I realized that you used MLE to infer the parameters for DESeq2 model. I am just curious whether SGD is suitable to infer these parameters.
We use IRLS which is related to Newton-Raphson method. I suppose you can use any optimization method you prefer. We also sometimes use L-BFGS-B for rows that don't converge in a maximum number of step.
thanks. Michael Love
Login before adding your answer.
Use of this site constitutes acceptance of our User Agreement and Privacy