Adding dose independent null distribution comparisons to original supplementary figure 4.
This required me to resample the null distribution without restricting that the random samples had to be measured at the same dose. I added this analysis for both Cell Painting and L1000, and I regenerated supplementary figure 4:
Modified original legend
Supplementary Figure S4. Percent strong scores with different input Cell Painting and L1000 data.
We calculated percent strong for profiling readouts across the data sets as described in Supplementary Figure S3. (a) Replicate and non-replicate distributions of median pairwise correlations of the five inputs as specified above. (b) Percent strong summary. The dose independent scores are comparing replicate median scores to a relaxed null distribution that did not require the random replicates to be randomly sampled from the same dose.
Adding dose independent null distribution comparisons to original supplementary figure 4.
This required me to resample the null distribution without restricting that the random samples had to be measured at the same dose. I added this analysis for both Cell Painting and L1000, and I regenerated supplementary figure 4:
Modified original legend
Supplementary Figure S4. Percent strong scores with different input Cell Painting and L1000 data. We calculated percent strong for profiling readouts across the data sets as described in Supplementary Figure S3. (a) Replicate and non-replicate distributions of median pairwise correlations of the five inputs as specified above. (b) Percent strong summary. The dose independent scores are comparing replicate median scores to a relaxed null distribution that did not require the random replicates to be randomly sampled from the same dose.