Developing and validating predictive decision tree models from mining chemical structural fingerprints and high–throughput screening data in PubChem

Table 2 Recognition rate of Decision Tree models

		Active compounds			Inactive compounds
Bioassay	PubChem Assay ID	TP	FN	Sensitivity	TN	FP	Specificity	Overall accuracy	MCC	Model Complexity (Number of Nodes/Number of Leaves/Number of Features)
5HT1a agonist	567	362	4	98.9%	64,394	146	99.8%	99.8%	0.84	(321/161/149)
5HT1a antagonist	612	360	56	86.5%	60,909	281	99.5%	99.5%	0.70	(1135/568/261)
HIV-1 RT RNase H inhibitor	565	1,128	122	90.2%	63,070	896	98.6%	98.4%	0.70	(3003/1502/412)
HIV-1 RT RNase H inhibitor	372	640	130	83.1%	98,463	535	99.5%	99.3%	0.67	(2511/1256/370)

TP = true positives, the number of correctly recognized active compounds;
FN = false negative, the number of active compounds that the model is unable to recognize;
TN = true negative, the number of inactive compounds that successfully recognized by the model;
FP = false positive, the number of inactive compounds that the model is unable to recognize.

ISSN: 1471-2105