Fisher exact test

The Fisher exact test is a statistical approach for comparing the composition of two collections (organsisms, genes, etc.). It is used to test the hypothesis that a given gene is not differentially represented in two cDNA libraries, as sampled by random sequencing. The result of the equation

p= [ N! M! c! C! ] / [(N+M)! a! b! A! B!]

is the probability value p.

The null hypothesis is rejected if

p <= 0.05/G, where 0.05 is the level of statistical certainty and G is a correction factor based on the Bonferroni inequality, equal to the total number of possible tests. In the case of DDD, this value is:
G = #UniGene clusters represented in analysis * #Pools * (#Pools - 1) / 2

Therefore, the p value depends on both the difference in the number of times a gene is represented in two cDNA libraries AND the total number of genes in the library.