COMPLEX SYSTEMS
PROTEIN FOLDING
At the present time there is an intense interest in predicting the three-dimensional (3D) structures of protein molecules, given their genome-derived amino acid sequences. The current effort to predict protein structures computationally is of immense importance world wide since the spatial structure of the protein must be known for detailed studies it its function.
 RIBBON REPRESENTATION three dimensional structure of a protein using a "ribbon representation" |
This research seeks to advance the state of the art in protein structure prediction via the ab initio folding method. Ab initio folding is based on the global optimization of a potential energy function and in general does not use knowledge of experimentally determined protein structures. Present ab initio folding methods require intense and exhaustive computing time, which increases as a function of the length of the protein. This limitation is due in part to the assumption that the initial condition for the ab initio folding protein is the linear sequence of residues comprising the protein as encoded by the gene. It is also due to optimizing based on all atom potential energy functions and the use of suboptimal global optimization techniques.
In our approach, we are applying the Terminal Repeller Unconstrained Subenergy Tunneling (TRUST) global optimization scheme developed by Jacob Barhen and his collaborators to perform the ab initio folding. It is a deterministic and continuous method that ensures escaping from local minima in a fast, reliable, and computationally efficient manner.
In a 1997 Science paper, TRUST was demonstrated to be superior (faster and more accurate) to several competing global optimization methods for a standard set of nonconvex functions. Our approach decouples the protein structure prediction problem into two parts. TRUST is first used to obtain a refined solution for the folds of the protein's backbone. It is a refinement process because in our approach the starting point for the global optimization is an approximate solution (prediction) for the folds of the target protein's backbone as determined by the PROtein Structure Prediction and Evaluation Computer Toolkit (PROSPECT), one of the best back fold recognition threading programs available. Upon obtaining an improved backbone prediction, we then apply TRUST to obtaining a refined solution for the target protein's side chains.
The global optimization to determine the structure of the side chains is also a refinement process because the starting point for the protein is its improved backbone prediction and an approximate solution for the side chains determined by rotamer library methods.
[