透過MA-plot上lowess curve之變化，本文提出一統計量E(|I|)以評估晶片的品質。當E(|I|)值愈大，表示晶片品質愈差。透過統計模擬，發現E(|I|)可以有效區別晶片間的品質。此外我們更進一步找出診斷品質的臨界值|I|*。當一晶片品質統計量值高於|I|*時，表示該晶片品質不良。我們亦利用此品質測度值來評估三家公司所製造的microarray data的品質，並以RT-PCR來(間接)驗證所提之E(|I|)的判別績效，得到不錯的結果。本文所提出之E(|I|)不僅可評估各公司之晶片品質，亦可利用它來比較各公司之製作品質，因此在實務之晶片品質品管，甚或了解實驗室之晶片製作上，深具實用之價值。

Microarray experiment is a useful way to observe the display for thousands of genes simultaneously. To improve the reliability of dowe-stream data analysis, a quantity to evaluate the quality of microarray chip is established. Motivated by the MA-plot of Dudoit et al. (2002), we develop a statistic E(|I|) to measure the quality of one microarray chip. Simulation study shows that E(|I|) is able to distinguish the chip quality effectively. Furthermore, we suggest one threshold value |I|* for purpose of diagnosis. The quality of one chip is acceptable if the E(|I|) value for the chip is smaller than |I|*. Three sets of microarray chips are presented to illustrate the application of E(|I|), with RT-PCR being used to validate the efficacy of the suggested quality statistic.

[1]Chan, S. H., Chen, L. J., Chow, N. H., Liu, H. S. (2005). An ANCOVA Approach
to Normalize Microarray Data, and Its Performance to Existing Methods.
Journal of Bioinformatics and Computational Biology, Vol. 3, No. 2,
257-268

[2]Dudoit, S., Yang, Y. H., Speed, T. P., and Callow, M. J. (2002). Statistical
Methods for Identifying Differentially Expressed Gene in Replicated
cDNA Microarray Experiments. Statistica Sinica, 12, 111-139.

[3]Johnson, R. A., Wichern, D. W. Applied Multivariate Statistical Analysis.
Upper Saddle River, N.J. :Prentice Hall, 4th ed., (1998), 732-733.

[4]Kerr, M. K., Martin, M., Churchill, G. A., (2000). Analysis of Variance for
Gene Expression Microarray Data. Journal of Computational Biology, 7,
819-837.