机构地区: 上海大学
出 处: 《生物信息学》 2010年第3期202-205,共4页
摘 要: 基于蛋白质结构字母的预测和分析方法,一个必然的步聚,是将目标蛋白质离散成结构字母序列。本文在对蛋白质结构字母序列空间,及其最小根均方偏差变化,穷举分析的基础上,提出了一种新的蛋白质结构字母序列优化算法,全局贪婪算法。全局贪婪算法避免了基本贪婪算法过度依赖候选集大小,计算量过大、以及过早收缩于局部最小等缺点。经实验分析,全局贪婪算法在性能上优于基本贪婪算法和局部最优方法。。 For Any Fragment-based protein structure prediction and analysis,it is essential to discrete each protein structure into a series of protein structure alphabet.The paper proposed a new method to optimizing the structure alphabet sequences of proteins with the base of exhaustive analysis all protein structure sequence and the variation of least root mean square deviation(LRMSD),global greed algorithm.The method avoids the weaknesses of basic greed algorithm which highly depends on candidate set size,demands more computer power,degrade rapidly.Experimental result show,the performance of global greed algorithm is superior to basic greed algorithm and local optimal methods.
领 域: [自动化与计算机技术] [自动化与计算机技术]