---
title: BioAgent - AI Bioinformatics Demo
emoji: 🧬
colorFrom: blue
colorTo: green
sdk: docker
app_file: app.py
pinned: false
license: mit
---

# 🧬 BioAgent: AI-Powered Bioinformatics

**BioAgent** demonstrates AI-augmented bioinformatics workflows combining Claude's reasoning capabilities with real scientific database APIs.

## Features

### 💊 Drug Discovery Pipeline
- Search compounds via PubChem API (real data)
- Lipinski's Rule of Five assessment
- AI-powered drug-likeness analysis
- 2D structure visualization

### 🧬 Sequence Analysis
- UniProt protein search (real data)
- Amino acid composition analysis
- AI-powered sequence interpretation
- Hydrophobicity and structure predictions

### 🎯 Patient Stratification (NEW)
- Multi-omics integration (GEX, Mutations, CNV)
- Interactive UMAP patient clustering
- Modality attention visualization
- Kaplan-Meier survival analysis
- Biomarker discovery pipeline
- AI-generated clinical insights

### 📚 Literature Intelligence
- PubMed search integration (real data)
- AI-synthesized research summaries
- Theme and gap identification
- Citation management

### 🔬 Single-Cell Analysis Demo
- Interactive UMAP visualization
- Cluster statistics
- AI-powered biological interpretation

## Technology Stack

- **AI Backend**: Claude 3.5 Sonnet via Anthropic API
- **Scientific Skills**: [K-Dense Claude Scientific Skills](https://github.com/K-Dense-AI/claude-scientific-skills)
- **Data Sources**: PubChem, PubMed, UniProt (live APIs)
- **Frontend**: Streamlit with custom scientific theme

## Usage

1. Enter your Anthropic API key in the sidebar
2. Select an analysis module
3. Input your query (compound name, protein, sequence, etc.)
4. Get AI-augmented insights combined with real data

## Skills Integration

This demo implements Claude Scientific Skills as **context injection** - loading relevant SKILL.md documentation into the system prompt so Claude follows domain-specific best practices for:

- BioPython (sequence analysis)
- RDKit (molecular properties)
- Scanpy (single-cell analysis)
- PubMed/UniProt/ChEMBL APIs

## Built By

**Excelra Data Sciences** - Powering drug discovery with AI-augmented bioinformatics.

Part of the **Risentia** platform for Bioinformaticians as a Service.

## License

MIT License - Free for commercial and non-commercial use.