Pathway expression scores based on the mean squared rank out-performed the mean rank, as assessed by precision-recall curves for the tissue-species data.
The authors ranked genes by expression level and computed the mean squared rank.
The mean squared rank was chosen based on a survey of statistical approaches for gene set analysis [ 34], and out-performed other summary statistics in a series of classification benchmarks based on tissue-specific pathway expression (see benchmarking section below).
In addition, an identical analysis pipeline was also constructed using the GSEA algorithm, as applied to single samples [ 25], as the initial step, in place of the mean squared rank.
There was also a much greater computational burden associated with running GSEA on 180,000 arrays compared with using the mean squared rank on the same number of assays.
For a pathway, P, of size k, represented in an array by genes G 1, G 2...G n, the pathway expression score, En P), is defined by the mean squared rank E n (P ) = 1 n × ∑ i = 1 n R i 2, where R i is the rank of gene G i in a pathway containing n genes.
