Statistics and Data Analysis for Microarrays Using R and Bioconductor, 2nd Edition

918 Statistics and Data Analysis for Microarrays Using R and Bioconductor

zero. These features point to the exponential distribution as an appropriate

model for the random variable P F :

f(x) = λe

−λx

(28.7)

Under the null hypothesis, differentially expressed genes would fall on the

pathway randomly, and would not interact with e ach other in any co nc erted

way. In other words, in the second term in Eq. 28.5 (which captures the influ-

ence of the genes upstream) roughly half of those influences will be positives,

and half negative, canceling each other out. In such circumstances, the per-

turbation of each gene would be limited to its own measured fold