation support the Coelomata Clade of Animals From a Rigorous Analysis of the Pattern of Intron Conservation (ref 3).
