I have a question regarding the interpretation of sample size when using the imputed MERFISH data in the ABC Atlas.
If one uses the imputed dataset for analysis and publication, how should the effective sample size (n) be considered?
Should it be treated as data derived from a single animal (since the MERFISH experiment itself was performed on one mouse), or is it more appropriate to consider the larger number of cells used for imputation across multiple animals (n=100 or more)?
I would greatly appreciate your guidance on how this is typically interpreted.
Additionally, are there any recommended best practices or guidelines for reporting results based on imputed MERFISH data in terms of statistical independence and sample size?
The imputed MERFISH values should be treated as data originating from a single animal. The imputation procedure uses information from multiple animals to computationally predict gene expression values for cells in the MERFISH dataset, but the underlying spatial observations all come from one mouse. The additional animals contribute statistical power to the imputation model itself, not independent biological replicates to your analysis.
As a result, the cells in the imputed dataset are not statistically independent from one another in the way that cells from separate animals would be. They share a common biological source, and any cell-to-cell correlation structure present in the original tissue is preserved — and in some respects amplified — through the imputation process.