1: \begin{abstract}

2: Gene Set Enrichment Analysis (GSEA) is a basic tool for genomic data

3: treatment. From a statistical point of view, the centering of its

4: test statistic does not 

5: allow the derivation of asymptotic results.

6: A test statistic with a different centering is proposed.

7: Under the null hypothesis, the convergence in distribution of 

8: the new test statistic is proved, using the theory of empirical

9: processes. The limiting distribution can be computed by Monte-Carlo

10: simulation. The test defined in this way has been called Weighted

11: Kolmogorov Smirnov (WKS) test. The fact that the evaluation of the

12: asymptotic distribution serves for many different gene sets

13: results in shorter computing times. Using

14: expression data from the GEO repository, tested against the MSig

15: Database C2, a comparison between the classical GSEA test and the new

16: procedure has been conducted. Our conclusion is that, beyond its

17: mathematical and algorithmic advantages, the WKS test could be more

18: informative in many cases, than the classical GSEA test.

19: \vskip 2mm \noindent

20: \emph{Keywords:} GSEA, statistical test,

21: empirical processes, weak convergence, Monte-Carlo simulation

22: %Some other potential keywords: Slutsky's theorem, Continuous mapping theorem,

23: %Donsker classes, stochastic integral, Kolmogorov-Smirnov statistic

24: 

25: \vskip 2mm \noindent

26: \emph{AMS Subject Classification:} Primary 62F03; Secondary 60F17

27: 

28: \end{abstract}

29: