AI-powered platform to predict pathogenicity of DNA mutations using Evo2
A full-stack bioinformatics project for gene variant effect prediction, powered by the Evo2 large language model from Arc Institute. This project features: A Next.js (React) frontend A FastAPI backend deployed with Modal (H100 GPU) Deep analysis and classification of genetic variants (e.g. BRCA1)
AI-Powered Genomic Variant Pathogenicity Predictor
Built an AI-driven platform leveraging the Evo2 genomic foundation model to predict the pathogenicity of Single Nucleotide Variants (SNVs) at single-nucleotide resolution. Designed a scalable FastAPI backend deployed on serverless NVIDIA H100 GPUs (Modal) to compute Delta Likelihood scores across millions of DNA base pairs efficiently. Developed an interactive Next.js dashboard integrating AI predictions with ClinVar clinical data, enabling real-time diagnostic insights for high-risk genes such as BRCA1.