Figure 2

Illustration how the global p-value is calculated. On the left ((a) and (c)) the cumulative p-value distribution between 0 and 0.05 is shown for the data set (red line) and the random sets (black or gray lines). For each distribution its maximal rank is determined and the maximal rank of the data set (red arrow) is compared to the maximal ranks of the random sets ((b) and (d)). The upper two panels exemplify this principle with three random sets and the lower two panels show the result of testing the ontology molecular function for an excess of amino acid changes in primates (see results and discussion).