Table 1.
The statistics of the DTI datasets from [31].
Table 2.
The AUC and AUPR obtained under the setting CVS1.
Table 3.
The AUC and AUPR obtained under the setting CVS2.
Table 4.
The AUC and AUPR obtained under the setting CVS3.
Fig 1.
Performance trend of NRLMF on the benchmark datasets (a) Nuclear Receptor, (b) GPCR, (c) Ion Channel, and (d) Enzyme, measured by AUC with different settings of K1 under CVS1.
The best competitors on these datasets are (a) BLM-NII, (b) BLM-NII, (c) BLM-NII and CMF, and (d) BLM-NII, respectively.
Fig 2.
Performance trend of NRLMF on the benchmark datasets (a) Nuclear Receptor, (b) GPCR, (c) Ion Channel, and (d) Enzyme, measured by AUPR with different settings of K1 under CVS1.
The best competitors on these datasets are (a) BLM-NII, (b) CMF, (c) CMF, and (d) CMF, respectively.
Fig 3.
Performance trend of NRLMF on the benchmark datasets (a) Nuclear Receptor, (b) GPCR, (c) Ion Channel, and (d) Enzyme, measured by AUC with different settings of K2 under CVS1.
The best competitors on these datasets are (a) BLM-NII, (b) BLM-NII, (c) BLM-NII and CMF, and (d) BLM-NII, respectively.
Fig 4.
Performance trend of NRLMF on the benchmark datasets (a) Nuclear Receptor, (b) GPCR, (c) Ion Channel, and (d) Enzyme, measured by AUPR with different settings of K2 under CVS1.
The best competitors on these datasets are (a) BLM-NII, (b) CMF, (c) CMF, and (d) CMF, respectively.
Fig 5.
Performance trend of NRLMF on the benchmark datasets (a) Nuclear Receptor, (b) GPCR, (c) Ion Channel, and (d) Enzyme, measured by AUC with different settings of c under CVS1.
The best competitors on these datasets are (a) BLM-NII, (b) BLM-NII, (c) BLM-NII and CMF, and (d) BLM-NII, respectively.
Fig 6.
Performance trend of NRLMF on the benchmark datasets (a) Nuclear Receptor, (b) GPCR, (c) Ion Channel, and (d) Enzyme, measured by AUPR with different settings of c under CVS1.
The best competitors on these datasets are (a) BLM-NII, (b) CMF, (c) CMF, and (d) CMF, respectively.
Fig 7.
Performance trend of NRLMF on the benchmark datasets (a) Nuclear Receptor, (b) GPCR, (c) Ion Channel, and (d) Enzyme, measured by AUC with different settings of r under CVS1.
The best competitors on these datasets are (a) BLM-NII, (b) BLM-NII, (c) BLM-NII and CMF, and (d) BLM-NII, respectively.
Fig 8.
Performance trend of NRLMF on the benchmark datasets (a) Nuclear Receptor, (b) GPCR, (c) Ion Channel, and (d) Enzyme, measured by AUPR with different settings of r under CVS1.
The best competitors on these datasets are (a) BLM-NII, (b) CMF, (c) CMF, and (d) CMF, respectively.
Table 5.
The top 30 novel interactions predicted by NRLMF on GPCR dataset.
Table 6.
The fractions of true DTIs among the predicted top N (N = 10, 30, 50) interactions under CVS1.