A Mathematical Model for Affymetrix GeneChip Probe Level Data

Affymetrix GeneChip PM-MM probe pair is designed with the intension of measuring non-specific binding. Though the rationale behind the design id that a PM probe is expected to have a larger value than that of the MM probe, there are many exceptions in actual data. We gave an explanation for this inconsistency based on the assumption of functional states of a gene ‘ON/OFF’. Our hypothesis on PM-MM probe pairs is that the logarithmic of PM and MM values have the same distribution when gene is in OFF state. It means that the probability of MM > PM is expected to be equal to that of MM < PM for OFF genes. The validity of the hypothesis was given by inter-platform comparisons using common targets among three different types of platforms. Keywords: Affymetrix Gene Chip; Binomial distribution; Mathematical modeling; Oligonucleotide microarray; ON/OFF genes Three commercially available single-color oligonucleotide microarray platforms for profiling gene expression - CodeLinkTM Human Whole Genome Bioarray (~55,000 transcripts, GE Healthcare, Tokyo, Japan), Affymetrix GeneChipTM Human Genome U133 Plus 2.0 array (~47,000 transcripts, Affymetrix, Santa Clara, CA), and Agilent Whole Human Genome Oligo Microarray (~41,000 transcripts, Agilent Technologies, Tokyo, Japan), were compared using the same RNA sample extracted from a human neuroblastoma tissue sample.