摘要
目的比较SAM方法、Bonferroni校正法、BH法等6种基因表达谱数据筛选差异表达基因的方法,探讨了各种方法的筛选效果。方法采用6种待比较的方法处理不同样本量和方差条件下的模拟数据,比较筛选结果与模拟设定的差异,计算相应的考核指标,探讨6种方法筛选差异表达基因的效果。结果Bonferroni校正法、Sidak校正法和Hochberg法可将FWER、FDR控制在很低的水平,但是筛选出的差异表达基因数比较少;成组t检验方法筛选的差异表达基因数最多,但是不能有效地控制FWER、FDR水平,筛选出的差异表达基因假阳性数过多;相同样本量和方差条件下,SAM方法和BH法筛选差异表达基因数、假阳性数、FWER和FDR均相差不大,均筛选出较多的差异表达基因,且控制了多重检验错误率。结论SAM方法适用于基因表达谱数据筛选差异表达基因的数据分析。
Objective To compare the efficiency of the SAM,Bonferroni adjustment method, BH procedure, et al, six kinds of screening differentially expressed genes methods. Methods Six methods were used to analyze the simulated data with different sample size and vari ance. The results were compared with the simulated initialization, and the assessing indexes were calculawA to compare the efficiency. Result Bonferroni adjustment method, Sidak adjustment method and Hchberg method can Control FWER and FDR at a very low level,but the number of screening differentially expressed genes of the three methods was small. The number of screening differentially expressed genes of t-test was the largest, but the false positive number was the largest, too. The t-test can't control FWER and FDR effectively. There was little difference in the number of screening differentially expressed genes, the false positive number, FWER and FDR between SAM and BH procedure on the same sample size and vafiance conditions. The two methods can not only screen more differentially expressed genes, but also control multiple test error rates. Conclusion The SAM was suitable to screen differentially expressed genes for gene expression profiles.
出处
《中国卫生统计》
CSCD
北大核心
2009年第3期250-254,共5页
Chinese Journal of Health Statistics
基金
陕西省科技计划项目(2008K04-02)
国家自然科学基金资助项目(39900126)
陕西省自然科学基金资助项目(2003F11)