DeepTYLCV, developed at Sungkyunkwan University, uses artificial intelligence and viral genome sequence data to predict tomato yellow leaf curl virus virulence. Published in Plant Communications, the model identifies mild and severe TYLCV strains before field symptoms appear, supporting early disease surveillance, resistance breeding, precision agriculture and improved tomato crop protection worldwide.
The post AI Model Predicts Tomato Virus Severity With 100% Accuracy appeared first on Seed World.
A research team at Sungkyunkwan University has developed DeepTYLCV, an artificial intelligence model designed to predict the virulence of tomato yellow leaf curl virus, one of the most damaging viral diseases affecting tomato crops worldwide.
The team was led by Professor Balachandran Manavalan of the Department of Integrative Biotechnology, with co-first authors Dr. Nattanong Bupi, Hariharan Sangaraju and Duong Thanh Tran. The study was published in Plant Communications.
TYLCV can cause severe symptoms in tomato plants, including leaf curling, yellowing, stunted growth and major yield losses. In recent years, highly virulent strains have continued to spread across regions, and some have overcome genetic resistance in tomato cultivars. This has created an urgent need for disease surveillance tools that are accurate, early, scalable and based on viral sequence information, according to a press release.
Professor Manavalan’s team has worked extensively at the intersection of biology and artificial intelligence, developing AI tools for peptide therapeutics, RNA and DNA modification prediction, protein function analysis, toxicity prediction, plant science and biomedical applications.
In 2023, the team developed IML-TYLCV, the first genome-based tool for predicting TYLCV severity. However, because that model was trained mainly on Korean isolates, its ability to assess globally diverse TYLCV strains was limited. This led to the development of DeepTYLCV, a more robust AI framework for predicting TYLCV virulence across global viral isolates.
Unlike field diagnosis or image-based AI models, which rely on visible symptoms and can be affected by environmental conditions, DeepTYLCV uses viral genome-derived sequence information. This allows the model to identify mild and severe strains before symptoms are confirmed in the field, offering a scalable approach for monitoring emerging viral variants.
DeepTYLCV combines protein language model embeddings with a hybrid architecture that includes a Transformer encoder and a multi-scale convolutional neural network. This enables the model to detect both broad sequence patterns and local motifs linked to virulence.
By integrating deep sequence representations with optimized conventional feature descriptors, DeepTYLCV outperformed the earlier IML-TYLCV model.
The study also included experimental validation. The team conducted blind predictions for 15 TYLCV isolates, including international reference isolates and Korean field isolates. These predictions were tested through tomato plant infection assays, symptom severity scoring and viral accumulation analysis.
DeepTYLCV achieved 100% agreement between predicted and experimentally observed virulence classes, highlighting its potential for identifying emerging severe TYLCV variants.
The work shows how AI, viral genomics and plant pathology can be combined to support precision agriculture and crop disease management. DeepTYLCV could become a valuable tool for early viral surveillance, resistance breeding programs and rapid assessment of newly emerging TYLCV strains.
The post AI Model Predicts Tomato Virus Severity With 100% Accuracy appeared first on Seed World.