2

I have RNA sequencing data (= expression of a certain protein) from the lung tumor of 14 patients, 5 non-smokers and 9 smokers.

If I want to compare the expression level of that protein of smokers vs non-smokers, which test should I use? The dataset doesn't seem parametric, so I usually try a Wilcoxon test then, but these are not paired data, no?

Thank you very much!

mdewey
  • 17,806
Kobe
  • 21
  • If you want to compare the smoker and non-smoker group, why don't you just use a regression instead? Actually 14 is a very small sample size for whatever you do to generalize your comment on the population. That is may be another problem. – Blain Waan Nov 13 '13 at 19:42
  • There are two 'Wilcoxon tests'; the signed rank test for paired data and the Wilcoxon-Mann-Whitney test for unpaired data (Wilcoxon invented the rank sum test, Mann & Whitney made substantial advances in the theory of such tests - via U statistics - and in the ability to tabulate the statistic in question). – Glen_b Nov 13 '13 at 23:05
  • How is 'expression' measured? – Glen_b Nov 13 '13 at 23:05
  • Consider negative binomial. – SmallChess Apr 04 '17 at 15:58

0 Answers0