AI for plant protein-protein interactions prediction

Area of Science:

• Plant biology
• Computational biology
• Bioinformatics

Background:

• Protein-protein interactions (PPIs) are crucial for plant cell functions, including metabolon assembly, signal transduction, and organelle communication.
• Traditional experimental and computational methods for PPI prediction have limitations in comprehensively mapping plant interactomes.

Purpose of the Study:

• To systematically review and categorize current and emerging methodologies for predicting plant protein-protein interactions.
• To highlight the potential of artificial intelligence (AI) in advancing plant interactome mapping.

Main Methods:

• Categorization of PPI prediction methods into four families: sequence-centric, structure-based, network-level, and geometric/generative.
• Review of AI-driven approaches, including protein language models, coevolutionary signals, graph neural networks, and symmetry-aware networks.

Main Results:

• AI offers unprecedented opportunities for comprehensive plant interactome mapping.
• Current AI methods are categorized into sequence, structure, network, and geometric/generative approaches.

Conclusions:

• Plant AI PPI models face challenges like paralog expansion and data scarcity.
• Future models should be organelle-aware, multimodal, rigorously benchmarked, structure-gated, and condition-validated for improved accuracy and applicability.