Skip to main content
. 2014 Aug 21;42(16):10288–10306. doi: 10.1093/nar/gku722

Figure 5.

Figure 5.

Six variants of MIPA approach. In Variants 1–3 (panels A, B, C), competitive tests evaluate all five measures, just like self-contained tests. Variants 4–6 (panels D, E, F) use a single t-test between the distribution of pathway activity measures observed per sample in the given dataset and that in randomized gene sets. Variants 1 and 4 (panels A, D) combine P-values using Stouffer's Z-score method and then use the larger value of the two as the final P-value for a pathway. Variants 2 and 5 (panels B, E) directly combine the P-values from all tests into a single overall value using Stouffer's Z-score method. Variants 3 and 6 (panels C, F) assign the largest P-value as the final value for a pathway.