Why Pesticide Repurposing?

Agriculture is fundamental to national economies and global stability, directly underpinning food security and societal well-being. Major cereal crops such as rice, wheat, and maize constitute the cornerstone of human nutrition, yet they remain highly vulnerable to a spectrum of biotic stresses, including fungal, bacterial, and viral pathogens, which can lead to yield reductions of 20–40% under epidemic conditions. Traditional chemical control, while effective, faces two convergent crises: first, the development of novel active ingredients is prohibitively costly (exceeding $280 million per compound) and time-consuming (often over a decade); second, the over-reliance on a limited set of mode-of-action classes has accelerated the evolution of pathogen resistance, rendering once-effective pesticides obsolete. In this context, pesticide repurposing—identifying new uses for existing registered compounds—emerges as a strategic imperative to rapidly expand the utility of current chemistries, bridge protection gaps, and sustain crop productivity without the delay and expense of de novo discovery. By leveraging historical efficacy data and computational intelligence, repurposing offers a timely, economically viable, and resistance-aware pathway to reinforce integrated pest management and safeguard global food systems.

Just as in human medicine, where doctors rely on comprehensive analysis of various patient examination data to make precise diagnoses, agricultural production also requires such precision. Agricultural experts assess potential diseases affecting crops based on symptoms such as color, shape, and texture. Timely disease diagnosis helps farmers implement effective prevention and control measures, ensuring healthy crop growth and safeguarding the quality and yield of agricultural products. As shown in Figure 1.

Highlights the primary challenges in pesticide innovation: the dilemma of high-cost new discovery, the massive unexplored potential of existing stocks, and the complexity of ternary pesticide-crop-disease relationships. (b) Outlines the proposed workflow design to address these challenges, which progresses from Event-Centric Modeling and Prototype Refinement to the final Link Prediction for scoring repurposing candidates.

Challenges in Pesticide Repurposing

Current computational approaches to pesticide repurposing primarily rely on historical registration data and predictive modeling to identify new uses for existing agents. However, several critical challenges hinder their effective application in agriculture. First, most models are built on binary relationships (e.g., pesticide–disease), while effective crop protection inherently depends on the pesticide–crop–disease triad. Ignoring the host crop often leads to recommendations that are theoretically sound but agronomically invalid. Second, registration data exhibit a severe long-tail distribution, with abundant records for major crops but sparse information for specialty crops and emerging pathogens. This imbalance limits the ability of conventional graph models to generalize to rare but high-value repurposing scenarios. Furthermore, existing methods lack sufficient interpretability. Agricultural decisions require understanding the rationale behind recommendations—such as mode of action and environmental suitability—yet most models remain opaque “black boxes,” undermining trust and usability in field applications.

why PE-HGT?

To address this, we propose the Prototype-Enhanced Heterogeneous Graph Transformer (PE-HGT), a novel and interpretable framework for AI-driven pesticide repurposing. PE-HGT constructs an event-centered knowledge graph that explicitly models the complete pesticide-crop-disease triad, preserving essential agronomic context. To mitigate extreme data sparsity and long-tail distributions inherent in agrochemical records, a dedicated prototype refinement module learns latent, semantically coherent prototypes and enhances representations of sparse entities through alignment with these robust clusters.

Research Team

SAMLab