Hi,

I have a time series experiment that has three time points and two conditions. I am using the LRT test to make the initial DESeq object and then doing the specific pair-wise comparisons using the wald test.

I also want to extract the normalized counts for all the samples. To do this do I use the DESeq object from the LRT test or its better to make another DESeq object using the wald test and use that to extract the normalized counts. Basically my question is if there is a difference in the normalized counts obtained from the Wald test and from the LRT test and if yes then which one is more accurate?

Thank you

If the same samples are present in the object then the normalized counts are exactly the same independent of the design or test you use.

The reason for this is that the normalization always runs via estimateSizeFactors() and this is agnostic of the design and test, see ?estimateSizeFactors().