Fig. 3: Mechanistic relevance of KDDANet prediction results.
From: Genome-wide discovery of hidden genes mediating known drug-disease association using KDDANet
a KDDANet resulting subnetwork mediating phylloquinone (DB01022)-Alzheimer’s disease (AD, 104300) association. b Top ten enriched KEGG terms of subnetwork genes mediating phylloquinone-AD association. c Expression level of STAT3, GNAI1, and GC in normal individuals and AD patients, **p value < 0.01, calculated by Mann–Whitney U test. d KDDANet resulting subnetwork for heparin (DB01109)-obesity (601665) association. e Top ten enriched KEGG terms of subnetwork genes mediating heparin-obesity association. f ROC and PR curves of KDDANet gene subnetwork mediating heparin-obesity association. In the subnetworks, the size of a gene node was proportional to its network degree; The thickness of a network edge was proportional to its flow amount. FPKM Fragments Per Kilobase Of Exon Per Million Fragments Mapped.
