Table 3.

Performances with angles using two DeepSnap systems of the four MIE targets.

PubChem		720725_GR_Ant		1347030_TRHR_Ago		1347032_TGF_Beta_Python
Assay AID		Python	DIGITS	Python	DIGITS	Python	DIGITS
ROC_AUC	average	0.832 ± 0.048	0.856 ± 0.029	0.875 ± 0.031	0.886 ± 0.028	0.879 ± 0.015	0.907 ± 0.020
	max_ROC_AUC	0.926	0.910	0.915	0.918	0.911	0.932
	max_angle	185	95	176	185	185	75
	max_split	7:1:2	5:5:1	3:1:2	5:3:4	7:1:2	5:3:2
BAC	average	0.762 ± 0.044	0.791 ± 0.023	0.811 ± 0.032	0.829 ± 0.023	0.805 ± 0.015	0.849 ± 0.030
	max_BAC	0.864	0.837	0.868	0.876	0.844	0.930
	max_angle	185	95	176	355	185	176
	max_split	7:1:2	3:3:1	3:1:2	5:3:4	7:1:2	5:3:2
MCC	average	0.248 ± 0.065	0.282 ± 0.030	0.141 ± 0.017	0.155 ± 0.022	0.309 ± 0.025	0.384 ± 0.044
	max_MCC	0.451	0.354	0.194	0.208	0.373	0.478
	max_angle	176	75	176	355	176	165
	max_split	7:1:2	4:4:1	3:1:2	1:1:1	7:1:2	2:2:1
Acc	average	0.790 ± 0.058	0.812 ± 0.044	0.781 ± 0.030	0.769 ± 0.060	0.770 ± 0.029	0.833 ± 0.033
	max_Acc	0.917	0.939	0.856	0.902	0.834	0.896
	max_angle	176	155	176	125	176	165
	max_split	7:1:2	4:4:1	3:1:2	1:1:1	7:1:2	2:2:1
loss_val	average	0.383 ± 0.115	0.108 ± 0.014	0.189 ± 0.070	0.032 ± 0.007	0.316 ± 0.061	0.113 ± 0.011
	min_loss_train	0.110	0.065	0.194	0.024	0.197	0.087
	min_angle	185	195	176	325	350	230
	max_split	7:1:2	7:1:2	3:1:2	3:1:2	3:2:1	7:1:2
loss_train	average	0.413 ± 0.153		0.247 ± 0.088		0.280 ± 0.120
	min_loss_train	0.038		0.047		0.044
	min_angle	176		176		176
	max_split	2:2:1		1:1:1		1:1:1
PR_AUC	average	0.335 ± 0.117		0.103 ± 0.041		0.315 ± 0.056
	max_PR_AUC	0.660		0.194		0.453
	max_angle	176		176		176
	max_split	7:1:2		3:1:2		3:1:1
F	average	0.853 ± 0.039		0.868 ± 0.020		0.833 ± 0.020
	max_F	0.935		0.914		0.876
	max_angle	176		176		176
	max_split	7:1:2		3:1:2		7:1:2