I am sorry I am a newbie in statistics and my question might sound very basic. Do the coefficients from dispersion estimation follow normal distribution? so we use parametric tests to calculate p values.
edgeR makes no assumption that any coefficient follows a normal distribution. Parametric tests are not limited to the normal distribution.
Nine weeks ago you seem to have done an analysis using limma-voom. Did that not work for yout?
Thank you, Gordon. Yes, It worked well for me. I am reading the statistical part of the analysis. I read some of your papers and I am trying my best to understand them to compare and contrast with other published libraries. I was curious about test statistics such as the moderated T test or Wald test in DESeq2. I had a gut feeling that maybe coefficients (might be random independent variables following a normal distribution, had Galton board imagination). So, I thought I should ask you if there might be some statistical dot I am missing. Thanks a lot for your time :-).
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
See the papers of the methods and previous questions, for example logfoldStandard Error and wald-statistics in DESeq2 results