1. Can you please briefly summarize the paper? 

This study addresses a critical limitation in predicting the virulence of Tomato Yellow Leaf Curl Virus (TYLCV), particularly the limited global generalizability of the existing IML-TYLCV model. Traditionally, TYLCV severity has been assessed through growers' visual inspection or image-based AI approaches. However, these methods depend on visible symptoms and may not be effective for detecting pre-symptomatic or asymptomatic infections. In our previous work, we developed IML-TYLCV, a genome-based prediction tool trained on Korean TYLCV isolates. However, its generalizability to globally diverse TYLCV isolates was limited. To overcome this limitation, we constructed a comprehensive global dataset of TYLCV genome sequences through extensive literature curation and developed DeepTYLCV, an interpretable deep learning framework. It integrates multiple protein language model embeddings with a hybrid architecture that combines a transformer encoder and a multi-scale convolutional neural network, enabling the model to capture both global and local virulence-associated motifs. By combining deep sequence representations with optimized conventional feature descriptors, DeepTYLCV outperformed IML-TYLCV in global prediction tasks. The interpretability analysis further showed that DeepTYLCV focuses on biologically meaningful regions of the viral proteins. Mapping these regions back to the viral genome may help identify sequence features associated with symptom severity and support future resistance-breeding strategies. To assess the real-world applicability of DeepTYLCV, we performed blind predictions on 15 TYLCV isolates, including international reference isolates and Korean field isolates. These predictions were experimentally validated using tomato plant infection assays, symptom severity scoring, and viral accumulation analysis. DeepTYLCV achieved 100% concordance between predicted and experimentally observed virulence classes.

Overview of the DeepTYLCV framework

2. Can you please tell us the main difficulties you had in the laboratory work and how you overcame them?

One of the main challenges in this study was constructing a reliable global TYLCV dataset. Because no readily available database with clearly annotated TYLCV isolates and their symptom severity, we could not simply download sequences and use them directly for model development. Instead, we had to collect complete TYLCV genome sequences from public databases and manually verify the virulence information by reviewing the corresponding literature. This process was time-consuming because the information was often scattered across different papers, and symptom descriptions were not reported consistently. In some cases, we had to carefully compare the reported phenotypes and exclude isolates that did not have clear or reliable severity information. Although this step required significant effort, it was essential for building a high-quality dataset. We overcame this challenge through systematic literature screening, careful manual curation, and repeated cross-checking of the collected information. This curated dataset became the foundation for developing DeepTYLCV and improving its ability to predict TYLCV severity across global isolates.

3. Please introduce your laboratory, university or organization to bio-researchers in Korea.

The Department of Integrative Biotechnology is part of the College of Biotechnology and Bioengineering at Sungkyunkwan University (SKKU). The department defines integrative biotechnology as a cutting-edge field focused on isolating, manipulating, and reintroducing genes to address critical challenges in food, health, and the environment. It offers a comprehensive curriculum spanning molecular biology, protein engineering, genetics, and immunology, with the goal of nurturing specialists who can develop novel biomaterials and therapeutic solutions. The department is also known for its faculty's active research contributions and for producing graduates who have successfully entered diverse fields, including the pharmaceutical, medical, and biotechnology sectors. Our Computational Biology and Bioinformatics Laboratory (CBBL) serves as an interface between biology and artificial intelligence. The laboratory, supervised by Associate Professor Balachandran Manavalan, focuses on developing AI-based solutions for peptide therapeutics, RNA/DNA modification prediction, protein function analysis, toxicity prediction, plant science, and biomedical applications. According to ScholarGPS, Prof. Manavalan is recognized as a Highly Ranked Scholar - Prior Five Years, placing within the top 0.05% of scholars, with rankings of #18 in Bioinformatics and #26 in Peptide.

4. Please tell us your experiences and your thoughts related to research activities abroad.

I am currently pursuing a Combined Master's and Ph.D. program in the CBBL at Sungkyunkwan University. Although our lab is based in Korea, our research team comprises talented members from different countries, creating a truly international environment. Working collaboratively with such a diverse team has helped me expand my knowledge from foundational concepts to emerging research approaches. Sharing diverse academic perspectives and advanced methodologies has been instrumental in solving complex problems and in developing this study.

5. Can you provide some advice for younger scientists who have plans to study abroad?

For younger scientists planning to study abroad, I would advise them to focus first on building a strong foundation in their field. A solid understanding of fundamental concepts and techniques gives students the confidence to adapt to new research topics, understand problems more clearly, and contribute meaningfully to their projects. At the same time, it is important to be open-minded and communicative. Studying abroad offers opportunities to work with people from diverse academic and cultural backgrounds, and this diversity can greatly improve the way we think about science. Young researchers should feel comfortable discussing their ideas. Because research often involves unexpected results and challenges, open communication can help them find better solutions, learn from different perspectives, and grow as independent scientists.

6. Future plan?

My future goal is to develop a comprehensive tool that integrates multi-omics data to predict viral severity across entire viral genera or families. Unlike species-specific models, a broad predictive framework will enable us to shift from reactive to preemptive surveillance, assessing the threat posed by novel or uncharacterized viruses before an outbreak occurs. Furthermore, by identifying conserved virulence mechanisms within a viral family, the tool will help breeders develop broad-spectrum, durable resistance in crops rather than short-term, species-specific solutions.

7. Do you have anything else that you would like to tell Korean scientists and students?

I would like to tell Korean scientists and students to protect their curiosity throughout their research journey. In a competitive research environment, it is easy to focus mainly on publications, deadlines, and results. These are important, but I believe the real strength of a scientist comes from the ability to keep asking meaningful questions and to stay excited about finding answers. Research does not always move in a straight line. Sometimes experiments fail, or progress feels slow. During those moments, curiosity can become the strongest motivation to continue. It helps us look at problems from a new angle, learn new methods, and turn difficulties into opportunities for growth. From my experience in Korea, I have seen many hardworking and dedicated students. My message to them is not to lose the joy of learning while pursuing scientific success. Skills and knowledge will continue to grow with time, but curiosity is what keeps research meaningful and creative.