Book of Abstracts: Albany 2009

category image Albany 2009
Conversation 16
June 16-20 2009
© Adenine Press (2008)

Pharmacophore Modeling Using a Reduced Protein Representation: Application to the Prediction of ATP, GTP, Sialic Acid, Retinoic Acid and Heme-Bound and -Unbound Nitric Oxide Binding Proteins

Due to increased activity in high-throughput structural genomics efforts around the globe, there has been a steady accumulation of experimentally solved protein 3D structures lacking functional annotation, thus generating a need for structure-based protein function assignment methods. Prediction of ligand binding sites is a well-established protein function assignment method. Here we apply the specific ligand binding site (LBS) detection algorithm we recently described (Reyes, V.M., 2008a) to 801 functionally unannotated experimental structures in the PDB, screening for the binding sites of 6 biologically important ligands: GTP in small Ras-type G-proteins, ATP in ser/thr protein kinases, sialic acid, retinoic acid, and heme-bound and unbound nitric oxide. Validation of the algorithm for the GTP- and ATP-binding sites has been previously described; here, validation for the binding sites of the 4 other ligands showed acceptable specificity and sensitivity as well. Of the 801 structures screened, 1.0% tested positive for GTP binding, 7.6% for ATP binding, 4.4% for sialic acid binding, 16.5% for retinoic acid binding, 4.1% for heme-bound nitric oxide binding, and 1.2% for unbound nitric oxide binding. Using the ?Cutting Plane? and the ?Tangent Sphere? methods described previously, (Reyes, V.M., 2008d), we also determined the degree of burial of the ligand binding sites detected. These ligand burial measures were compared with those in the respective training structures, and the degree of similarity between the two values as taken as a further validation of the predicted LBSs.

Vicente M. Reyes

Dept. of Biological Sciences
Rochester Institute of Technology
Rochester, NY 14623

Ph: 585-475-4115