On the design of optimization algorithms for prediction of molecular interactions

Darby Tien Hao Chang, Jung Hsin Lin, Chih Hung Hsieh, Yen Jeng Oyang

研究成果: Article同行評審

3 引文 斯高帕斯(Scopus)

摘要

This article presents a comprehensive study on the main characteristics of a novel optimization algorithm specifically designed for simulation of protein-ligand interactions. Though design of optimization algorithms has been a research issue extensively studied by computer scientists for decades, the emerging applications in bioinformatics such as simulation of protein-ligand interactions and protein folding introduce additional challenges due to (1) the high dimensionality nature of the problem and (2) the highly rugged landscape of the energy function. As a result, optimization algorithms that are not carefully designed to tackle these two challenges may fail to deliver satisfactory performance. This study has been motivated by the observation that the RAME (Rank-based Adaptive Mutation Evolutionary) optimization algorithm specifically designed for simulation of protein-ligand docking has consistently outperformed the conventional optimization algorithms by a significant degree. The experimental results reveal that the RAME algorithm is capable of delivering superior performance to several alternative versions of the genetic algorithm in handling highly-rugged functions in the high-dimensional vector space. This article also reports experiments conducted to analyze the causes of the observed performance difference. The experiences learned provide valuable clues for how the proposed algorithm can be effectively exploited to tackle other computational biology problems.

原文English
頁(從 - 到)267-280
頁數14
期刊International Journal on Artificial Intelligence Tools
19
發行號3
DOIs
出版狀態Published - 2010 六月

All Science Journal Classification (ASJC) codes

  • 人工智慧

指紋

深入研究「On the design of optimization algorithms for prediction of molecular interactions」主題。共同形成了獨特的指紋。

引用此