基因芯片表达数据的标准化策略研究

目的　从统计学、数学和生物信息学相结合的角度 ,探讨针对不同密度基因芯片 (microarray ,genechip)如何进行数据标准化处理与分析 ,寻找表达上有显著性差别的基因。方法　在Excel和Spss软件中 ,采用总强度标准化法 (totalintensitynormalization ,TIN )和局部加权线性回归标准化法 (locallyweightedlinearregressionnormalization ,LWLRN)对高低密度芯片数据进行分析处理。结果　这些方法能有效地减少系统误差产生的影响 ,使处理后的数据更加具有可比性和可靠性。结论　本研究提供的分析方法在减少系统误差基础上能较好地找出芯片上有显著性差异表达的基因 ,不过标准化方法还处于起步和发展的阶段 ,需进一步完善 ;但随着非线性技术的不断完善 ,计算机软硬件的快速发展 ,芯片海量数据分析处理方面的研究将得到新的突破

Normalization and analysis of expression data of gene chips

Objective To explore how to conduct normalization and analysis of different density of gene chips based on statistics, maths, and bioinformatics for the purpose of finding significantly differentially expressed genes. Methods By the aid of Excel and SPSS softwares, data from high or low density chips were analyzed by total intensity normalization (TIN) and locally weighted linear regression normalization (LWLRN) methods. Results These methods could effectively reduce the influence resulted from the system errors, so the processed data were more comparable and reliable. Conclusion The methods mentioned in this paper can be helpful for the finding of significantly differentially expressed genes on the basis of decrease of system errors. However, normalization methods, still at the beginning and developing stages, need to be further improved. With the development of non linear technology, software, and hardware of the computer, great achievements will be made in the studies of the analysis of enormous amount of microarray data.