Classification and Differentiation of Homologs and Structural Analogs

It is both meaningful and useful to study protein sequence, structure, and function in the context of evolution. In divergent evolution, homologs, or proteins having descended from a common ancestor, usually share sequence, structure, and functional properties, and an unknown protein's structure and function can be hypothesized from its experimentally characterized homologs. In convergent evolution, proteins from distinct evolutionary lineages converge to similar structures or functions, and these proteins are called "analogs". To classify proteins into evolutionary families, it is necessary to differentiate these two opposite scenarios. Statistically significant sequence similarity is commonly accepted as adequate evidence for homology. Yet in the absence of significant sequence similarity, discrimination between homology and analogy frequently requires manual work. This dissertation describes an effort in developing an automatic tool to differentiate remote homologs and structural analogs.