启动子的潜在语义索引差异识别算法  

Genomic Promoter Recognition Algorithm Based on Difference Latent Semantic Indexing Model

作　　者：秦洋[1] 王立宏[1] 武栓虎[1] 宋宜斌[1] 

机构地区：[1]烟台大学计算机科学与技术学院,山东烟台264005

出　　处：《烟台大学学报（自然科学与工程版）》2010年第3期211-216,共6页Journal of Yantai University(Natural Science and Engineering Edition)

基　　金：国家自然科学基金资助项目(60772028);山东省自然科学基金资助项目(Y2008G08;Y2006G22)

摘　　要：启动子是基因前面的一个短序列,定位启动子即能找到基因,因此启动子的识别具有重要意义.潜在语义索引差异模型(DLSI)能够扩大类间的差异,降低噪声的干扰,从而提高识别的精度.本文提出基于DLSI的启动子识别算法,利用潜在语义索引差异模型进行特征词选择,通过将样本变换到潜在语义空间中实现降维,建立了启动子-外显子、启动子-内含子分类器,把两分类器的串联结果作为最终的分类结果.实验验证了算法的有效性.Promoter is a subsequence before gene,and locating promoter is equivalent to the locating of gene,so promoter prediction is significant in genomic data analysis.Difference latent semantic indexing（DLSI） can enlarge the difference of classes and reduce noise,thus improve the accuracy of classifier.In this paper,a promoter prediction algorithm based on DLSI is proposed,which selects feature terms based on the difference of feature terms of two classes,and maps the sample into latent semantic space to reduce dimension.Promoter-exon classifier and promoter-intron classifier are set up,and the conjunction of the two classifiers gives the final class label.Simulating experiment illustrates the validity of this method.

关 键 词：潜在语义索引 差异模型 启动子识别 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]