Significance Analysis of Prognostic Signatures
Description
Functions implementing the Significance Analysis of
Prognostic Signatures method (SAPS). SAPS provides a robust
method for identifying biologically significant gene sets
associated with patient survival. Three basic statistics are
computed. First, patients are clustered into two survival
groups based on differential expression of a candidate gene
set. P_pure is calculated as the probability of no survival
difference between the two groups. Next, the same procedure is
applied to randomly generated gene sets, and P_random is
calculated as the proportion achieving a P_pure as significant
as the candidate gene set. Finally, a pre-ranked Gene Set
Enrichment Analysis (GSEA) is performed by ranking all genes by
concordance index, and P_enrich is computed to indicate the
degree to which the candidate gene set is enriched for genes
with univariate prognostic significance. A SAPS_score is
calculated to summarize the three statistics, and optionally a
Q-value is computed to estimate the significance of the
SAPS_score by calculating SAPS_scores for random gene sets.