Table 2 The details of the present KwLPR based QSAR/QSAAR models
Case study | Bandwidth method selection | Bandwidth | Local polynomial's degree | Kernel function | |||||
|---|---|---|---|---|---|---|---|---|---|
Case study 1 | Least Squares Cross-Validation Method | LogP | 0.410 | 0(Constant) | Gaussian | ||||
pEC50 (mM) (D. magna) | 0.213 | ||||||||
nT = 254; R2 = 0.85; RMSEC = 0.60; Q2LOO = 0.79; RMSECV = 0.70 | |||||||||
nV = 64; Q2F1 = 0.88; Q2F2 = 0.88; Q2F3 = 0.88; CCC = 0.93; RMSEP = 0.54 | |||||||||
Case study 2 | Expected Kullback–Leibler cross-validation Method | LogP | 0.416 | 0 (Constant) | Gaussian | ||||
pEC50 (D. magna) | 0.292 | ||||||||
nT = 235; R2 = 0.83; RMSEC = 0.66; Q2LOO = 0.79; RMSECV = 0.74 | |||||||||
nV = 59; Q2F1 = 0.91; Q2F2 = 0.91; Q2F3 = 0.91; CCC = 0.95; RMSEP = 0.48 | |||||||||
Case study 3 | Least Squares Cross-Validation Method | pEC50 (D. magna) | 0.357 | 0 (Constant) | Gaussian | ||||
GATS1e | 0.585 | ||||||||
nT = 35; R2 = 0.95; RMSEC = 0.34; Q2LOO = 0.88; RMSECV = 0.51 nV = 15; Q2F1 = 0.83; Q2F2 = 0.83; Q2F3 = 0.83; CCC = 0.90; RMSEP = 0.61 | |||||||||
Case study 4 | Direct Plug-in Method | pT (T. pyriformis) | 0.399 | 1 (Local linear) | Gaussian | ||||
nT = 31; R2 = 0.81; RMSEC = 0.28; Q2LOO = 0.72; RMSECV = 0.34 nV = 10; Q2F1 = 0.83; Q2F2 = 0.82; Q2F3 = 0.83; CCC = 0.91; RMSEP = 0.27 | |||||||||
Case study 5 | Least Squares Cross-Validation Method | MLOGP | 0.417 | 0 (Constant) | Gaussian | ||||
CIC0 | 0.584 | ||||||||
SM1_Dz(Z) | 0.512 | ||||||||
GATS1i | 0.535 | ||||||||
NdsCH | 0.781 | ||||||||
NdssC | 0.521 | ||||||||
nT = 726; R2 = 0.85; RMSEC = 0.57; Q2CV = 0.57; RMSECV = 0.93 nV = 182; Q2EXT = 0.68; RMSEEXT = 0.87; CCC = 0.79 | |||||||||