Table 3.

Comparison of various short molecular fingerprints

ECa	GPCR	MFPb
		MACCS	ECFP2	ECFP4	ECFP8	ECFP10	FCFP2	FCFP4	FCFP8	FCFP10	NGFP	WDL
RMSE (↓)	P08908	1.85*	1.88*	1.97*	2.30*	2.41*	1.77*	1.98*	2.22*	2.25*	1.56	1.50
	P50406	1.81*	1.90*	2.10*	2.54*	2.61*	1.83*	2.04*	2.39*	2.55*	1.60*	1.45
	P08912	2.17*	2.15	2.33*	2.92*	3.13*	2.11	2.16	2.41*	2.84*	2.30*	2.14
	P35348	2.20*	2.25*	2.58*	2.69*	3.01*	2.28*	2.06*	2.94*	2.77*	2.03*	1.72
	P21917	2.08*	2.25*	2.41*	2.91*	3.27*	2.21*	2.47*	2.92*	3.02*	1.99*	1.76
	Q9Y5N1	2.17*	2.46*	2.66*	3.23*	3.45*	2.17*	2.30*	3.03*	3.08*	1.62*	1.42
	P30968	1.33*	1.53*	1.57*	2.10*	2.25*	1.49*	1.59*	1.89*	2.02*	1.23	1.21
	P24530	1.32*	1.58*	1.73*	2.01*	2.16*	1.49*	1.67*	1.80*	1.92*	1.11*	1.04
	Q99705	1.82*	1.92*	2.10*	2.48*	2.58*	1.80*	1.90*	2.28*	2.44*	1.41*	1.33
	P35372	1.74*	1.87*	2.13*	2.47*	2.52*	1.80*	1.93*	2.24*	2.45*	1.53	1.51
	P46663	2.19*	2.29*	2.67*	3.42*	3.39*	2.49*	2.46*	2.93*	3.25*	2.33*	1.77
	P35346	1.53*	1.70*	2.00*	2.21*	2.21*	1.37*	1.57*	2.28*	2.01*	1.18*	1.13
	P21452	2.36*	2.50*	2.42*	2.75*	3.14*	2.46*	2.27*	2.67*	2.85*	2.07*	1.86
	P30542	1.14*	1.45*	1.79*	2.07*	2.21*	1.30*	1.49*	1.86*	1.96*	1.00	0.96
	Q99500	1.29*	1.78*	1.88*	2.23*	2.44*	1.67*	1.92*	2.02*	2.06*	1.33*	1.02
	Q9Y5Y4	1.41*	1.70*	1.74*	2.41*	2.54*	1.48*	1.78*	2.08*	2.35*	1.46*	1.28
	P34995	2.00*	2.12*	2.11*	2.76*	3.28*	3.10*	2.45*	2.59*	2.91*	2.06*	1.70
	P51677	1.57*	1.64*	2.00*	2.29*	2.52*	1.90*	1.75*	2.11*	2.51*	1.41*	1.15
	P48039	1.65	1.72	2.06*	2.38*	2.53*	1.60	1.79	2.25*	2.32*	1.84*	1.69
	Q8TDU6	0.99*	1.37*	1.25*	1.81*	2.00*	1.41*	1.13*	1.37*	1.88*	0.87	0.87
	Q8TDS4	1.37	1.45	2.61*	3.15*	3.00*	2.03*	2.12*	2.93*	2.94*	1.59	1.52
	Q9HC97	1.22*	1.31*	1.36*	1.47*	1.65*	1.41*	1.46*	1.22*	1.40*	1.39*	0.85
	P47871	1.17*	1.56*	1.71*	2.47*	2.51*	1.49*	1.54*	2.08*	2.25*	1.06	1.05
	P41180	1.08*	1.28*	1.34*	2.07*	2.16*	1.31*	1.10*	1.44*	1.90*	1.16*	0.76
	Q14416	1.08*	1.23*	1.42*	1.81*	1.91*	1.12*	1.42*	1.88*	1.70*	1.05*	0.81
	Q99835	1.25*	1.48*	1.44*	2.31*	2.12*	1.43*	1.08	1.61*	1.73*	1.11	1.01
r2 (↑)	P08908	0.58*	0.56*	0.53*	0.29*	0.22*	0.61*	0.51*	0.37*	0.33*	0.67	0.70
