Hi Michael Love,
I am pretty new in this. I have "wt" and "ko" RNA-seq samples, each three replicates for differential gene expression analysis. sample description as follows:
wt group = 2 males and 1 female ko group = 3 female
I noticed that my samples are grouping based on "sex" more than "genotype". I do not have more mice to repeat this experiment but I want to make some sense out of it.
Could you please help me with how I can correct effect of sex in my analysis or how I can remove sex-based influence using DEseq analysis?
Thank you, Mahendra
Hi James,
Thanks for your reply. Does ANOVA solve this problem since it provides p-values for both comparisons- genotype and sex? Then I can remove genes those have significantly differentially expressed in sex comparison to just interpret differentially expressed genes in genotype comparison.
Thanks, Mahendra
No, ANOVA doesn't solve the problem. Your WT samples are all male, and all but one female is KO. If you fit a model like
~ genotype + sex
, then you will be apportioning the variability into two buckets that are almost the same and are likely to lose any significant results. Here's an exampleNow let's make the females all really similar.
Because sex and genotype are almost completely aliased, each one separately looks significant, but together they are not. And it's not possible to say if the difference is due to the genotype or to sex differences, or some combination of the two.
That makes sense. Thank you so much for your detailed explanation.