February 28, Thu 2013
4:00 pm, MRB 200 Conference Room
Dr. Zhong-Ru Xie
Skaggs School of Pharmacy & Pharmaceutical Sciences, University of California, San Diego
Ligand Docking and Structure-based Drug Design
Drug design and discovery is one of the most significant purposes for the research of computational chemistry and biology. A scoring function of docking computation is often developed and evaluated separately and considered as the biggest challenge in docking. I have created a non-energy based scoring function for docking, MotifScore, which does not consider the interactions between protein and ligand atoms independent to others. The statistical significance of the co-occurrences of several atom-atom interactions was used to score the possible binding poses generated by the sampling algorithm of docking.
Ligand binding site is the prerequisite knowledge of docking computation, and the accuracy of the binding site prediction method will affect the accuracy of docking when we do not have the ligand binding complex structure of the target protein. I used network approach to create a novel binding site prediction method, LISE, by using the protein triangles which consist of three protein atoms and have different propensities to occur in the ligand binding site.
Ligand binding site comparison provides us a new way for drug discovery. Many proteins with different sequences, structures and functions bind to the same small molecules because of the similarity of the local structure of the binding sites. We can develop drug candidates for the binding pockets of interested targets by searching similar pockets with binding ligands. However, we still need some good test sets to evaluate and compare the performance of different comparison methods.