	P50406	0.68*	0.68*	0.60*	0.37*	0.33*	0.70*	0.64*	0.48*	0.38*	0.75	0.79
	P08912	0.46	0.49	0.40*	0.18*	0.14*	0.53	0.53	0.38*	0.20*	0.47	0.47
	P35348	0.61*	0.54*	0.43*	0.37*	0.29*	0.57*	0.59*	0.29*	0.38*	0.65*	0.70
	P21917	0.54*	0.44*	0.40*	0.21*	0.12*	0.51*	0.42*	0.25*	0.18*	0.59*	0.66
	Q9Y5N1	0.66*	0.55*	0.47*	0.28*	0.20*	0.64*	0.60*	0.36*	0.35*	0.80	0.83
	P30968	0.85	0.82	0.80*	0.60*	0.52*	0.82	0.80*	0.72*	0.64*	0.85	0.86
	P24530	0.80*	0.71*	0.62*	0.44*	0.31*	0.73*	0.69*	0.59*	0.48*	0.82	0.85
	Q99705	0.69*	0.69*	0.64*	0.40*	0.32*	0.69*	0.71*	0.53*	0.42*	0.80	0.82
	P35372	0.67*	0.62*	0.50*	0.29*	0.26*	0.64*	0.59*	0.43*	0.30*	0.73	0.74
	P46663	0.71*	0.66*	0.53*	0.27*	0.17*	0.60*	0.62*	0.45*	0.34*	0.66*	0.78
	P35346	0.74*	0.69*	0.58*	0.50*	0.48*	0.80*	0.73*	0.45*	0.56*	0.85	0.86
	P21452	0.53	0.44*	0.59*	0.31*	0.16*	0.48*	0.55*	0.41*	0.36*	0.66	0.70
	P30542	0.83	0.75*	0.58*	0.35*	0.23*	0.75*	0.71*	0.50*	0.43*	0.84	0.86
	Q99500	0.80*	0.52*	0.51*	0.45*	0.35*	0.60*	0.48*	0.47*	0.38*	0.82*	0.87
	Q9Y5Y4	0.84	0.74*	0.75*	0.52*	0.44*	0.81*	0.73*	0.60*	0.52*	0.82	0.86
	P34995	0.56*	0.52*	0.45*	0.26*	0.50*	0.54*	0.35*	0.27*	0.17*	0.54*	0.64
	P51677	0.76*	0.72*	0.58*	0.50*	0.43*	0.65*	0.69*	0.56*	0.40*	0.80*	0.86
	P48039	0.75	0.75	0.56*	0.53*	0.44*	0.74	0.67	0.52*	0.51*	0.74	0.72
	Q8TDU6	0.88	0.77*	0.82*	0.60*	0.51*	0.76*	0.85	0.77*	0.58*	0.91	0.92
	Q8TDS4	0.86	0.84	0.44*	0.22*	0.32*	0.65*	0.67	0.25*	0.21*	0.76	0.77
	Q9HC97	0.77*	0.72*	0.70*	0.64*	0.56*	0.70*	0.67*	0.75*	0.68*	0.72*	0.89
	P47871	0.87	0.78*	0.73*	0.47*	0.43*	0.80*	0.77*	0.58*	0.50*	0.89	0.90
	P41180	0.86	0.81*	0.77*	0.51*	0.47*	0.81*	0.86	0.76*	0.59*	0.85*	0.93
	Q14416	0.86	0.82*	0.78*	0.62*	0.60*	0.86	0.80*	0.66*	0.69*	0.88	0.92
	Q99835	0.85	0.79*	0.80*	0.51*	0.58*	0.81*	0.89	0.76*	0.71*	0.89	0.90

^aEvaluation Criterion: ↑ (↓) indicates the larger (smaller), the better the model performance; the best results on each evaluation criterion are highlighted in boldface.

^bMolecular Fingerprints: * indicates the performance of the compared short molecular fingerprint is significantly worse than that of WDL based on Wilcoxon signed-ranked test.