About Viral Sentry AI Platform

GitHub Release Preprint
Viral Sentry AI is a research prototype designed to explore the prediction of zoonotic potential in viruses and assist in identifying hypothetical threats to public health. At its core is VirSentAI, an architecture fine-tuned from the pretrained HyenaDNA model to investigate human host tropism. The framework is engineered to scan public DNA databases for new viral sequences and metadata, utilizing a local Large Language Model (LLM) with Web search plugin to infer host context when the host is unknown. For viruses identified with potential zoonotic risk, the system conducts exploratory drug repurposing via PLAPT to estimate affinity interactions between ChEMBL-indexed compounds and viral proteins. Our goal is to leverage AI and bioinformatics to provide preliminary computational leads for the research community. As a tool for hypothesis generation, all VirSentAI predictions—particularly those regarding therapeutic prioritization—require rigorous structural and experimental validation to confirm biological plausibility. In support of transparent research, the VirSentAI code is available as open-source software on GitHub.

Web Framework

Our Team