Simbios Talk by Jean-Claude Latombe, Stanford University, November 14, 2007

Title:  Computational Tools to Study Protein Flexibility

Folded proteins are not rigid.  Their flexibility has long been recognized as a crucial property for achieving their functions.  For example, the so-called conformational selection model suggests that a protein exists as an ensemble of rapidly inter-converting folded conformations and that a ligand selects the most proper conformations to bind to from this ensemble. In this talk, I will present several tools to analyze and/or predict protein flexibility (mainly of protein fragments). I will consider three distinct topics.  First, I will argue that kinematic singularities in a protein backbone play a significant role in its stabilization, by making it difficult for atoms to move along certain directions. Next, I will present a new approach to modeling structural heterogeneity from X-ray crystallographic data.  Instead of extracting a single conformer for an electron-density map (EDM), we extract a distribution of conformers (each with some fractional occupancy) that provides a near-optimal explanation of the data in the EDM.  Finally, I will discuss new results with our loop conformation sampler (LoopTK) to sample and recognize Calcium-binding loop conformations.
This talk is based on work done with Ankur Dhanik, Peggy Yao, Henry van den Bedem (JCSG), Guanfeng Liu, Jim Milgram (Math Dept.), Tsai-Yen Li (Natl. Chengchi Univ.), Danny Neil, Nathan Marz, Ryan Propper, and Charles Ku. The new results with LoopTK to recognize Calcium-binding loop conformations are part of a joint research project with Russ Altman and Inbal Halperin.