Nuclear transfer techniques (a.k.a. mitochondrial replacement therapies) are currently under development to provide a route to eliminating particular instances of mitochondrial disease from the germline. Before these kinds of techniques are implemented clinically it is of primary concern that their safety and efficacy is established. In a recent paper, Hyslop et al. (Nature 534:383-386, 2016. doi: 10.1038/nature18303 ) utilized a specific version of pronuclear transfer to investigate the consequences for gene expression in the developing embryo, which may indicate whether or not developmental pathways have been perturbed. However, the study was only able to include a small number of blastocysts within each treatment group, although a larger number of single cell expression profiles from each blastocyst were acquired. Using simulated datasets we show that the size and experimental design of this study cannot provide conclusive evidence that expression profiles of manipulated or control samples are indistinguishable from one another due to low power. These simulations also illustrate why visual inspections of principle component analyses used in the study cannot replace statistical modeling of treatment effects.