The results of application of the machine learning methods to the initial ECFP6 fingerprints and to the 2D mapped space (multiclass classification).

Descriptor set	ML method	Accuracy
		GPCR ligands	NR ligands
ECFP6 descriptors	kNN	0.829	0.526
	SVM	0.821	0.549
	XGBoost	0.821	0.540
	Random forest	0.788	0.537
pTSNE mapping (2D space)	kNN	0.763	0.383
	SVM	0.704	0.336
	XGBoost	0.764	0.394
	Random forest	0.745	0.360
PCA mapping (2 components)	kNN	0.739	0.296
	SVM	0.735	0.345
	XGBoost	0.743	0.360
	Random forest	0.735	0.349
MDS mapping (2D space)	kNN	0.725	0.326
	SVM	0.543	0.250
	XGBoost	0.712	0.333
	Random forest	0.707	0.328