Deseq and EdgeR developers and user hi,
I'm testing an RNA-Seq data set for few developmental time-points (t0,t4,t12,t18,t21,t24,t48) with no replicates (more details here: https://support.bioconductor.org/p/75773/#75851). All samples come from the same embryos culture, and can be considered as samples taken from a single individual at different times. Biologically, we know that from time "t0" to "t4" there is no transcription, and existing mRNAs are degraded. Therefore I would expect that as compared to "t0", most genes will be under-expressed in "t4". Yet, as far as I read, EdgeR and Deseq2 default normalization of read-counts expect symmetrical under/over expression.
attached a diagnostic histogram based on the Deseq developers suggestions ( Normalization by DEseq ), where the log-fold changes of t4 is based on the geometric mean of all samples. I am suspecting, though cannot know for sure, that the true scaling factor is different (e.g., possibly red line should be shifted even more to the right, since I expect only degradation in t4). Is there something I can do here to improve the scaling factor estimation ?