Fig. 3.

MP bound strikes a balance between overfitting and underfitting. The percent accuracy in identifying ligands from the verification set and rejecting ligands randomly selected from ChEMBL is shown as a function of the number of eigenvectors included in $\mathbf{V}$ for (A) AA2AR, (B) ADRB1, (C) ${\mathrm{\mu }}_1$ opioid receptor, and (D) cannabinoid CB1 receptor.