Files
2026-02-11 00:58:17 -05:00

38 lines
2.2 KiB
ReStructuredText

Case Study: Hypercholesterolemia Drug Discovery
===============================================================
.. image:: https://colab.research.google.com/assets/colab-badge.svg
:target: https://colab.research.google.com/drive/1UwJ6RwyUoqI5risKQ365EeFdDQWOeOCv?usp=sharing
:alt: Open in Google Colab
**Explore a full AI-driven drug discovery workflow using ToolUniverse and Gemini 2.5 Pro!**
This tutorial walks you through the hypercholesterolemia case study presented in our paper, "Democratizing AI scientist systems using ToolUniverse." You can run the entire workflow interactively in Google Colab: `ToolUniverse Case Study Notebook <https://colab.research.google.com/drive/1UwJ6RwyUoqI5risKQ365EeFdDQWOeOCv?usp=sharing>`_.
What does the case study cover?
-------------------------------
- **End-to-end drug discovery pipeline** for hypercholesterolemia
- **AI scientist reasoning**: See how Gemini 2.5 Pro uses ToolUniverse tools to select targets, analyze tissue expression, choose drugs, and optimize molecules
- **Human-in-the-loop**: Integrates expert feedback for final decision making
- **Tool calls and outputs**: All code and tool calls are shown step-by-step
Sections in the tutorial:
-------------------------
1. **Target Selection**: Identify top protein targets for hypercholesterolemia using OpenTargets and literature review
2. **Target Characterization**: Analyze tissue/organ expression of selected target to assess drug side effects
3. **Drug Selection**: Review current drugs targeting the protein and select a molecule to optimize
4. **Analog Search & ADMET Prediction**: Find structural analogs and predict pharmaceutical properties
5. **IP Review**: Check intellectual property status for the optimized molecule
How to use this tutorial:
-------------------------
- **Open the Colab notebook**: Click the badge or link above
- **Follow the step-by-step code**: Each section includes code, tool calls, and reasoning
- **Learn by example**: See how ToolUniverse enables agentic workflows for real-world biomedical research
Useful links:
-------------
- `ToolUniverse Documentation <https://aiscientist.tools>`_
- `ToolUniverse GitHub <https://github.com/mims-harvard/ToolUniverse>`_
- `Gemini 2.5 Pro <https://deepmind.google/technologies/gemini/>`_