cidalsDB: an AI-empowered platform for anti-pathogen therapeutics research

Table 2 Variation of the FPs composition in zero values with respect to the vector size (in bits) and the subsequent performances of RF in molecule classification based on the different FPs versus the molecular descriptors

Fingerprints	Vector size	% of Zeros	Metrics
Fingerprints	Vector size	% of Zeros	Precision	Recall	F1-score
RDK	256	5	0.565	0.121	0.200
	512	18	0.622	0.205	0.309
	1024	39	0.650	0.238	0.349
	2048	61	0.643	0.280	0.390
Atom Pair	256	43	0.705	0.0586	0.101
	512	63	0.700	0.080	0.143
	1024	79	0.810	0.114	0.200
	2048	89	0.770	0.128	0.210
Topological Torsion	256	84	0.692	0.128	0.216
	512	92	0.696	0.154	0.252
	1024	96	0.723	0.158	0.259
	2048	98	0.698	0.168	0.271
Morgan	256	72	0.811	0.048	0.091
	512	85	0.807	0.083	0.152
	1024	92	0.758	0.109	0.191
	2048	96	0.766	0.147	0.257
Molecular descriptors	208	–	0.609	0.170	0.274

ISSN: 1758-2946