🧬 Variant Analysis with Evo2

AI-Powered DNA Mutation Pathogenicity Prediction

Predict whether a DNA mutation is disease-causing or harmless using a state-of-the-art AI model, real clinical databases, and GPU-accelerated inference — all wrapped in a modern full-stack web app.

---

🚀 Why This Project?

DNA mutations play a critical role in diseases like cancer — but understanding their impact is slow, complex, and expensive.

This project demonstrates how modern AI + cloud GPUs can:

• Analyze DNA mutations in seconds
• Predict disease risk
• Compare results with real clinical databases
• Present everything in a clean, beginner-friendly UI

🔥 No biology background required — the app handles the complexity for you.

---

✨ What You Can Do

• 🧬 Predict if a DNA mutation is pathogenic or benign
• ⚖️ Compare AI predictions with ClinVar medical classifications
• 🔍 Search genes like BRCA1 or browse entire chromosomes
• 🌍 Choose genome assemblies (hg38, etc.)
• 📊 Get prediction confidence scores
• ⚡ Run AI inference on NVIDIA H100 GPUs
• 🧪 Explore real genomic and clinical data interactively

---

🧠 How It Works (Simple)

• DNA is made of A, T, G, C
• A single letter change = mutation
• Some mutations cause disease, some don’t

This app:

1. Takes a mutation
2. Runs it through an AI model (Evo2)
3. Compares results with real medical data
4. Shows a clear, understandable result

---

🏗️ Architecture (High Level)

Frontend (Next.js)
        ↓
FastAPI Backend
        ↓
Evo2 AI Model (GPU)
        ↓
ClinVar + UCSC APIs

• Serverless GPU inference
• FastAPI REST endpoints
• Modern React UI

---

🧩 Key Features

• 🧬 Evo2 large language model for genomic analysis
• 🩺 Pathogenic vs benign prediction
• ⚖️ AI vs ClinVar comparison view
• 💯 Prediction confidence scoring
• 🗺️ Chromosome & gene browser
• 📜 Reference genome visualization (UCSC)
• 🔬 Real clinical variant data (NCBI ClinVar)
• ⚡ NVIDIA H100 GPU acceleration
• 🚀 Serverless deployment with Modal
• 📱 Fully responsive UI

---

🛠️ Tech Stack

Backend

• Python 3.12
• FastAPI
• Modal (Serverless GPUs)
• Evo2 LLM
• NCBI ClinVar API
• UCSC Genome API

Frontend

• Next.js
• React
• TypeScript
• Tailwind CSS
• Shadcn UI
• T3 Stack

---

📦 Getting Started

Clone the Repo

git clone https://github.com/GeneralSubhra/variant-analysis-evo2
cd variant-analysis-evo2

---

⚙️ Backend Setup

cd backend

Prerequisites

• Python 3.12
• uv package manager
  https://github.com/astral-sh/uv

Install & Run

uv venv --python 3.12
source .venv/bin/activate   # Mac/Linux
# or .venv\Scripts\activate # Windows

uv pip install -r requirements.txt
modal setup
modal run main.py

Deploy (Production)

modal deploy main.py

---

🎨 Frontend Setup

cd frontend
npm install
npm run dev

App runs at:

http://localhost:3000

---

📚 Evo2 Model

• 📄 Paper: https://www.science.org/doi/10.1126/science.ado9336
• 💻 GitHub: https://github.com/ArcInstitute/evo2

---

👉 Please star ⭐ the repo — it really helps!