Alzheimer’s Prevention Scorecard From Genetic Data

ISEF Category: Translational Medical Science

Subcategory: Disease Prevention  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

APOE gets a lot of attention in Alzheimer’s risk, but genes do not act alone. Your body also responds to sleep, blood pressure, blood sugar, and other modifiable factors. That makes this project a powerful question, which pathways matter most between genotype and disease? You can use public data to rank those pathways instead of guessing.

What Is It?

This project asks you to study how genetic risk and lifestyle-linked biology connect to Alzheimer’s disease. APOE is a gene that strongly affects risk, especially the APOE-e4 variant. But the gene does not work in a vacuum. Other traits, like cholesterol levels, blood pressure, inflammation, and body mass index, may sit between APOE and disease risk. Those middle steps are called mediators.

Think of it like a traffic jam. APOE may be the starting roadblock, but the real slowdown can happen at several exits. Mendelian randomization is a method that uses inherited genetic variants as natural test tools. If a variant linked to one trait also changes disease risk, that suggests the trait may play a causal role. By combining UK Biobank summary GWAS data with OpenTargets and public genetic instruments, you can compare possible mediators and build a scorecard that ranks which modifiable factors look most worth targeting.

Why This Is a Good Topic

This is a strong science fair topic because you can test real biological claims with public data, not just write a report. The question is narrow enough to study, but big enough to matter for disease prevention. You can learn genetics, epidemiology, causal inference, and data analysis in one project. Best of all, the final result can point to practical prevention targets, not just a list of risk factors.

Research Questions

• How does APOE-linked genetic risk relate to Alzheimer’s risk after accounting for candidate lifestyle mediators?
• What is the effect of genetically predicted LDL cholesterol on Alzheimer’s risk in public summary GWAS data?
• Does genetically predicted blood pressure show a causal link to Alzheimer’s risk across different instruments?
• To what extent do inflammation-related traits mediate the relationship between APOE genotype and Alzheimer’s risk?
• Which modifiable trait ranks highest as a mediator when you compare effect size, precision, and consistency across datasets?
• How does the mediator ranking change when you test only traits with strong genetic instruments?

Basic Materials

• Laptop with enough memory to handle large summary statistics files.
• Spreadsheet software for tracking variants, traits, and effect sizes.
• Python or R installed locally.
• Public GWAS summary statistics from UK Biobank-related studies.
• OpenTargets gene-disease and trait evidence pages.
• Reference files for APOE region coordinates and common Alzheimer’s risk traits.
• Basic statistics notes on odds ratios, confidence intervals, and multiple testing.

Advanced Materials

• University or institutional computer access for larger data files.
• Python, R, or both, plus Bioconductor or similar genetics packages.
• Access to curated GWAS summary statistic databases and harmonized phenotype tables.
• LD reference panels for clumping and instrument selection.
• Mendelian randomization packages such as TwoSampleMR or MR-Base-compatible tools.
• Fine-mapping or colocalization tools for follow-up analyses.
• Visualization software for forest plots, scatter plots, and network diagrams.

Software & Tools

• Python: Cleans summary statistics, runs statistical tests, and makes plots for mediator ranking.
• R: Fits Mendelian randomization models and compares effect estimates across traits.
• TwoSampleMR: Helps select instruments and run standard Mendelian randomization workflows.
• OpenTargets: Finds trait and gene evidence that can help you choose candidate mediators.
• ImageJ: Not needed for this project, so you can ignore image-based analysis and focus on genetics data.

Experiment Steps

1. Define the exact disease outcome and the set of candidate mediators you will test.
2. Gather public summary statistics and confirm that the traits use comparable ancestry and phenotype definitions.
3. Select valid genetic instruments for each mediator and check for weak or overlapping signals.
4. Run Mendelian randomization analyses to estimate which mediators may causally affect Alzheimer’s risk.
5. Compare results across sensitivity tests so you can separate signal from pleiotropy and other bias.
6. Build a prevention scorecard that ranks mediators by strength, consistency, and practical modifiability.

Common Pitfalls

• Mixing datasets with different ancestry groups, which can distort Mendelian randomization estimates.
• Using weak genetic instruments, which makes causal claims unstable and noisy.
• Treating correlation as causation before checking for pleiotropy and reverse causation.
• Testing too many candidate traits without correcting for multiple comparisons, which creates false positives.
• Building a scorecard from effect size alone and ignoring confidence intervals, instrument strength, and biological plausibility.

What Makes This Competitive

A class-level version of this project may stop at a single Mendelian randomization test. A stronger version compares several mediators, runs sensitivity analyses, and explains why some signals fail. You can make it stand out by using careful instrument selection, ancestry matching, and a transparent ranking system. The best entries also connect the statistics back to a real prevention question that a clinician or researcher could use.

Project Variations

• Test whether lipid traits or blood sugar traits make stronger mediators than inflammation traits for Alzheimer’s risk.
• Compare APOE-linked mediation patterns across European-only, multi-ancestry, or matched-subset summary GWAS data.
• Build a simpler scorecard that ranks modifiable traits by consistency across two independent public datasets.

Learn More

• NIH MedlinePlus Genetics: Clear background on APOE, Alzheimer’s disease, and genetic risk, found by searching MedlinePlus Genetics.
• PubMed: Search for review articles on Mendelian randomization, APOE, and Alzheimer’s disease.
• Open Targets Platform: Free gene-disease evidence and target prioritization resource, searchable by gene or disease name.
• UK Biobank Research Analysis Platform documentation: Background on summary data access and phenotype structure, available through UK Biobank support pages.
• MIT OpenCourseWare: Search for statistics, genetics, and computational biology course materials to strengthen your analysis.