Machine learning methods for metabolic pathway prediction

BMC Bioinformatics

Table 4 Best-performing discretized numeric features, ordered by information gain

Feature	ACC	SN	SP	FM	PR	RC	IG
enzyme-info-content-norm	0.824	0.912	0.801	0.685	0.548	0.912	0.19
enzymes-per-reaction	0.822	0.914	0.798	0.683	0.545	0.914	0.189
fraction-reactions-with-enzymes	0.824	0.91	0.801	0.684	0.548	0.91	0.188
num-reactions-with-enzymes	0.821	0.914	0.796	0.681	0.543	0.914	0.188
num-enzymes	0.821	0.914	0.796	0.681	0.543	0.914	0.188
enzyme-info-content-unnorm	0.821	0.914	0.796	0.681	0.543	0.914	0.188
evidence-info-content-norm-all	0.821	0.893	0.802	0.677	0.545	0.893	0.179
best-fraction-reactions-present-in-linear-path	0.842	0.85	0.84	0.693	0.584	0.85	0.179
fraction-reactions-present	0.83	0.869	0.82	0.682	0.562	0.869	0.176
fraction-reactions-present-or-orphaned	0.852	0.817	0.861	0.698	0.609	0.817	0.176

See section "Feature Extraction and Processing" and Section 1 of Additional table 2 for description of features. See Table 2 for explanation of column headings.

ISSN: 1471-2105