Table 9.

Summary of Contributions in Studies on Transfer Learning for Protein–Protein Interactions. Note that each study employed varied datasets, cross-validation methods, and simulation settings for evaluation, making direct comparisons potentially inconclusive. The highest reported accuracy is presented when models were assessed using multiple datasets.

Author	Metrics and Results	Contributions
Chen et al. [109]	ACC: 0.9745	Developed TNNM, a model for predicting essential proteins with superior performance on two public databases.
Derry and Altman [110]	AUC: 0.881	Proposed COLLAPSE, a framework for identifying protein structural sites, demonstrating excellent performance in various tasks including PPIs.
Si and Yan [111]	AvgPR: 0.576	Presented DRN-1D2D_Inter, a deep learning method for inter-protein contact prediction with enriched input features.
Yang et al. [112]	ACC: 0.9865 F1: 0.9236 AUPRC: 0.974	Utilized a Siamese CNN and a multi-layer perceptron for human-virus PPI prediction, applying transfer learning for human-SARS-CoV-2 PPIs.
Zhang et al. [113]	AvgPR: 0.6596	Introduced HDIContact, a deep learning framework for inter-protein residue contact prediction, showcasing promising results for understanding PPI mechanisms.