The Application Of Noise Characteristics Of Primary Affymetrix Genechip Data

Over the past decade, hybridization-based microarray technology is widely used in the fields of life sciences and biomedical. However, there is a serious problem exists in the measurement of gene expression level for hybridization-based microarray, since the computation accuracy of gene expression level is contaminated by much noise generated by cross-hybridization signals .Over the past few years, various mathematical models have been proposed to simulate the noise sources, in order to decrease the influence of the noise caused by cross-hybridizationThis thesis studies the probe-level data obtained from widely used Affymetrix GeneChip? and reveals the correlation between the middle three nucleotides of probes and the ratios of Mis-match/Perfect-Match probe intensities. This correlation is applied to an advanced probe-level analyses model mmgMOS, and obtained an improved new model mmgMOSⅡ. Experiment results on a standard spike-in dataset, a real mouse embryo dataset and an expanded, balanced Platinum Spike-in dataset have shown that the accuracy and efficiency of the new model are improved and the influence of the noise caused by cross _hybridization is decreased.