apply Kruskall or other non-parametric test for multiple comparisons

0

Entering edit mode

bhgyu ▴ 30

@bhgyu-13069

Last seen 6.4 years ago

Hello there,

I've been struggling to find a non-parametric test to determine significantly differentially expressed genes in sample sets with lots of features.

Running Wilcox etc always gives me the same p-value even if I do something like:

ttest <- apply(dataset, 2, kruskal.test)

Does anyone know how to determine top features non-parametrically?

Cheers!

r kruskal ttest nonparametric • 1.6k views

ADD COMMENT • link 7.0 years ago bhgyu ▴ 30

0

Entering edit mode

Why do you want a non-parametric test?

ADD REPLY • link 7.0 years ago chris86 ▴ 420

0

Entering edit mode

Hi Chris, cause I'm dealing with very small n of samples and can't be certain conditions for parametric tests are met.

ADD REPLY • link 7.0 years ago bhgyu ▴ 30

0

Entering edit mode

I'd just use limma or one of the other bioconductor packages like DESEQ2.

ADD REPLY • link 7.0 years ago chris86 ▴ 420

Login before adding your answer.