Dear edgeR users,

In a standard edgeR analysis, when using estimateDisp() on a class 'DGEList' for a specific design  estimateDisp(y,design), there is an estimation of the common dispersion, which we can access simply by y$common.dispersion.

When using estimateGLMRobustDisp() instead of estimateDisp(), there is no estimation of the common.dispersion (unless I am doing something wrong).

Is there a reason why the common dispersion is not computed?

Thank you in advance

The authors just didn't feel like putting it in, I guess. estimateGLMCommonDisp supports weights, so there's no reason to avoid calculation of the common dispersion. (It's probably not necessary to calculate it at every robustness iteration, where only the trended dispersion is used.) If you want it, you can just take the output of estimateGLMRobustDisp and feed it to estimateGLMCommonDisp; the weights will automatically be extracted and applied to compute the common dispersion.