Bt Research 2024, Vol.15, No.5, 240-247 http://microbescipublisher.com/index.php/bt 242 detect Bt-induced hormesis in controlled settings, this phenomenon may not be observed under field conditions with current levels of Bt expression in crops (Rabelo et al., 2020). Another limitation is the difficulty in studying the complex interactions and evolutionary dynamics of insect resistance to Bt toxins. Traditional methods may not adequately capture the spatial and temporal heterogeneity of insect populations and their environments. This is particularly important in the context of genetically engineered Bt crops, where the evolution of insect resistance can be influenced by factors such as spatial heterogeneity in crop planting patterns (Huang et al., 2017). The experimental elucidation of membrane insertion steps in Bt toxins is challenging, necessitating the use of advanced computational tools like Alphafold-2 to predict structural intermediates. 3 AI Techniques for Predicting Bt Toxin-Insect Interactions 3.1 Machine learning algorithms in toxin prediction Machine learning (ML) algorithms have shown significant potential in predicting Bt toxin-insect interactions by analyzing complex multivariate data. For instance, ML tools have been effectively used to assess the risk of agrochemicals on non-target insects, such as bees, by modeling multiple factors affecting insect health. These algorithms can identify agrochemical contamination with high accuracy, demonstrating their capability in predicting interactions based on behavioral and physiological changes in insects (Bernardes et al., 2021). ML-based systems like ToxNet have been developed to predict toxins based on symptoms and metadata, outperforming traditional methods and even experienced medical professionals in accuracy (Zellner et al., 2022). 3.2 Deep learning for interaction modelling Deep learning (DL) approaches have further advanced the prediction of Bt toxin-insect interactions by enabling the modeling of complex biological networks. For example, the Data Integration with Deep Learning (DIDL) method predicts inter-omics interactions by automatically extracting features from biomolecules and predicting novel interactions. This method has shown high accuracy in various biological networks, making it a robust tool for understanding the underlying mechanisms of Bt toxin interactions with insect receptors (Borhani et al., 2021). Tools like Alphafold-2 have been used to predict the structural intermediates of Bt toxins during membrane insertion, providing insights into the molecular dynamics of toxin-receptor interactions (Campagne et al., 2016). 3.3 Integrating multi-omics data with AI for accurate predictions Integrating multi-omics data with AI techniques has proven to be a powerful approach for accurate predictions of Bt toxin-insect interactions. By combining genomic, transcriptomic, and proteomic data, AI models can provide a comprehensive understanding of the interactions at multiple biological levels. For instance, the CryProcessor tool utilizes a robust pipeline to mine 3d-Cry toxins from sequencing data, allowing for the precise prediction of domain layouts in toxin sequences. This integration of multi-omics data enhances the ability to screen for novel toxins and understand their interactions with insect receptors (Shikov et al., 2020). The continuous evolution of Bt toxins through AI-driven methods has shown promise in overcoming insect resistance by evolving toxins with novel receptor affinities, thereby maintaining their effectiveness (Badran et al., 2016). 4 AI in Predicting Resistance Development 4.1 Identifying genetic markers of resistance using AI Artificial intelligence (AI) has shown significant potential in identifying genetic markers associated with resistance to Bt toxins. For instance, genetic markers for resistance in Western corn rootworm (WCR) to the Cry3Bb1 Bt toxin have been discovered and validated, which can help in predicting resistance and improving insect resistance management strategies. Similarly, the identification of mutations in an ABC transporter gene linked to Cry2Ab resistance in Helicoverpa armigera highlights the role of AI in pinpointing specific genetic changes that confer resistance (Tay et al., 2015; Wang, 2024). These advancements underscore the importance of AI in enhancing our understanding of the genetic basis of resistance and in developing targeted management strategies.
RkJQdWJsaXNoZXIy MjQ4ODYzNA==