DeepSA: a deep-learning driven predictor of compound synthesis accessibility

Table 3 Prediction results for the different types of molecules

Compound name	DeepSA	GASA	SAscore	RAscore	SYBA	SCScore	Synthesis steps	Refs.
Cularine	0.999	0.458	2.933	0.939	−0.556	3.706	16	[31]
Goniomitine	0.999	0.165	3.831	0.314	−16.213	4.376	12	[32]
Fusaequisin A	0.976	0.008	5.066	0.023	−51.766	3.925	20	[33]
Haliclonin A	0.960	0.002	6.513	0.131	−97.488	3.982	38	[34]
Hyacinthacine A1	0.959	0.108	3.847	0.638	−17.617	3.446	13	[35]
Hydroxyancepsenolide	0.744	0.029	3.892	0.018	−33.231	4.000	13	[36]
Kirkamide	0.630	0.475	4.142	0.264	−16.188	3.161	11	[37]
Longianone	0.727	0.190	4.666	0.668	−45.744	2.857	14	[38]
Simpotentin	0.522	0.075	4.560	0.201	25.489	3.643	14	[39]
Halomon	0.403	0.688	4.847	0.302	−6.849	2.588	10	[40]
Dihydropinidine	0.340	0.455	3.275	0.927	−29.819	3.102	9	[41]
Scorodonin	0.306	0.452	4.831	0.907	−26.430	2.160	6	[42]
Pinnatolide	0.294	0.718	3.534	0.547	−6.103	2.498	7	[43]
Tanikolide	0.289	0.371	3.072	0.797	2.726	3.218	7	[44]
Sedridine	0.128	0.924	3.324	0.990	−23.154	2.987	7	[45]
Isolaurepan	0.117	0.958	2.790	0.393	22.754	3.609	4	[46]
cis-Perhydroazulene	0.086	0.724	2.532	0.965	10.031	2.746	8	[47]
Gabapentin	0.071	0.770	2.400	0.973	2.213	2.161	3	[48]

ISSN: 1758-2946