Table 3 Transductive model evaluation based on training dataset with negative samples generated by DDB. The three values divided by the two vertical bars relatively are AUROC, Spearman coefficient correlation (ρ) between the positive sample degree and sample predicted interactive probability, and ρ between the negative sample degree and sample predicted interactive probability. Results represent mean of n = 15 independent runs

From: Negative sampling strategies impact the prediction of scale-free biomolecular network interactions with machine learning

		Noise-RF	Seq-RF	Seq-Deep
LPI	NPInterv4.0	0.548\|− 0.046\|0.228	0.624\|− 0.048\|0.295	0.862\|− 0.145\|0.102
LPI	RAID v2.0	0.764\|− 0.284\|0.634	0.770\|− 0.286\|0.603	0.617\|− 0.072\|− 0.116
PPI	InBioMap	0.828\|0.077\|0.420	0.858\|0.144\|0.265	0.924\|− 0.026\|− 0.285
	STRING	0.693\|0.141\|0.542	0.774\|0.306\|0.259	0.889\|0.265\|− 0.227
	BioGRID	0.689\|0.171\|0.329	0.743\|0.250\|0.246	0.767\|0.122\|− 0.108
	HuRI	0.759\|0.113\|0.261	0.797\|0.161\|0.240	0.782\|0.110\|− 0.112
DTI	DrugBank	0.494\|0.023\|0.229	0.825\|0.292\|0.117	0.821\|0.243\|− 0.134
DTI	DrugCentral	0.622\|0.300\|0.491	0.852\|0.354\|0.309	0.859\|0.206\|− 0.139

Back to article page

ISSN: 1741-7007

Contact us

General enquiries: journalsubmissions@springernature.com