Learning self-supervised molecular representations for drug–drug interaction prediction

BMC Bioinformatics

Table 4 Performance of SMR-DDI and other featurizers on Task 2

Featurizer	AUPRC	AUROC	ACC	F1	Precision	Recall
gin_supervised_contextpred	0.503 ± 0.039	0.908 ± 0.023	0.596 ± 0.005	0.583 ± 0.007	0.599 ± 0.01	0.596 ± 0.005
ECFP	0.502 ± 0.047	0.897 ± 0.036	0.601 ± 0.016	0.581 ± 0.022	0.616 ± 0.017	0.601 ± 0.016
gin_supervised_masking	0.494 ± 0.051	0.901 ± 0.019	0.594 ± 0.008	0.579 ± 0.009	0.602 ± 0.006	0.594 ± 0.008
gin_supervised_infomax	0.472 ± 0.065	0.884 ± 0.031	0.589 ± 0.004	0.576 ± 0.007	0.593 ± 0.01	0.589 ± 0.004
gin_supervised_edgepred	0.48 ± 0.036	0.905 ± 0.024	0.585 ± 0.01	0.57 ± 0.012	0.593 ± 0.013	0.585 ± 0.013
MOL2VEC	0.554 ± 0.022	0.913 ± 0.006	0.575 ± 0.024	0.565 ± 0.022	0.573 ± 0.023	0.575 ± 0.024
ChemBERTa-77M-MLM	0.491 ± 0.021	0.9 ± 0.02	0.58 ± 0.013	0.564 ± 0.017	0.585 ± 0.013	0.58 ± 0.013
ChemBERTa-77M-MLR	0.533 ± 0.016	0.911 ± 0.009	0.571 ± 0.014	0.562 ± 0.01	0.575 ± 0.013	0.571 ± 0.014
MACCKEYS	0.482 ± 0.028	0.891 ± 0.014	0.543 ± 0.015	0.528 ± 0.014	0.545 ± 0.014	0.543 ± 0.015
SMR-DDI	0.434 ± 0.025	0.896 ± 0.008	0.528 ± 0.005	0.51 ± 0.006	0.529 ± 0.007	0.528 ± 0.005
ChemGPT-4	0.447 ± 0.035	0.899 ± 0.01	0.503 ± 0.014	0.486 ± 0.018	0.499 ± 0.021	0.503 ± 0.014
ChemGPT-1	0.452 ± 0.039	0.898 ± 0.015	0.493 ± 0.01	0.476 ± 0.011	0.488 ± 0.011	0.493 ± 0.01

ISSN: 1471-2105