融合多机器学习方法的siRNA在线设计系统  

siRNA online design system based on multi-machine learning methods

作　　者：王璿[1] 张法[2] 陈伟[1,3] 

机构地区：[1]燕山大学信息科学与工程学院,河北秦皇岛066004 [2]中国科学院计算技术研究所,北京100083 [3]中国环境管理干部学院信息工程系,河北秦皇岛066102

出　　处：《燕山大学学报》2015年第5期458-463,共6页Journal of Yanshan University

基　　金：国家自然科学基金资助项目(61232001;61202210);河北省高等学校科学技术研究青年基金资助项目(QN2015133)

摘　　要：siRNA设计是RNAi研究中的一个重要部分。由于靶向基因可分割成数以千计的候选siRNA,找到其中最有效的siRNA具有一定的挑战性。本文融合特征分析研究成果和多机器学习方法,设计并实现了一个siRNA在线设计系统。将目标RNA的二级结构作为影响siRNA干扰效率的评分因素,以挑选靶向合适位置的siRNA序列。对于给定的目标基因,系统经过设计得出若干高效siRNA序列的沉默效率及其相关信息。实验测试结果表明,本系统具有较高的siRNA有效性预测精度。siRNA Design has attracted attention as an important issue in RNAi research. As a targeted gene could have thousands of potential siRNAs, finding the most efficient siRNAs among them had a huge challenge. In this paper, a siRNA online design system was developed. It included multiple technologies such as feature analysis, machine learning methods. For filtering siRNA sequences on suitable location, the secondary structure of target RNA was also acted as a scoring factor which effect siRNA interference efficiency. To given target genes, this system showed effective siRNA sequences and related information sorting by the silencing efficiency. Compared with existing software and methods, siRNA Online has higher prediction accuracy.

关 键 词：SIRNA 有效性预测 机器学习方法 在线设计 

分 类 号：TP393[自动化与计算机技术—计算机应用技术] Q811[自动化与计算机技术—计算机科学与技术]