How to test for gender bias in RNA seq experiment ?

Question

[mohsamir2016]

Dear All, I performed an RNA seq experiment comparing virus-infected animals vs control animals. The goal is to see how these two animal groups would differ in gene expression. I realized that infected animals were of 13 male + 11 female and in the control animals: 15 male , 16 female. As you know sex chromosome could have genes that bias the analyses. How can I test if this gender variation between the two groups could bias the analyses, so could shift somehow the expression of some gene towards one sex. I performed chi-square to test association and it was non significant. Any ideas

[cpad0112]

if you have used DESeq2/EdgeR, include sex in model (~ condition+sex). If other variables in experiment are influenced by sex, include interaction in the model.

[mohsamir2016]

Thanks for the answers. I already created an nMDS. Here I run 3 nMDS:

How do you think is the sepration. I run Permanova to test significant differences among groups (see P-value above the figure). Now which is these figures are best answering the question ? Beacuse I assume looking into each group (e.g. within infected). if male and female seprated then there is a sex bias inside this group, meaning that infection did affects gene expression in a sex-dependent manner ? am I right

I also though about logistic regression analyses as a statitical approach to test this ? how do you think ?

[Jeremy Leipzig]

use DESeq2/EdgeR as suggested and get back to us if you need help

Answer 1

[fracarb8]

You could run an initial analysis without accounting for the sex (e.g. ~condition), and look at how your samples split based on few PCs.

If there is a clear difference between the two groups, then you need to account for that variability like cpad0112 suggested (e.g. ~Sex + condition).

In the figure above, the samples are not splitting by sex, so it should be safe to avoid the correction.