Fig. 4: Parameter optimization for constant budget.
From: scPower accelerates and optimizes the design of multi-sample single cell transcriptomic studies
Maximizing detection power by selecting the best combination of cells per individual and read depth for a DE study with a budget of 10,000€ (a) and an eQTL study with a budget of 30,000€ (b). Sample size is uniquely defined given the other two parameters due to the budget restriction and visualized using the point size. c–f Overall detection power dependent on cost determining factors. Influence of the cells per individual given the optimized read depth (c, e) and of the read depth given the optimized number of cells per individual (d, f). Corresponds to the DE study in (a), visualized in (a) by the red frame around the row with the optimal number of cells (corresponding to (c)) and the red frame around the column with the optimal read depth (corresponding to (d)). Same frames for (e, f) in the eQTL study (b). The optimal sample size values are shown in the upper x axes for (c–f). Vertical line in the subplots marks the optimal parameter combination. Effect sizes were chosen as in Fig. 3. Gene expression is defined as detected in >50% (DE analysis) or >9.5% (eQTL analysis) of individuals.
