Figure 8
From: Predicting cytotoxicity from heterogeneous data sources with Bayesian learning
Prediction network of percent inhibition models. Nodes represent assays with arbitrary assay number. Node size is proportional to number of molecules in assay. The presence of edges between two nodes indicates that a model from one set is predictive for the other and vice versa. All data sets are Pfizer assays except for 33 (Scripps) and 34 (NCGC). Assays with fewer than 10 actives were removed.