Structure prediction of quaternary, or multi-chain, proteins using ab initio methods is a promising approach to determining how proteins fold into their native conformations. Computational structure prediction from protein sequences is critical because many quaternary proteins involved in cancer and other diseases have known sequences but structures, and thus possible biological mechanisms that are undetermined. Unlike knowledge-based methods such as homology modeling and threading, ab initio physics-based methods predict the structure of a protein from its sequence alone by finding the global minimum of a potential energy function describing the protein's interatomic interactions. The objectives of the proposed research are to incorporate innovative quaternary-specific protein representations and conformational searches into the ab initio hierarchical protein structure prediction procedure developed by Scheraga and co-workers, and to assess these new approaches in a series of test cases. Successful quaternary structure predictions would demonstrate understanding of how interatomic interactions determine structure and provide the opportunity to apply this understanding to applications such as drug design and other biological and health-related challenges. ? ?