Badhan Das earned a bachelor of science degree in computer science from Bangladesh University of Engineering and Technology (BUET). When deciding to further her education, she was drawn to Virginia Tech and the Sanghani Center for their strong emphasis on interdisciplinary research that bridges computer science, data science, and real-world applications. 

“The center’s collaborative environment and focus on advancing AI and machine learning for complex data problems align perfectly with my research interests in computational biology. I was particularly inspired by the opportunity to work with researchers who integrate high-performance computing and data-driven modeling to address impactful scientific questions,” said Das.

As a Ph.D. student in computer science, Das said her work in large-scale genomic data analysis and model development has been greatly supported at the Sanghani Center, where she is advised by Lenwood Heath.

Her research focuses on modeling viral evolution using graph-theoretic and machine-learning frameworks. She develops computational representations that capture how viruses mutate and evolve, which is beyond what traditional phylogenetic trees can describe. 

Specifically, she created the Variant Evolution Graph (VEG) and the Mutation Learning Graph (MLG), which represent viral genomes as interconnected nodes and edges reflecting mutational changes and temporal order.

"By combining graph neural networks and transformer-based genome models, I aim to predict the emergence of new viral variants. For example, in studying SARS-CoV-2, my models analyze mutation patterns to identify potential future variants, offering insights to guide public health preparedness and vaccine design,” Das said.

She became interested in viral evolution after realizing how differently viruses evolve compared to other organisms. Their high mutation rates create mutant clouds (populations in which ancestral and descendant variants coexist), making it difficult for traditional phylogenetic trees to capture their actual dynamics. 

“Inspired by quasispecies theory, I began modeling viral evolution as a graph rather than a tree, bridging evolutionary theory with graph modeling,” Das said. “This approach reveals complex events unique to viruses -- such as back mutations, convergent evolution, and recurrent mutations -- that conventional methods often overlook.”

Her published work includes: 

·      "Variant evolution graph: Can we infer how SARS-CoV-2 variants are evolving?”, PLoS One, June 2025

·      "DeePSP-GIN: identification and classification of phage structural proteins using predicted protein structure, pretrained protein language model, and graph isomorphism network," in proceedings of the 15th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics,  December 2024

·    "HT-ARGfinder: A Comprehensive Pipeline for Identifying Horizontally Transferred Antibiotic Resistance Genes and Directionality in Metagenomic Sequencing Data," Frontiers in Environmental Science, June 2022

Das will graduate this month and plans to continue in academia, pursuing a postdoctoral or faculty position. 

“I aim to expand my work on graph-based and machine learning models of evolution to study not only viral systems but also cancer genomics, where similar mutational dynamics drive disease progression,” she said.

Her long-term goal is to lead an interdisciplinary research group that bridges computational modeling with real-world applications in health and disease prediction.