Building mycotoxin-resilient supply chains with AI
Mycotoxins have long been a pervasive challenge in agriculture and food production, threatening both animal and human health. As naturally occurring toxins produced by certain moulds, mycotoxins can contaminate crops under favourable environmental conditions, leading to significant economic losses and health risks.
MYCOTOXINS 2025: Utilising technology to detect & mitigate – read all articles
Decades of expertise in mitigating mycotoxin risks reside within dsm-firmenich, where innovative tools such as the Mycotoxin Survey and Prediction Tool have been integrated and further developed to enhance risk management strategies.
The Mycotoxin Survey, one of the most comprehensive databases of mycotoxin contamination worldwide, provides invaluable insights into regional and temporal trends of mycotoxin occurrence. Accompanying this, the Mycotoxin Prediction Tool leverages historical data, weather patterns, and other critical factors to forecast mycotoxin contamination risk with high levels of accuracy.
Artificial intelligence driven statistics
With the advent of artificial intelligence, these efforts have reached new heights. Artificial intelligence -driven analytics provide deeper insights into mycotoxin development, helping to refine prediction models with real-time environmental data. This allows for more proactive and precise interventions, ultimately safeguarding feed and food supplies from mycotoxin contamination.
Beyond prediction, artificial intelligence is also accelerating the development of novel solutions for mycotoxin detoxification.
Enzyme engineering i.e. relies on slow, manual approaches where scientists made incremental mutations to improve enzyme performance. This method, while effective, was costly and inefficient due to the vast search space of possible enzyme sequences. The company is now leveraging machine learning (ML) to streamline enzyme development—generating, predicting, and optimising sequences far more efficiently while ensuring structural and functional integrity.
One of the most significant challenges in enzyme design is navigating the immense sequence space. The number of potential enzyme variations surpasses the number of atoms in the universe, making exhaustive search methods impractical.
AlphaFold2
A game-changing breakthrough in this field is AlphaFold2—a Nobel Prize-winning AI technology that accurately predicts protein structures. Prior to AlphaFold2, structural biology relied heavily on experimental methods like X-ray crystallography, which are expensive, time-consuming, and sometimes infeasible. With AlphaFold2, structure-function analysis can now be conducted within minutes instead of months or years. This capability has been instrumental in the development of mycotoxin-neutralising enzymes and broader enzyme applications in health and nutrition.
Protein language models (pLMs), inspired by natural language processing, are another frontier in AI-driven enzyme engineering. By recognising evolutionary relationships and functional constraints, pLMs help optimise enzyme sequences for enhanced stability, solubility, and catalytic efficiency. dsm-firmenich integrates these artificial intelligence -generated insights with wet-lab validation, ensuring that computationally designed enzymes undergo rigorous experimental testing to confirm their real-world effectiveness.
Significant findings have resulted from the companies internal studies and collaborations with global research institutions. Artificial intelligence models have highlighted how specific regions are disproportionately affected by certain mycotoxins, enabling tailored risk management strategies. The impact of climate change on accelerating mycotoxin prevalence highlights the need for preparedness of those through the value chain. Additionally, advanced analytics have illuminated interactions between different mycotoxins, providing a clearer understanding of their combined toxic effects.
Commitment to quality and compliance
But how does it all work? The company’s products are designed with a robust Mode of Action (MoA) to tackle mycotoxin contamination at multiple levels. Solutions include binders that physically sequester mycotoxins, enzymatic treatments that degrade toxic compounds (e.g.: ZENzyme, and bioprotective agents that inhibit mould growth. These products have been rigorously tested and proven to reduce mycotoxin exposure effectively.
Furthermore, there is a strong commitment to quality and compliance which is evidenced by both EU and FDA authorisations. Both underscore the safety and efficacy of the entire product line. These certifications provide assurance to customers, reinforcing dsm-firmenich’s position as a leader in mycotoxin risk management. Customers worldwide rely on these solutions to safeguard operations, mitigate risks, and ensure the safety of their products. From feed manufacturers to livestock producers, the company’s comprehensive and innovative AI approach continues to deliver measurable value across the supply chain.
Paving the way for the future
As machine learning continues to evolve, its impact on enzyme engineering is expanding, allowing researchers to push beyond the limitations of conventional techniques. By combining AI-guided sequence generation with biochemical expertise, dsm-firmenich is paving the way for breakthroughs in industrial biotechnology, pharmaceutical applications, and beyond.
Join 13,000+ subscribers
Subscribe to our newsletter to stay updated about all the need-to-know content in the dairy sector, two times a week.
