AlphaFold Demo for Biology Students

An interactive Streamlit application demonstrating AI-powered protein structure prediction using AlphaFold/ESMFold. Perfect for biology students with no programming background!

What This Demo Does

• Predicts 3D protein structures from amino acid sequences
• Visualizes proteins in interactive 3D
• Explains the difference between creating vs using AI models
• Introduces modern bio-AI tools and resources
• Encourages hands-on exploration ("vibe coding")

Quick Start (For Non-Programmers)

Prerequisites

You'll need:

• A computer (Windows, Mac, or Linux)
• Internet connection
• 15 minutes for setup

Easy Installation (Recommended)

We've included automated setup scripts to make installation easier!

Windows:

1. Double-click setup.bat
2. Wait for installation to complete
3. Double-click run.bat to start the app

Mac/Linux:

1. Open Terminal in the project folder
2. Run: bash setup.sh
3. Run: bash run.sh to start the app

The scripts will automatically:

• Create a virtual environment
• Install all dependencies
• Launch the app in your browser

Manual Installation Steps

Step 1: Install Python

Windows:

1. Go to https://www.python.org/downloads/
2. Download Python 3.10 or newer
3. Run the installer
4. IMPORTANT: Check the box "Add Python to PATH"
5. Click "Install Now"

Mac:

1. Open Terminal (search for "Terminal" in Spotlight)
2. Install Homebrew if you don't have it:

   /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

3. Install Python:

   brew install python

Linux (Ubuntu/Debian):

sudo apt update
sudo apt install python3 python3-pip

Step 2: Download This Project

Option A: Using Git (if you have it)

git clone <your-repository-url>
cd alphafolddemo

Option B: Download ZIP

1. Click the green "Code" button on GitHub
2. Click "Download ZIP"
3. Extract the ZIP file
4. Open Terminal/Command Prompt and navigate to the folder:

   cd path/to/alphafolddemo

Step 3: Create Virtual Environment and Install Packages

A virtual environment keeps this project's packages separate from your system Python.

Windows:

# Create virtual environment
python -m venv venv-windows

# Activate it
venv-windows\Scripts\activate

# Install packages
pip install -r requirements.txt

Mac/Linux:

# Create virtual environment
python3 -m venv venv

# Activate it
source venv/bin/activate

# Install packages
pip install -r requirements.txt

You should see (venv-windows) (Windows) or (venv) (Mac/Linux) at the beginning of your command prompt, indicating the virtual environment is active.

This will install:

• streamlit - The web app framework
• py3Dmol - 3D molecule visualization
• stmol - Streamlit integration for py3Dmol
• requests - For making API calls

Step 4: Run the App

Make sure your virtual environment is activated (you should see (venv) in your prompt), then run:

streamlit run app.py

The app will automatically open in your web browser at http://localhost:8501

If it doesn't open automatically, copy that URL into your browser.

Important: Every time you open a new terminal to run the app, you need to activate the virtual environment first:

• Windows: venv-windows\Scripts\activate
• Mac/Linux: source venv/bin/activate

Using the Demo

1. Choose a protein: Select from the dropdown menu or paste your own sequence
2. Click "Predict Structure": Wait 30-60 seconds for the prediction
3. Explore the 3D visualization: Rotate, zoom, and examine the structure
4. Read the educational content: Check the sidebar and tabs for learning materials
5. Download structures: Save PDB files for further analysis

What's Included

Files in This Project

• app.py - The main Streamlit application
• requirements.txt - Python package dependencies
• PRESENTATION.md - Complete presentation slides (20+ slides)
• README.md - This file!
• setup.sh / setup.bat - Automated setup scripts (Mac/Linux & Windows)
• run.sh / run.bat - Quick run scripts (Mac/Linux & Windows)
• venv/ - Mac/Linux virtual environment folder (created after setup.sh)
• venv-windows/ - Windows virtual environment folder (created after setup.bat)

Features

• Interactive 3D protein visualization with py3Dmol
• Pre-loaded example proteins (insulin, lysozyme, myoglobin)
• Educational sidebar explaining:
  • What is AlphaFold?
  • Creating vs Using AI models
  • Other bio-AI tools (Benchling, PubMed, BioRender, etc.)
  • Career paths in computational biology
• Three educational tabs:
  • About Proteins (biology basics)
  • About the AI (how AlphaFold works)
  • Vibe Coding Tips (getting started with programming)
• Download predictions as PDB files
• Free API usage (ESMFold from Meta)

Example Proteins Included

1. Human Insulin - Hormone regulating blood sugar
2. Human Lysozyme - Antibacterial enzyme
3. Myoglobin - Oxygen storage protein
4. Custom - Enter your own sequence!

