Small Sample Size Question

[davebridges]

Quynh, can you please address this comment

the statistical analysis and models used should be discussed here to explain how the limitation of small number of subjects was overcome, at least in part.

[qtran1]

Another limitation in our study is the small sample size, especially the number of biological replicates in Cushing’s group (n=5). Adding a covariate such as BMI or age in the model further reduces the sample size to 2 or 3 replicates. Although this sample size is small, it is common and reasonable for a controlled high through put sequencing experiment, especially of a rare disease such as Cushing’s. Realizing our limitation, we chose DESeq2 as the statistical method for our RNA-seq data. DESeq2 overcomes the small sample size problem by pooling information across genes. Maximum likelihood estimation is applied to estimate the dispersion or variance of a gene across all replicates in a group. Then, an empirical Bayes approach is used to get maximum a posterior as the final dispersion estimate. This method utilizes the available data to the maximum extent; therefore, help avoiding potential false positives (DESeq2 REF).