I am looking to develop transcriptional signatures that are associated with sensitivity to therapeutics in a screen of patient tumor derived organoids. I am trying to figure out how to estimate the power for a biostatistical plan. We will be screening 100 organoids and performing whole transcriptome RNAseq.

I would like to use the methodology as described in Tiric, et al. Cancer Discovery. 2018. For each therapeutic they tested, for each individual gene, they plotted each organoids gene expression count vs therapeutic specific AUC to determine the spearman correlation. They then filtered by P value and r2 of each gene to develop the therapeutic specific gene signature.

The only power calculators I have seen for gene signatures involve comparing two groups with different characteristics. For example, I could estimate for the third most sensitive vs third most resistant, but this does not reflect the analysis I would like to perform.