Troubleshooting

"streamlit: command not found"

Make sure your virtual environment is activated first:

Windows:

venv-windows\Scripts\activate
streamlit run app.py

Mac/Linux:

source venv/bin/activate
streamlit run app.py

If that doesn't work, run streamlit via Python:

Windows:

python -m streamlit run app.py

Mac/Linux:

python3 -m streamlit run app.py

"Module not found" errors

Make sure your virtual environment is activated, then reinstall:

# Activate venv first (see above)
pip install --upgrade -r requirements.txt

Virtual environment not activating

Windows (PowerShell): You may need to enable script execution:

Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser

Then try activating again:

venv-windows\Scripts\activate

Mac/Linux: Make sure you're using source:

source venv/bin/activate

Prediction fails or times out

• Check your internet connection
• Try a shorter protein sequence (< 200 amino acids)
• Wait a few minutes and try again (API might be busy)

Port already in use

Change the port:

streamlit run app.py --server.port 8502

Browser doesn't open

Manually open: http://localhost:8501

Understanding the Code (For the Curious)

Main Components

# 1. Import libraries
import streamlit as st  # Web app framework
import requests          # API calls
import py3Dmol          # 3D visualization
from stmol import showmol  # Display molecules

# 2. Set up the page
st.set_page_config(...)  # Configure appearance

# 3. Create UI elements
st.header("...")         # Headers
st.text_area("...")      # Input boxes
st.button("...")         # Buttons

# 4. Make predictions
response = requests.post(api_url, data=sequence)

# 5. Visualize results
view = py3Dmol.view()    # Create viewer
view.addModel(pdb_data)  # Add structure
showmol(view)            # Display

How the Prediction Works

Your Sequence
    ↓
requests.post() sends to ESMFold API
    ↓
ESMFold AI model predicts structure
    ↓
Returns PDB file (3D coordinates)
    ↓
py3Dmol visualizes in browser

Extending This Project

Easy Modifications (No Experience Needed)

1. Add your own protein examples:

   • Find a protein on UniProt (https://www.uniprot.org/)
   • Copy the sequence
   • Add to example_proteins dictionary in app.py

2. Change colors:

   • Find view.setStyle({'cartoon': {'color': 'spectrum'}})
   • Change 'spectrum' to 'red', 'blue', 'green', etc.

3. Add information:

   • Edit the sidebar expanders
   • Add your favorite proteins or diseases

Intermediate Modifications (Some Python)

1. Add protein properties:

   • Calculate molecular weight
   • Count amino acid types
   • Predict isoelectric point

2. Batch predictions:

   • Upload a FASTA file with multiple sequences
   • Predict all structures
   • Download as a ZIP file

3. Compare structures:

   • Predict two proteins
   • Visualize side-by-side
   • Calculate RMSD (structural similarity)

Advanced Modifications (More Programming)

1. Use real AlphaFold:

   • Set up Google Colab
   • Run full AlphaFold2
   • Get highest accuracy predictions

2. Add analysis tools:

   • Ramachandran plots
   • Secondary structure prediction
   • Binding site identification

3. Create database:

   • Store predictions in SQLite
   • Track prediction history
   • Compare multiple versions

Resources for Learning

Biological Background

• Protein Data Bank: https://www.rcsb.org/
• UniProt: https://www.uniprot.org/
• NCBI: https://www.ncbi.nlm.nih.gov/

AlphaFold Resources

• AlphaFold Database: https://alphafold.ebi.ac.uk/
• ESMFold: https://esmatlas.com/
• Original Paper: Nature (2021) "Highly accurate protein structure prediction with AlphaFold"

Learning to Code

• Rosalind (bioinformatics problems): http://rosalind.info/
• Python for Biologists: https://pythonforbiologists.com/
• BioPython Tutorial: https://biopython.org/wiki/Documentation
• Kaggle (data science): https://www.kaggle.com/learn

Visualization Tools

• PyMOL: https://pymol.org/ (professional visualization)
• ChimeraX: https://www.cgl.ucsf.edu/chimerax/ (free, powerful)
• Mol*: https://molstar.org/ (web-based)

AI Tools for Biology

• Claude: https://claude.ai/ (AI assistant with bio MCP servers)
• Benchling: https://www.benchling.com/ (lab management)
• BioRender: https://biorender.com/ (scientific figures)
• PubMed: https://pubmed.ncbi.nlm.nih.gov/ (literature)

Using the Presentation

The PRESENTATION.md file contains a complete 20-slide presentation covering:

1. Introduction to protein folding
2. The AlphaFold breakthrough
3. Creating vs using AI models
4. Foundation models in biology
5. Other bio-AI tools
6. Career paths
7. Vibe coding philosophy
8. Hands-on exercises
9. Real-world impact
10. Getting started guide

How to Present

Option 1: Convert to PowerPoint

Use a markdown-to-slides tool:

• Marp: https://marp.app/
• Slidev: https://sli.dev/
• Deckset (Mac): https://www.deckset.com/

Option 2: Present from Markdown

• Open in a markdown viewer
• Use presentation mode in VS Code
• Convert to PDF and present

Option 3: Create Custom Slides

• Use the content as a script
• Create slides in PowerPoint/Google Slides
• Add images and animations

Suggested Presentation Flow

1. Introduction (5 min)

   • Slides 1-3: Problem setup

2. AlphaFold Breakthrough (10 min)

   • Slides 4-6: The solution

3. Creating vs Using (10 min)

   • Slides 5-8: Key distinction

4. Live Demo (10 min)

   • Slide 10: Use the app!

