Methods for detecting inter-protein covarying sites

Covarying sites are defined to be sites in a protein whose rate of evolution changes over time. We design software to group protein sites into three rate pools: conserved, variant, and temporary invariant. Other software is written to find sites which are closely correlated. The algorithms used by the software require a multiple sequence alignment and phylogenetic tree as input and rely heavily on tree-corrected information entropy. Through a study of the protein Cu, Zn Superoxide Dimutase it is shown that temporary invariant sites have interactions with at least one site which is either closely correlated or binary-switching. From this result it is reasonable to assume that temporary invariant sites which interact with no such intra-protein sites must be sites of protein-protein interaction. Temporary invariant sites are also shows to reflect the animal plant divergence.