Fig. B1.

Problems introduced by batch effects. This figure shows a toy example using simulated data that reflects how aggregating across measurements from different batches (e.g., measurements performed on different days) can ignore batch-to-batch variation leading to an overestimate of measurement precision. Each of the 3 dots represents replicate fitness measurements of a single genotype. In this example, batch effects are at least as strong as the effect of the drug on fitness. Since both drug and control conditions were included in each batch, the aggregate panel accurately captures the fitness difference between the drug and control, but overestimates the precision of this measurement. Had the experiment been designed differently, with drug and control conditions being surveyed on different days (dotted box), the effect of the drug on fitness may have been obscured by batch effects