During my research at the University of California, Berkeley and at the Lawrence Berkeley National Lab, I've worked on serveral projects, some of which are described below.
Shortly after joining Sung-Hou Kim's lab we created the first three dimensional map of the known protein universe. This research was published in the Proceedings of the National Academy of Sciences, USA in 2003.
With the success and welcome reception of our map of the protein universe we developed a 3D map of nucleic acid conformational space. This research was published in the journal, Nucleic Acids Research in 2003.
The Ramachandran or phi-psi plot is a familiar sight to mostly everyone who studies protein structure. It depicts the favorable conformations of the two backbone torsion angles, phi and psi, in protein structure. The structures of native proteins conform to a given small set of expected phi-psi values, representing the conformations of short 3 amino acid residue segments. We explore the concept of higher order phi-psi plots which cluster the conformations of fragments longer than 3 amino acids, represented by multiple pairs of phi-psi angles. We gathered torsion angles from high resolution protein crystal structures (< 1.0 Å) and segmented the backbone chain via a sliding window into various lengths from 3 to 12 residues long. The fragments are then mapped via a dimension reduction technique called Multidimensional Scaling (MDS) to yield 3D maps of conformational space. These maps show clustering of peptide conformations that reveal a dramatic reduction of conformational space. This research was published in the Proceedings of the National Academy of Sciences, USA in 2005.
We are using the conformational information from the protein conformational space to develop a software package, HOPPscore for analyzing the structural quality of theoretical and experimental models of protein structure. The HOPPscore server is available at http://hoppscore.lbl.gov. We are not currently distributing a stand-alone version. We described the results of this research in Proceedings of the National Academy of Sciences, USA in 2006.
Another project in development is the mapping of the drug-like chemical universe. In much the same way that protein conformational space is restricted to a given set of preferred conformations, we believe that the chemical universe also exists as discrete sets of compounds which are related to each other by atomic connectivites and properties.
Current Resources we're finding useful in the lab now.Updated Mar. 16, 2007