Hello,
I have two conditions (condition A and condition B) and 4 groups (group 1, group 2, group 3).
I want to compare condition A to condition B within each of the groups, performing differential expression analysis using DESeq2. Then I want to contrast the results between groups, to determine which genes are uniquely differentially expressed in each group.
My problem is that my group sizes are very unequal:
Group 1: 37 x condition A; 50 x condition B
Group 2: 14 x condition A; 33 x condition B
Group 3: 5 x condition A; 5 x condition B.
I was wondering whether it is recommended to use DESeq2 as normal in this instance, using adjusted p-values and LFC values, or whether it is recommended to obtain the standardized effect sizes or something similar? I am concerned that I will obtain a falsely large number of SDE genes for group 1 compared to group 2 and 3 as a result of the much larger sample size in group 1.
Many thanks in advance for any tips that anyone has!
If you get false results then from the smaller groups and not vice versa, as larger groups better capture biological variability and have more statistical power and better estimation of the true effect size. You can combine the results to see which genes are consistently significant, e.g. by rank aggregation, see https://cran.r-project.org/web/packages/RobustRankAggreg/index.html
Hi @ATpoint, thanks so much for your reply!
I am not sure if rank aggregation is what I am looking for - I would like to compare the results between groups rather than aggregate them all. I suppose I am looking for an approach that can account for the sample size in the discovery of SDE genes thus enabling more fair comparison between groups.
Then you will have to downsample I guess. Rank aggregation would just tell you which genes are consistently called as differential.