[Response to Review] Update Figure 5 (Deep learning analysis)

[gwaybio]

One of the primary reviewer comments was wondering how the assays compared in their ability to predict pathways.

Here, I calculate average precision and precision-recall curves for the GO term models.

In addition, I also:

• Calculate AP and precision recall curves for MOA models
• Remove bortezomib from the test set predictions (it was biasing model eval)
• Update visualizations for Figure 5 by adding:
  • Precision recall curves for select high-performing models (in response to reviewers comments about accurate performance descriptions)
  • Add an explainer figure of our deep learning pipeline
• I also create two new supplementary figures (19 and 20), and remove old supplementary figure 19
  • These figures describe certain important aspects of the MOA and GO term predictions
• Add supplementary table 6 to describe average precision for all predictions

Updated Figure 5

Updated Supplementary Figure 19

New Supplementary Figure 20