A relatively simple question. I wonder if it is possible to extract gene-wise variance covariance matrices of the GLM coefficients in edgeR. I would like to use it for non-standard multiple comparison procedures.
Thank you.
A relatively simple question. I wonder if it is possible to extract gene-wise variance covariance matrices of the GLM coefficients in edgeR. I would like to use it for non-standard multiple comparison procedures.
Thank you.
The simple answer is no, edgeR does not store the covariance matrices so they can't be extracted. Note that every gene will have its own covariance matrix.
You can fairly easily compute the covariance matrices yourself from the edgeR results if you know how. However it should be noted that covariance matrices are only relevant for Wald tests, and edgeR doesn't use Wald tests because likelihood ratio tests have much better properties.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
I also would like to use the per-gene covariance matrices. How do I calculate them from the edgeR results?
Thank you,
Lee