摘要
挖掘特征基因是分析基因表达谱的一项基础工作,但众多的特征基因挖掘方法所挖掘的特征基因集合并不完全一致。本研究试图从生物功能和样本分类能力两个方面,分析比较不同方法所挖掘的特征基因的功能一致性,并对RankGene中的八种不同特征基因挖掘方法以及基于八种方法的集成法进行了功能一致性分析。结果显示:无论是生物功能分析还是样本分类能力分析,二分规则、Gini指数、方差总和三种方法的一致性较高;样本分类能力分析中,各方法所挖掘的特征基因对样本的分类准确率均较高,但不能明确区分方法间样本分类能力的优劣。
Gene mining is the basic work in the gene expression profile analysis.And lots of gene mining methods turned up with a different discriminatory gene list with the ones selected by the existed methods.We can analyze the function concordance of the genes selected by different gene mining methods.The function concordance is not only about the classification of the genes,but also the biological function.In this study,we applied the function concordance analysis on the eight gene mining methods in RankGene,as well as an integrated method based on the result of the former eight methods.The result showed that there was little difference among twoing rule,Gini index and sum of variation on both biological function and classification of selected genes.The concordance of the nine gene mining methods was good in classification,and it was hard to rank the performance of the eight methods by classification,though some methods were not quite concordant in biological function.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2010年第2期212-219,共8页
Chinese Journal of Biomedical Engineering
基金
国家自然基金资助项目(30871394)
国家高技术研究发展(863)计划(2007AA02Z329)
黑龙江科技攻关(ZJG0501)
