Fig. 4

Loaded data table of PIDGIN results. The results table includes a row for each target defined by their HGNC symbol (hyperlinked to the UniProt database entry) and preferred names, the predicted probability of activity of the query compound lapatinib, the most structurally similar compound (nearest neighbour—NN) in the model training set (hyperlinked to the ChEMBL database entry), the Tanimoto similarity of the NN compared to lapatinib (where 1 indicates the compounds are exactly the same), and the experimental pChEMBL value of the NN against the target. Here the top three predicted targets (ERBB4, EGFR and ERBB2/HER2), are selected for further analysis to recapitulate lapatinib’s MoA in HER2 + cells