Model Nanobody Affinity Maturation Strategies

ISEF Category: Cellular and Molecular Biology

Subcategory: Cellular Immunology  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

A single mutation can help a virus slip past a binder that used to work well. That makes antibody design feel like trying to hit a moving target in the dark. Your project asks a smart question, which search strategy finds better binders faster, greedy steps or evolutionary search? You can answer it with data, not guesswork.

What Is It?

Nanobodies are tiny antibody fragments from camelids. They bind to a target shape, such as the receptor binding domain, or RBD, of the SARS-CoV-2 spike protein. Affinity means how tightly the nanobody sticks. Breadth means how well it still binds after the target changes. Your project models that process with a fitness landscape, which is just a map that shows which versions of a sequence work better than others.

Think of it like hiking across hills in fog. A greedy search always moves uphill in the steepest nearby direction. An evolutionary search tries many paths, keeps some diversity, and can sometimes cross a dip to reach a higher hill later. In this project, you compare those strategies in silico, which means on a computer, using sequence and binding data or simulated scores.

Why This Is a Good Topic

This is a strong science fair topic because you can turn a real biology problem into a clear computational test. You can compare two search strategies, measure binding score, and track breadth across variant targets. That gives you a real performance question, not just a description of a protein. You also learn skills that matter in modern biology, like sequence analysis, modeling, and statistical comparison.

Research Questions

• How does greedy search compare with evolutionary search in finding high-affinity nanobody sequences against a SARS-CoV-2 RBD model?
• What is the effect of target variant diversity on the breadth of nanobody sequences found by each search strategy?
• Does allowing neutral mutations improve the final affinity landscape reached by evolutionary search?
• To what extent does starting from different seed nanobody sequences change search success in silico?
• Which search strategy finds sequences with the best balance of affinity and breadth across multiple RBD variants?
• How does the number of search iterations affect convergence speed and final binding score for each strategy?

Basic Materials

• Laptop or desktop computer with enough memory to run sequence analysis tools.
• Public amino acid sequence data for nanobodies and SARS-CoV-2 RBD variants from databases such as NCBI or UniProt.
• Spreadsheet software for organizing scores and comparing strategies.
• Free programming environment such as Python with Jupyter Notebook.
• Access to published binding scores or a simple scoring rule derived from sequence features.
• Basic graphing tool for plotting fitness landscapes and search paths.

Advanced Materials

• Workstation or university server access for larger simulation runs.
• Curated nanobody and RBD sequence datasets from literature or structural databases.
• Structural models from the Protein Data Bank for feature extraction.
• Python scientific stack for simulation, optimization, and statistics.
• Software for sequence alignment and mutation mapping.
• Access to molecular modeling tools if you add docking or structure-based scoring.

Software & Tools

• Python: Runs simulations, parses sequence data, and compares search strategies.
• Jupyter Notebook: Keeps code, notes, plots, and parameter tests in one place.
• Biopython: Handles sequence reading, alignment, and mutation analysis.
• NumPy: Stores scores and runs fast numerical calculations.
• Matplotlib: Plots fitness landscapes, convergence curves, and breadth metrics.

Experiment Steps

1. Define the exact target set, the nanobody starting points, and the score you will treat as fitness.
2. Choose one greedy strategy and one evolutionary strategy, then write down the rules for each before you run them.
3. Build a scoring framework that turns sequence changes into comparable numbers across all target variants.
4. Plan a way to test breadth, not just one best score, so your result reflects variant coverage.
5. Decide how you will repeat the search many times with different random seeds to check whether the result is stable.
6. Set up the plots and statistics you will use to compare convergence speed, final affinity, and breadth.

Common Pitfalls

• Using only one starting sequence, which can make the whole result depend on a lucky or unlucky seed.
• Comparing strategies with different scoring rules, which makes the final numbers unfair.
• Measuring only best affinity and ignoring breadth, which hides whether a sequence fails on variant targets.
• Treating simulated scores as if they were real binding constants, which overstates what the model can prove.
• Running too few repeats, which makes random variation look like a true strategy effect.

What Makes This Competitive

A strong version of this project does more than compare two algorithms. It tests whether one strategy wins across many starting points, many variant sets, and more than one scoring rule. You can raise the level by adding confidence intervals, effect sizes, and sensitivity checks. A very strong entry also explains what the model predicts about real antibody design, and where the model could break down.

Project Variations

• Compare greedy and evolutionary search on a different viral target, such as influenza hemagglutinin or a second coronavirus variant.
• Replace the sequence-only score with a structure-informed score based on predicted contact changes or docking features.
• Test whether multi-objective search finds better tradeoffs between affinity, breadth, and mutation count than single-objective search.

Learn More

• NCBI Virus: Search for viral sequence data and variant records through the NCBI database.
• Protein Data Bank: Find experimentally solved protein structures and related metadata through RCSB PDB.
• UniProt: Read protein annotations, domains, and sequence information in the UniProt database.
• Biopython Tutorial and Cookbook: Learn sequence handling and analysis, available from the Biopython documentation.
• Nature Reviews Immunology: Search PubMed for review articles on antibody affinity maturation and immune escape.