Strongly Connected Components can Predict Protein Structure

Finding the three-dimensional structure of proteins is one of the fundamental problems in molecular biology today. The improvements in throughput of classical methods for determining the structure | e.g., using x-ray di raction analysis or NMR | could not keep up with the ever-increasing speed with which proteins are sequenced. This resulted in a desire for methods for structure prediction solely from sequence data, either ab initio, modeling the molecular folding process, or homology based, using protein sequences with known structures as a template. The main idea is based on the fact that sequence similarity allows to detect homology, i.e., the existence of a common evolutionary predecessor, and thus to infer similar structure and even function virtue of this shared history [10,15]. Note, that the same structure or function does not imply a common ancestor, likewise a common ancestor does not imply a common function, but probably a shared fold.