How does the presence of DEG (affecting total count and proportionally the Ratio) impact the potential for error (type 1 or 2) within the analysis of ratio based on the vignette? If so can one mitigate this incidence ?
The total count is conditioned on when using the beta binomial, it is taken as fixed. It will affect power but not bias the ratio or coefficients.
Am I correct in thinking it would diminish the power of the analysis for those genes (exhibiting DEG) ? Also, I am a bit confused about the meaning of biasing the coefficient in this context ?
If one would take the same would analyze the Allelic Expression shift between condition that result in DEG, it would diminish the ability to correctly reject the null hypothesis of genes with DEG and nothing else ?
Thank you very much.
Lots of things are possible but generally low total counts will lead to higher sampling variance on the allelic ratio, so lower power.
To the extent that DE may lead to lower counts in one group it may affect power.
Login before adding your answer.
Use of this site constitutes acceptance of our User Agreement and Privacy