5. Career Paths (10 min)

   • Slides 11-12: Future opportunities

6. Hands-On (15 min)

   • Slide 13: Let students try it

7. Q&A (10 min)

   • Slide 19: Discussion

Total: ~70 minutes (adjust as needed)

Common Use Cases

For Students

• Learn structural biology: See how sequence determines structure
• Explore proteins: Visualize your favorite proteins
• Class projects: Use for presentations and reports
• Research: Predict structures for your lab work

For Educators

• Lecture demos: Show real AI in action
• Lab exercises: Let students predict structures
• Assignments: "Predict and analyze protein X"
• Inspiration: Encourage computational thinking

For Researchers

• Quick predictions: Fast structure estimation
• Hypothesis generation: "What if this mutant..."
• Preliminary analysis: Before expensive experiments
• Visualization: Share structures with collaborators

Citation

If you use this demo in your work or teaching, please cite:

AlphaFold:

Jumper, J., Evans, R., Pritzel, A. et al. Highly accurate protein structure
prediction with AlphaFold. Nature 596, 583–589 (2021).
https://doi.org/10.1038/s41586-021-03819-2

ESMFold:

Lin, Z., Akin, H., Rao, R. et al. Evolutionary-scale prediction of atomic-level
protein structure with a language model. Science 379, 1123-1130 (2023).
https://doi.org/10.1126/science.ade2574

License

This demo is for educational purposes. The AlphaFold and ESMFold models have their own licenses:

• AlphaFold: Apache 2.0 License
• ESMFold: MIT License
• This demo code: MIT License (free to use, modify, share)

Contributing

Want to improve this demo? Great! Here's how:

1. Fork the repository
2. Make your changes
3. Test thoroughly
4. Submit a pull request
5. Describe what you changed and why

Ideas for contributions:

• Add more example proteins
• Improve visualization options
• Add more educational content
• Fix bugs or improve performance
• Translate to other languages
• Add accessibility features

Support

Getting Help

• GitHub Issues: Report bugs or request features
• Discussions: Ask questions and share ideas
• Email: [Your email if you want to provide support]

Known Limitations

• ESMFold is slightly less accurate than AlphaFold2
• Long sequences (>400 aa) may be slow or fail
• No support for protein complexes (use AlphaFold-Multimer instead)
• Requires internet connection
• API may have rate limits

Acknowledgments

• DeepMind for AlphaFold
• Meta AI for ESMFold and the free API
• Streamlit for the web framework
• 3Dmol.js for visualization
• The open-source bioinformatics community

FAQ

Q: Is this real AlphaFold? A: This demo uses ESMFold, a similar but faster model from Meta. For the highest accuracy, use AlphaFold2.

Q: Can I use this for my research? A: Yes! But validate critical predictions experimentally.

Q: Do I need a GPU? A: No! The prediction happens on Meta's servers.

Q: Is it free? A: Yes! ESMFold's API is currently free to use.

Q: How accurate is it? A: ~85-90% accuracy on average, similar to experimental methods for many proteins.

Q: Can I predict protein complexes? A: Not with this demo. Use AlphaFold-Multimer for complexes.

Q: What's the maximum sequence length? A: Technically ~1000 amino acids, but shorter (<400) is recommended for speed.

Q: Can I run this offline? A: No, it requires internet for the API. You could set up local AlphaFold/ESMFold for offline use.

Q: Can I modify and share this? A: Yes! It's open source (MIT License).

What's Next?

After trying this demo, consider:

1. Explore AlphaFold Database: Download pre-computed structures
2. Learn PyMOL/ChimeraX: Professional visualization tools
3. Try BioPython: Analyze sequences and structures programmatically
4. Take a course: Rosalind, Coursera, or edX bioinformatics
5. Join a lab: Get hands-on research experience
6. Build something: Modify this app or create your own!

Version History

• v1.0 (2024): Initial release
  • Basic prediction and visualization
  • Educational content
  • Example proteins

Contact

Questions? Suggestions? Found a bug?

• GitHub: [Your GitHub]
• Email: [Your email]
• Twitter/X: [Your handle]

---

Remember: You don't need to understand everything to get started. Just start exploring!

The best way to learn is by doing. Pick a protein, predict its structure, and see what you discover.

Happy folding!