文章基本信息

标题：Improving Protein Docking Using Sustainable Genetic Algorithms
本地全文：下载
作者：Emrah Atilgan ; Jianjun Hu
期刊名称：International Journal of Computer Information Systems and Industrial Management Applications
印刷版ISSN：2150-7988
电子版ISSN：2150-7988
出版年度：2011
卷号：3
页码：248-255
出版社：Machine Intelligence Research Labs (MIR Labs)
摘要：AutoDock is a widely used automated protein docking program in virtual screening of structure-based drug design. Several search algorithms such as simulated annealing, traditional genetic algorithm (GA), and Lamarckian genetic algorithm (LGA) are implemented in AutoDock to find optimal conformation with the lowest binding energy. However, the docking performance of these algorithms is still limited by the local optima issue of simulated annealing and traditional evolutionary algorithms (EA). Due to the stochastic nature of these search algorithms, users usually need to run multiple times to get reasonable docking results, which is time-consuming. We have developed a new docking program AutoDockX by applying a sustainable GA named ALPS to the protein docking problem. We tested the docking performance over three different proteins (pr, cox and hsp90) with more than 20 candidate ligands for each protein. Our experiments showed that the sustainable GA based AutodockX achieved significantly better docking performance in terms of running time and robustness than all the existing search algorithms implemented in the latest version of AutoDock. AutodockX thus has unique advantages in large-scale virtual screening. var currentpos,timer; function initialize() { timer=setInterval("scrollwindow()",10);} function sc(){clearInterval(timer); }function scrollwindow() { currentpos=document.body.scrollTop; window.scroll(0,++currentpos); if (currentpos != document.body.scrollTop) sc();} document.onmousedown=scdocument.ondblclick=initialize evolutionary algorithms are able to find better results while traditional genetic algorithms get stuck in local optima. According to a recent survey of protein docking algorithms [1], there are more than 50 protein-ligand docking softwares, mo st of which still used traditional GAs for conformation search optimization. Our experiments implied that other modern protein-ligand docking programs can also be potentially improved b y the sustainable genetic algorithms. II. Background A. Protein-Ligand Docking Protein docking is a method that predicts the bound conformation of one protein to another protein or a ligand. A docking algorithm aims to find the best orientation of these two molecules such that they have the minimum binding energy as scored by a predefined scoring function. There are two key components in a docking algorithm: a good scoring function with high selectivity and efficiency that distinguishes between correctly or incorrectly docked structures and a search algorithm that can efficiently do global minimization of the scoring function [34-36]. Protein-ligand docking algorithms can be classified into two methods. In early docking algorithms, both protein and ligand are considered as rigid bodies and they have only six degrees of translational and rotational freedo m to search for best orientations. Since the number of degrees of freedom is large if the proteins are modeled as flexible, it is impractical to perform exhaustive conformational search. Most of current docking algorithms consider the flexibility of ligands to find the best binding position between small molecules (ligands) such as sub strates or drug candidates and structurally known target proteins (see Figure 1). Interaction between proteins produces no change in conformation. Flexibility of ligands comes fro m the rotatable bonds (also called torsions) of a ligand (see Figure 2). The number of optimization variables is composed of six degrees of freedom for rotation and translation plus the number of torsion angels. The ligand finds its position into the protein's active site after a certain number of moves (searches) in its conformational space. Flexibility modeling allo ws the ligand to change its structure with the torsions angles. Each move costs energy, and after moves are completed, total energy is computed by the system. Our go al is to minimize this binding energy to find the best conformation. Figure 1. An example of protein-ligand docking B. Search Algorithms in AutoDock AutoDock (Auto mated Docking So ftware for Predicting Optimal Protein-Ligand Interaction) is a suite of automated docking tools. AutoDock is widely used as a docking engine Figure 2. A ligand with rotatable bonds (torsions). in virtual screening [37-39] for predicting ho w small molecules bind to a receptor of kno wn 3D structure. In AutoDock [40], a ligand and a protein are defined by a set of values describing the translatio n, orientation and conformation of the ligand with respect to the protein. The target protein is represented as a grid. This three dimensional grid surrounds all ato ms of the protein. Each atom in a protein has its o wn points in the space. The representation of a ligand consists of 3 coordinates o f the location of the ligand , followed by the 4 quaternio n parameters , which define the orientation of the small molecule, and followed by the number of torsions , depending on how many rotatable bonds the ligand has [8] (see Figure 3). These are the state variables of the ligand, and each state variable corresponds to a gene. The ligand 's state corresponds to the genotype, and the atomic coordinates of the state corresponds to the phenotype [27]. Figure 3. Representation of a ligand as a vector Autodock implements three conformation search algorithms for docking including simulated annealing (SA), traditional genetic algorithm (GA), and Lamarckian genetic algorithm (LGA). 1) Simulated Annealing In early versions of AutoDock, Simulated Annealing (SA) was used as the major optimization method [41-43]. Simulated annealing is a generic probab ilistic method for global optimization. The algo rithm starts fro m a random or specific state with an initial temperature parameter (T0) and a specific cooling scheme [41]. At each step of the simulation, the ligand explores the conformation space by adding a small random displacement in each degree of freedom and evaluating the binding energy for the new conformation, which is composed of the intermolecular energy between the protein and the ligand and the intra-molecular energy of the ligand. It repeatedly searches the neighborhood and selects a neighbor as a new state. New energy is compared to the energy of the previous step. If the new energy is lower, the step is accepted. Otherwise, if the new energy is higher, the decision is made probabilistically based on a temperature (T) parameter. Because simulated annealing is a kind of a Monte Carlo method, different runs may produce different solutions 249 Atilgan and Hu
关键词：autodock; protein docking; genetic algorithm; HFC; ; sustainable evolutionary algorithms