Skip to main content

Table 2 Performance of selectivity classification models

From: Quantitative prediction of selectivity between the A1 and A2A adenosine receptors

Classification modelClassMCCSensitivitySpecificityPPVNPVROC
QSAR 2-classA1AR0.40 ± 0.130.62 ± 0.160.76 ± 0.110.57 ± 0.120.86 ± 0.070.87 ± 0.06
A2AAR0.40 ± 0.130.76 ± 0.110.62 ± 0.160.86 ± 0.070.57 ± 0.120.87 ± 0.06
QSAR 3-classA1AR0.22 ± 0.150.25 ± 0.150.96 ± 0.020.31 ± 0.140.93 ± 0.020.76 ± 0.06
A2AAR0.20 ± 0.170.33 ± 0.160.88 ± 0.020.35 ± 0.140.83 ± 0.040.62 ± 0.09
Dual0.10 ± 0.090.81 ± 0.030.29 ± 0.100.75 ± 0.020.35 ± 0.080.58 ± 0.06
QSAR 3-classA1AR0.00 ± 0.070.11 ± 0.050.88 ± 0.070.10 ± 0.060.91 ± 0.020.64 ± 0.09
A2AAR0.36 ± 0.120.47 ± 0.140.85 ± 0.100.59 ± 0.110.85 ± 0.020.65 ± 0.09
Non-binder0.07 ± 0.130.67 ± 0.110.39 ± 0.120.70 ± 0.050.37 ± 0.120.50 ± 0.09
QSAR 4-classA1AR0.11 ± 0.100.12 ± 0.090.97 ± 0.010.18 ± 0.110.95 ± 0.010.70 ± 0.07
A2AAR0.25 ± 0.160.29 ± 0.130.94 ± 0.020.39 ± 0.160.90 ± 0.020.67 ± 0.09
Dual0.09 ± 0.050.51 ± 0.050.58 ± 0.080.50 ± 0.050.60 ± 0.020.58 ± 0.05
Non-binder0.15 ± 0.060.51 ± 0.090.64 ± 0.060.47 ± 0.040.68 ± 0.050.57 ± 0.07
  1. Means of fivefold cross-validation with standard error of the mean (SEM). The class indicates the performance for that particular selectivity class: A1AR-selective, A2AAR-selective, dual (non-selective), and non-binders
  2. MCC Matthews Correlation Coefficient, PPV positive predictive value, NPV negative predictive value, ROC receiver operating characteristic