Fusing graph transformer with multi-aggregate GCN for enhanced drug–disease associations prediction

BMC Bioinformatics

Table 2 Performance of WMAGT and other compared methods on three benchmark datasets

Datasets	NIMCGCN	DRWBNCF	Ghasemian’s model	VAGE	DRGBCN	WMAGT
AUROC
Fdataset	0.7428 ± 0.0276	0.8781 ± 0.0192	0.8902 ± 0.0328	0.9163 ± 0.1052	0.9326 ± 0.013	0.9353 ± 0.012
Cdataset	0.7928 ± 0.0248	0.8928 ± 0.0154	0.9114 ± 0.0292	0.9551 ± 0.0842	0.9454 ± 0.0091	0.9458 ± 0.0114
LRSSL	0.8661 ± 0.0165	0.8297 ± 0.0161	0.8791 ± 0.0359	0.8856 ± 0.0536	0.9437 ± 0.005	0.9434 ± 0.0083
Average	0.8006	0.8669	0.8936	0.9189	0.9405	0.9415
AUPR
Fdataset	0.0558 ± 0.0106	0.4638 ± 0.0548	0.4046 ± 0.0683	0.0589 ± 0.0429	0.4087 ± 0.0281	0.5231 ± 0.0487
Cdataset	0.0751 ± 0.0138	0.5801 ± 0.0332	0.4881 ± 0.1047	0.0608 ± 0.0355	0.4517 ± 0.0423	0.6 ± 0.0429
LRSSL	0.1807 ± 0.0204	0.4033 ± 0.0201	0.4925 ± 0.1166	0.0381 ± 0.0144	0.2558 ± 0.033	0.3651 ± 0.026
Average	0.1039	0.4824	0.4617	0.0526	0.3721	0.4961

ISSN: 1471-2105