Recent Advances and Application of Machine Learning for Protein-Protein Interaction Prediction in Rice: Challenges and Future Perspectives.

Protein-protein interactions (PPIs) are significant in understanding the complex molecular processes of plant growth, disease resistance, and stress responses. Machine learning (ML) has recently emerged as a powerful tool that can predict and analyze PPIs, offering complementary insights into traditional experimental approaches. It also accounts for proteoforms, distinct molecular variants of proteins arising from alternative splicing, or genetic variations and modifications, which can significantly influence PPI dynamics and specificity in rice. This review presents a comprehensive summary of ML-based methods for PPI predictions in rice (Oryza sativa) based on recent developments in algorithmic innovation, feature extraction processes, and computational resources. We present applications of these models in the discovery of candidate genes, unknown protein annotations, identification of plant-pathogen interactions, and precision breeding. Case studies demonstrate the utility of ML-based methods in improving rice resistance to abiotic and biotic stresses. Additionally, this review highlights key challenges like data limits, model generalizability, and future directions like multi-omics, deep learning and artificial intelligence (AI). This review provides a roadmap for researchers aiming to use ML to generate predictive and mechanistic insights on rice PPI networks, hence helping to achieve enhanced crop improvement programs.