Drug Repurposing for Rare Disease Target Discovery

ISEF Category: Biomedical and Health Sciences

Subcategory: Genetics and Molecular Biology of Disease  ·  Difficulty: Advanced  ·  Setup: Home Setup  ·  Time: Full Year

The Hook

A single gene defect can leave a patient with no approved treatment, even when the protein's shape is already known. You can still hunt for a rescue drug on a laptop. AlphaFold helps you predict the target's structure, AutoDock helps you test FDA-approved compounds, and molecular dynamics helps you check whether the best hits stay put. That makes this a real drug-repurposing project, not just a list of guesses.

What Is It?

This project starts with a rare-disease protein, often one that has a known gene but no approved therapy. AlphaFold predicts the protein's 3D shape from its amino acid sequence. AutoDock then tests how well each drug may fit into a pocket on that shape, kind of like trying a stack of keys in one lock.

Molecular dynamics, or MD, adds a second check. It simulates how the protein and drug move over time, so you can see whether a docked pose stays stable or falls apart. That helps you sort out hits that look good for a moment from hits that may actually hold up better in a real biological setting.

Why This Is a Good Topic

This is a strong science fair topic because you can test it with clear numbers, not just opinions. You can compare docking scores, pose quality, contact patterns, and MD stability across a defined drug set. It connects to a real need, rare diseases often lack approved therapies, so repurposing can move faster than inventing a brand-new drug. You also learn useful skills in structure reading, data cleaning, molecular modeling, and result ranking.

Research Questions

• Which FDA-approved drugs rank highest against the chosen rare-disease target by docking score and pose quality?
• How does using an AlphaFold structure versus a homology model change the top-ranked drug list?
• Does adding a second docking run with a different search box change the most stable candidates?
• To what extent do molecular dynamics stability metrics agree with the initial docking ranks?
• What is the effect of filtering the library by approved use, safety, or blood-brain barrier penetration on the final hit list?
• Which predicted contacts appear most often among the top candidates, and do they match known active-site chemistry?

Basic Materials

• Laptop or desktop computer with at least 8 GB RAM.
• Stable internet access for downloading structures and compound files.
• Google account for Colab notebooks.
• AlphaFold Protein Structure Database access.
• PubChem compound records for FDA-approved drugs.
• AutoDock Vina installed in Colab or on your machine.
• ChimeraX or PyMOL for viewing docking poses.

Advanced Materials

• Linux workstation or cluster account with GPU support.
• GROMACS or AMBER for longer molecular dynamics runs.
• Curated access to OMIM, ClinVar, and UniProt annotations.
• A larger FDA-approved compound library with clean 3D structures.
• Ligand preparation tools such as Open Babel and RDKit.
• Protein pocket analysis tools such as fpocket or ChimeraX.

Software & Tools

• Google Colab: Runs docking and dynamics notebooks without local hardware.
• AutoDock Vina: Scores how well each drug may fit the target pocket.
• UCSF ChimeraX: Lets you inspect binding poses, contacts, and structure confidence.
• GROMACS: Checks whether top poses stay stable during molecular dynamics.
• PubChem: Provides drug structures and identifiers for building a clean screening library.

Experiment Steps

1. Choose one rare-disease gene, one protein isoform, and one binding question so your screen stays focused.
2. Decide whether your target structure will come from AlphaFold, an experimental structure, or a homology model, then document why.
3. Clean and standardize the FDA-approved drug library so every compound enters docking in the same format.
4. Set up one docking protocol, one pocket definition, and one ranking rule before you screen anything.
5. Re-score the best hits with molecular dynamics and compare stability against the starting docked pose.
6. Compare your final candidates with disease biology, known drug safety, and practical repurposing barriers.

Common Pitfalls

• Docking the wrong protein region, which gives high scores for poses that never reach the real active site.
• Ignoring AlphaFold confidence scores, which can make a weak loop or flexible tail look like a trustworthy pocket.
• Leaving drug protonation and tautomer states inconsistent, which changes charge and flips the ranking.
• Trusting one docking run, which hides unstable hits that only look strong by chance.
• Skipping post-docking stability checks, which lets short-lived poses look like real repurposing leads.

What Makes This Competitive

This project gets stronger when you compare more than one structure and more than one scoring view. A competitive version tests whether the same drugs stay near the top after you change the pocket definition, run repeated docking trials, and add molecular dynamics stability metrics. You also raise the bar when you explain why the best hits fit the disease biology, not just the score. That turns a simple screen into a careful filter for real repurposing candidates.

Project Variations

• Screen only drugs with known brain exposure if the rare disease affects the nervous system.
• Compare a wild-type protein model with a pathogenic mutant to see whether one mutation creates a new pocket.
• Test two ranking methods, docking score and contact-based rescoring, to see which one keeps the best candidates.

Learn More

• PubChem: Search FDA-approved drug structures, identifiers, and bioactivity records on the NIH compound database.
• OMIM: Read gene and disease summaries for Mendelian targets and phenotype context.
• AlphaFold Protein Structure Database: Find predicted protein structures when no solved structure exists.
• RCSB Protein Data Bank: Compare your model to experimental structures and ligands.
• PubMed: Search review articles on drug repurposing, molecular docking, and molecular dynamics.
• MIT OpenCourseWare: Find free lectures on structural biology, biochemistry, and computation for background.