Unequal Sample Number

Question

[ARich]

Hi,

I have very a simple question. Actually I working with miseq and I have genotype with two conditions mutant and nonmutant. The number of mutant samples are more than non-mutants. So my question is does it effect on the box plots and wilcoxon test? Ofcourse box plot show the median value of each group but will it effect the p values on doing the statistical analysis?

Thanks in advance

Answer 1

[Sean Davis]

I am assuming the context is RNAseq, but correct me if I am wrong. An unequal number of samples should not be a problem for an appropriate statistical test.

[ARich]

Its miseq bacterial 16s rRNA samples basically DNA. So how many minimum number of samples are required to decide which test to proceed with. I means if number of samples are 4 to 9 for each group will wilcoxon test be appropriate?

[Sean Davis]

I'd suggest testing using RNA-seq specific tools, but I don't do rRNA analysis. edgeR or DESeq would probably be fine. The sample size will depend on the expected signal. In general, you will need larger sample sizes if the signal is small relative to the noise.