A FC-GSEA Approach to Identify Significant Gene-Sets Using Microarray Gene Expression Data - Advances in Computational Science and Engineering - page 117

Fig. 1. A summary of gene set enrichment analysis approach [1]
[STEP 2] Estimate statistical significance of the computed ES
- Generate k different expression datasets by performing k random permutations on
class labels of gene expression profiles.
- Perform the computation of ES values on the generated k different datasets and
obtain a null distribution of ES.
- Use the null distribution to calculate a nominal p-value of the ES obtained in
STEP 1.
[STEP 3] Adjust significance level for multiple hypothesis tests
- For each gene-set, obtain a normalized-ES by normalizing the computed ES to
account for the gene-set size.
- Adjust false positive rate by calculating false discovery rate (FDR) for each nor-
malized-ES and finally determine statistically significant gene-sets
3 FC-GSEA : Fisher's Criterion Based GSEA
The FC-GSEA method is a gene-set enrichment analysis approach to identify signifi-
cant gene-sets showing statistically significant differences in gene expression intensi-
ties between two classes, where the Fisher's criterion is employed for gene-ranking.
The detailed description of FC-GSEA is given as below.
3.1 Motivation
In the original GSEA approach [1, 2], the statistical significance of gene-sets is
estimated based on the enrichment score (ES) calculated for each gene-set. Also the
ES computation is made by calculating the Kolmogorov-Smirnov(KS)score with
the ordered entire gene-list generated by SNR-based gene ranking method. That is,
the entire gene-list should be rearranged according to a specific ranking statistic and
the original GSEA employs the SNR for gene ranking. The signal-to-noise ratio
(SNR) for a gene i is given as follows.