hello, I did GO analysis based on two hundred interested genes and did KS test, I found some of the terms with 0 significant hit but high confidence and rank ahead. Why would this happen? are they enriched as parent terms?
GO.ID Term Annotated Significant Expected KS
GO:0003743 translation initiation factor activity 109 0 0.71 1.30E-09
GO:0046982 protein heterodimerization activity 44 0 0.29 8.40E-06
1 answer
Because the Kolmogorov-Smirnov works based on ranking all genes according to some score, and it ignores the "significative / non-significative" classification. The KS test checks if the score distribution is random between some group (the GO categories) and the rest of the population (all other genes not in the GO category being tested), or if the score distribution is enriched for the group - that is, does some group have an unusual amount of high scores?
The most famous implementation of this king of test is GSEA (paper and site).
Log in to answer this question.
