Discovery and Removal of Systematic Pattern of Errors in Large cDNA Microarrays by Statistical Analysis

• Activity Number: 258
• Type: Contributed
• Date/Time: Tuesday, August 13, 2002 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section*
• Author(s): Shinsheng Yuan*+ and Ker-Chau Li
• Affiliation(s): University of California, Los Angeles and University of California, Los Angeles
• Keywords: Principal Component Analysis ; Microarray data analysis

Abstract

Genes that express rapid oscillatory patterns are amazingly abundant in the Stanford's Yeast Cell Cycle Data (http://genome-www.stanford.edu/cellcycle/). This was first reported in the cdc15 data by Li, Yan, and Yuan (2002); 190 genes move up-down-up-down while 363 genes go the opposite direction, down-up-down-up, for 19 consecutive time points. The affected genes amount to nearly 10 percent of the full yeast genome. The encoded proteins involve in all kind of biological activities including transportation, cytokinesis, RNA processing, transcription, translation and so on. It is hard to explain this kind of massive synchronization biologically. A more likely reason would be some systematic error caused by chip-to-chip variation. However, the standard chip to chip adjustment by removing the means does not help resolving the problem at all. In this article, we examine the data more closely. The green and the red channels are analyzed separately. We discover even more intriguing oscillation patterns in each channel. Statistical methods are applied in order to reduce the systematic bias. The strategy we take can be useful for discovery and correction of gross errors in large cDNA arrays.