Skip to main content

A combined approach for protein structure prediction and protein-protein docking


EMSL Project ID
8491b

Abstract

Mapping the network of protein interactions in the cells of living organisms is an intriguing puzzle that we are just beginning to understand with the knowledge of completed genomes. Predicting the structure of individual protein complexes within the cell is crucial for verification of the protein interactions and gaining insight into their function. Given two independently determined protein structures described by their atomic coordinates, the protein-protein docking problem computationally predicts the best fit between them for constructing a protein complex. Computational approaches are necessary because experimental approaches will not be practical for predicting the structures of all the thousands of complexes that could form. The related protein structure prediction problem of determining the tertiary structure of a protein given its primary sequence of amino acids remains one of the grand challenges in modern science. The solution of these two problems in structural biology may benefit from knowledge-based methods when information from homologous proteins or protein interactions is available. Lacking this information, expensive physics-based methods derived from physical principles must be utilized.

We have developed a methodology for protein structure prediction that uses results from known proteins in combination with a physics-based method. This methodology is composed of two phases: the first phase generates initial configurations using secondary structure prediction as well as fold recognition servers. The second phase improves those initial configurations through a sophisticated global minimization algorithm that treats the full-dimensional global optimization problem as a series of small-dimensional ones. The optimization code has been running efficiently on the EMSL Linux cluster MPP2 for the past 2 years and this resource enabled our group to 1) compete in CASP6 achieving the best prediction for one of the most difficult targets of the competition (target T0238) and 2) explore improvements to our method subsequent to the CASP competition. A significant portion of the procedure developed for protein structure prediction will be applicable to the problem of protein-protein docking. What follows is a description of tools developed by our collaboration for protein structure prediction that will also aid in the development of our new docking method.

To support our methodology for structure prediction we developed ProteinShop, an interactive visualization tool for protein modeling and manipulating protein structures with pinpoint control, guided by the user's biological and experimental instinct. ProteinShop takes a given sequence of amino acids and generates structures containing predicted secondary structure. It allows researchers to twist and turn these configurations using concepts from robotics and animation so that structures can be reconfigured without breaking them. The use of inverse kinematics enables all the angles in the structure to move as jointed segments simulating the way joints in our bodies interact. In addition, ProteinShop allows us to monitor the changing energy profile of structures. Furthermore, ProteinShop allows us to manipulate structures generated by the global optimization process as it runs and subsequently schedule them for further optimization - thus dynamically steering the search through the vast conformational space.

The features of ProteinShop which provide a unique platform for developing and exploring protein structure prediction algorithms are mirrored in a new tool: DockingShop. DockingShop provides an interactive molecular docking environment that includes real-time visual feedback to steer the docking process for rapid estimation of the conformational binding. It allows users to bring two molecules close together and see them bind in real-time. It also provides an interface to an analytical-based method for detecting binding sites.
DockingShop visualizes computational parameters used in scoring functions to understand the behavior of structures during molecular interaction, and to aid in discrimination between native and non-native bindings. We propose to incorporate a framework for comparison of scoring functions used within the context of optimization methods, and to use DockingShop to develop a new docking method. We hope to utilize the EMSL cluster for the expensive global optimization runs of both the structure prediction and docking methods.

Project Details

Project type
Exploratory Research
Start Date
2006-05-01
End Date
2007-06-26
Status
Closed

Team

Principal Investigator

Silvia Crivelli
Institution
Lawrence Berkeley National Laboratory

Team Members

Elizabeth Eskow
Institution
University of Colorado

Lianjun Jiang
Institution
University of Colorado