Integrating synthetic accessibility with AI-based generative drug design

Table 9 Some statistics about the molecules in the blueprint for the 6 generations

Synth constraint	Count	Average RScore3min	Feasible	Good RScore3min	Standard Murckos	Generic Murckos	Unique
None	5005	0.08	1959	282 (6%)	59 (21%)	36 (13%)	34
RA	3574	0.11	2660	360 (10%)	79 (22%)	47 (13%)	64
SC	211	0.35	202	127 (60%)	19 (15%)	14 (11%)	64
SA	311	0.56	311	286 (92%)	40 (14%)	31 (11%)	145
RScore1min	850	0.49	843	706 (83%)	69 (10%)	46 (7%)	314
RSPred	985	0.46	971	655 (66%)	104 (16%)	73 (11%)	357

Standard Murckos and Generic Murckos: the number of different Murcko. Feasible/good RScore: number of molecules with RScore \(>0\) / \(\ge 0.5\). Unique: # of molecules that are only in this generation (and not in any of the other five). All the columns after ’good RScore3min’ refer to the molecules in the blueprint with a good RScore3min

ISSN: 1758-2946