63 - Quantitative Modeling of DNA-protein Interactions
Panayiotis V. Benos, Alan S. Lapedes, Dana S. Fields, Gary D. Stormo, Washington University
We are investigating the rules behind the DNA-protein interactions, using probabilistic algorithms similar to Boltzmann machines. The network is trained on data, collected from the literature, in a multipass method. In the poster we present the general training algorithm as well as some preliminary results/predictions of the method.
64 - Recognition of Tertiary Packing
Motifs in Protein Structures using Delaunay Tessellation
Stephen A. Cammer, Alexander Tropsha, University of North Carolina at Chapel Hill
An approach to recognizing residue packing motifs in proteins has been developed based on Delaunay tessellation of protein structure. The proposed methodology termed Simplicial Neighborhood Analysis of Protein Packing (SNAPP) can be used to locate recurrent tertiary contacts in non-homologous structures as well as functionally relevant patterns in related proteins.
65 - Molecular Dynamics Simulation Study of the Human Topoisomerase I-DNA Covalent Complex
G. Chillemi, T. Castrignano, CASPUR, Supercomputing Center for University and Research Italy; A. Desideri, University of Rome
Here we present one nanosecond molecular dynamics (MD) simulation of the covalent topoisomerase-DNA complex carried out in periodic boundary condition, using parallel architecture, starting from the crystallographic structure. Structural and dynamical information obtained from MD simulations allows us to identify the DNA-enzyme interactions important for the stability of the complex.
66 - Essential Molecular Dynamics of Post-translationally Modified Gamma-B Crystallin
L. R. Cooper, D. Corne, J. C. Crabbe, University of Reading
Purpose: Simulate the structural changes that occur in Gamma-B-crystallin following post-translational modification; one of the major causes of cataract. Results: The mechanism of unfolding through simulated heating and glycation is analysed and discussed.
67 - Multi-class Protein Fold Classification Using Support Vector Machines and Neural Networks
Chris H. Q. Ding, Inna Dubchak, Lawrence Berkeley National Laboratory
Most current discriminative methods for protein fold prediction use the one-against-others approach, which has the well-known "False Positives" problem. We investigated two advanced approaches--the unique one-against-others and the all-against-all approaches--both improve prediction accuracy by 13-20% on a 27 folds problem. Support vector machine and neural networks are used.
68 - Sequence Independent Comparison of Flexible Molecules by Geometric Hashing
Zipora Y. Fligelman, Ruth Nussinov, Haim J. Wolfson, Tel Aviv University
We present an efficient general scheme for structural alignment and docking of flexible molecules. The method is protein sequence order independent and allows handling of molecules with numerous pre-defined internal degrees of freedom (e.g. hinges). The technique employs the Geometric Hashing Method and graph theoretic method.
69 - Application of a Computational Docking Procedure to Predict the Location of Heparin Binding Sites in Growth Factors
Mark J. Forster, Barbara Mulloy, National Institute for Biological Standards and Control, UK
A protocol has been developed for docking model heparin pentasaccharide fragments to known protein structures. Two validation cases, basic fibroblast growth factor and antithrombin, are shown. Predictive docking is reported for hepatocyte growth factor and vascular endothelial growth factor; predicted binding sites are compared with available experimental data.
70 - Consensus Macromolecular Surfaces and the Molecular Recognition Problem
Murad Nayal, Barry Honig, Columbia University
A novel approach is devised to detect similar binding sites in a set of proteins based on surface shape and chemical properties. The technique is applied to a number of protein families and a library of consensus binding sites is constructed for each. These consensus molecular surfaces highlight potentially critical molecular recognition determinants and are shown to be distinctive of each family studied.
71 - Studies of Amino Acid Frequencies in Integral Membrane Proteins from Different Subcellular Organelles
Johan Nilsson, Bengt Persson, Karolinska Institutet, Sweden
We have analysed transmembrane and loop regions of integral membrane proteins to find sequence features characteristic of different organelles. To eliminate homology bias, sequences were grouped into families to obtain a representative data set. The resulting families and amino acid distributions arrived at for each organelle type will be presented.
72 - Gamma Gamma ADH: A Docking Study
Erik Nordling, Bengt Persson, Karolinska Institutet, Sweden
A flexible side-chain docking technique was used to investigate gamma gamma ADHs role in bile acid metabolism. We demonstrate that gamma gamma ADH is involved in the conversion of iso-bile acids to bile acids by catalysing the oxidation of the 3 beta hydroxyl group.
73 - A Method for Comparing Flexible Objects Applied to Protein Molecules
S. D. O'Hearn, National Research Council of Canada; A. J. Kusalik, University of Saskatchewan
A new method and computer program, MolCom3D, is described for spatially decomposing protein molecules for comparison and categorization. Molecules are subjected to a spatial decomposition using octtrees. The octtrees constitute multiple structure alignments providing an efficient comparison space and compact 3-D descriptions of properties. http://www.octtech.com/MolCom3D, http://www.cs.usask.ca/projects/combi/ohearn
74 - Improved Spatial Contact Prediction from Correlated Mutation Analysis
Astrid Reinhardt, Olivier Lichtarge, Baylor College of Medicine
A new method was developed to identify spatial proximity between amino acid resiued-pairs within a protein based on correlations in their mutational behaviour. Phylogenetic tree information is used to classify mutational correlation in light of its evolutionary impact. This approach seems to improve the signal-noise ratio compared to existing methods.
75 - Structural Studies Probe the Evolution
of Spherical Viruses
John J. Rux, Stacy D. Benson, Carmen San Martin, Roger M. Burnett, The Wistar Institute
Similarities in the molecular structures and the packing arrangement of the major coat proteins in human adenovirus and bacteriophage PRD1 suggest an evolutionary link between the viruses. Structural alignments and optimized multiple-sequence alignments of the adenovirus hexons have revealed conserved elements, which are being used to search for remote homologs.
76 - Consensus Binding Site Structures at the Atomic Level: The Adenine Ring Moiety as a Test Example
Vladimir Sobolev, Yosef Kuttner, Alexander Raskind, Marvin Edelman, Weizmann Institute of Science
We have derived an algorithm to search for similar spatial arrangements of atoms around a given chemical moiety, using the adenine ring as a test example. Such a consensus structure might serve as a sufficient identifying motif to search for unknown, potential binding sites.
77 - Optimized and Evolved Protein Residue Interactions
D. M. Taverna, R. A. Goldstein, University of Michigan
We compare these optimized interactions of lattice model proteins to average contact energies derived from a population of evolving proteins. We find that highly designable structures maintain a higher average stability. This indicates highly designable structures will be very flexible toward adopting novel contact interactions for a variety of functions.
78 - Defining Functional Residues in Furin: Structure-based Evolutionary Analysis
Yezhen Ye, Dafu Ding, Chinese Academy of Sciences Academia Sinica
To explain the inhibition of furin by eglin-C mutants rationally and give guidance to further inhibitors design, we study the key residues in the furin by sequence-structure analysis. "Weighted evolutionary trace" and "correlated mutation analysis" methods are applied to detect specific residues. Afterwards, those residues were mapped onto the furin 3-D structure constructed with the in-house program "Pmodeling" for further exploration.
