This project is based upon the idea that, once secondary structure is specified, generation of the native protein conformation via computer simulation can be accomplished with relatively simple potential functions and algorithms. We have successfully folded several proteins (myoglobin, 1-CTF, myohemerythrin, cytochrome b-256) to a resolution of ~ 4-6 A RMS using this strategy, and our code has been implemented on both IBM workstations and the CM-5 where efficient parallelization has been achieved. We are currently working on improving the potential functions to deal with difficult cases (R69) and developing specialized algorithms for treating twisted B-strands (which were not allowed in the original algorithm). We are also investigating a new minimization algorithm, based upon the branch and bound procedure of Floudas.
Showing the most recent 10 out of 22 publications