GWAS (Genome-Wide Association Studies) — Genome-wide association study (GWAS) concepts (intro) - Genetics

SECTION 1

Historical Context & Motivation

For most of human history, people noticed that certain diseases and traits "run in families," but nobody knew exactly why. Scientists understood that DNA (deoxyribonucleic acid) carried the instructions for building our bodies, yet finding the exact spots in our DNA connected to a particular disease felt like searching for a needle in a haystack. The human genome contains roughly 3 billion base pairs — that is a lot of hay!

Early approaches to finding disease genes required scientists to study large families with many affected members and trace how a disease moved through generations. This method, called linkage analysis, worked well for diseases caused by a single gene, like cystic fibrosis. However, many common conditions — such as diabetes, heart disease, and asthma — involve lots of genes working together, each contributing just a small effect. Linkage analysis was not powerful enough to catch those tiny contributions. Scientists needed a completely new strategy.

1990 • Human Genome Project Launches

An international effort begins to sequence the entire human genome. This project would eventually provide the reference map that makes GWAS possible.

2001 • First Draft of the Human Genome

Scientists publish the first draft of the human genome, revealing about 3 billion base pairs of DNA and roughly 20,000–25,000 genes.

2002 • International HapMap Project Begins

Researchers catalog common genetic variations (called SNPs) across different human populations. This catalog becomes the foundation for GWAS.

2005 • First Successful GWAS Published

A landmark study on age-related macular degeneration (an eye disease) identifies a genetic variant strongly linked to the condition — proving that GWAS works.

2007–Present • GWAS Explosion

Thousands of GWAS studies are published, linking genetic variants to hundreds of diseases and traits including height, diabetes, and cancer risk.

The key question GWAS set out to answer was simple but powerful: Which tiny differences in DNA are more common in people with a disease compared to people without it? Answering this question required new technology, massive datasets, and clever statistics — all of which came together in the early 2000s.

SECTION 2

Core Principles & Key Definitions

Before diving into how GWAS works, you need to understand a few important ideas. A genome-wide association study (GWAS) is a research method that scans the entire genome of many people to find genetic variations associated with a specific trait or disease. Let's break down the building blocks.

SNPs (Single Nucleotide Polymorphisms)

A SNP (pronounced "snip") is a single-letter change in the DNA code. For example, one person might have an A at a particular spot while another has a G. Humans share about 99.9% of their DNA, so SNPs are the tiny differences that make us unique. There are roughly 4–5 million SNPs in each person's genome.

Alleles

An allele is one version of a gene or SNP. At any SNP location, you have two alleles — one inherited from each parent. GWAS compares which alleles appear more often in people with a disease versus those without.

Case vs. Control Groups

GWAS compares two groups: the case group (people with the disease or trait) and the control group (people without it). By comparing SNP frequencies between these groups, scientists spot which variants may be linked to the condition.

Statistical Association

A GWAS does not prove that a SNP causes a disease. It shows a statistical association — meaning the SNP and the disease appear together more often than you would expect by chance. Think of it as a clue, not a verdict.

p-value & Genome-Wide Significance

Because GWAS tests millions of SNPs, scientists use a very strict cutoff called genome-wide significance. A SNP must have a p-value smaller than 5 × 10⁻⁸ to be considered significant. This extremely low threshold prevents false alarms.

✦

KEY TAKEAWAY

Imagine you have two jars of colored candies. Jar A holds candies eaten by people with allergies, and Jar B holds candies eaten by people without allergies. If you notice that red candies appear in Jar A way more often than in Jar B, you'd suspect that red candies might be linked to allergies. That is exactly what GWAS does — except instead of candy colors, it compares DNA letter variants (SNPs) between people with and without a disease.

SECTION 3

Visualizing a GWAS: The Manhattan Plot

The signature visual of every GWAS study is the Manhattan plot. It is named after the skyline of Manhattan in New York City because the tall spikes of data points look like skyscrapers rising above the rest of the buildings. Each dot on the plot represents one SNP, and its height shows how strongly that SNP is associated with the trait being studied.

Each dot represents one SNP. The x-axis shows the SNP's position across different chromosomes (alternating colors). The y-axis shows how strongly each SNP is associated with the trait (higher = stronger). The red dashed line marks the genome-wide significance threshold (p = 5 × 10⁻⁸). Only SNPs above this line, like the peak on chromosome 5, are considered truly significant.

In the Manhattan plot above, notice how most dots sit at the bottom of the chart. These SNPs show little or no association with the trait. The dots that shoot up like skyscrapers — especially the cluster on chromosome 5 — are the exciting ones. They cross above the red dashed significance line, meaning those SNPs are found at different rates in the case group versus the control group far more often than random chance would predict.

The y-axis uses a special scale called −log₁₀(p-value). This just means that smaller p-values (stronger evidence) get placed higher on the chart. A SNP with a p-value of 0.001 sits at 3 on this scale, while a SNP with a p-value of 0.00000001 (10⁻⁸) sits at 8. The higher the dot, the stronger the evidence that the SNP is truly associated with the disease.

SECTION 4

How GWAS Works — Step by Step

Running a GWAS involves several key steps. While the full statistical methods can get complex, the core logic is straightforward: compare DNA variants between two groups and look for differences that are too large to be caused by chance alone.

Step 1 — Recruit Participants

Scientists gather a large number of people — often thousands or even hundreds of thousands. They split participants into a case group (people who have the disease or trait) and a control group (people who do not). Larger studies give more reliable results.

Step 2 — Genotype the DNA

Each person's DNA is placed on a special SNP array (also called a DNA microarray or genotyping chip). This tiny chip reads hundreds of thousands to millions of SNPs across the genome. The result is a list of which allele each person carries at each SNP position.

Step 3 — Compare Allele Frequencies

For every single SNP, scientists calculate the allele frequency (how common each version of the SNP is) in the case group and the control group separately. If a SNP's allele shows up much more often in the case group, it might be associated with the disease.

ALLELE FREQUENCY

f = (number of copies of allele A) ÷ (total number of alleles tested)

Where f is the allele frequency, a decimal between 0 and 1. If f = 0.30, the allele appears in 30% of the chromosomes sampled.

Step 4 — Calculate Statistical Significance

For each SNP, a statistical test produces a p-value — a number that tells you how likely it is that the observed difference between groups happened by random chance. A very small p-value means the difference is probably real, not just a fluke.

GENOME-WIDE SIGNIFICANCE THRESHOLD

p < 5 × 10⁻⁸

Because GWAS tests approximately 1 million SNPs, the standard significance threshold is extremely strict: p must be less than 0.00000005. This is called a Bonferroni correction — dividing the usual 0.05 threshold by roughly 1 million tests.

Step 5 — Plot and Interpret Results

The results are displayed as a Manhattan plot (Section 3). Significant SNPs that pass the threshold are flagged for further study. Scientists then examine the genes near those SNPs to understand what biological pathways might be involved in the disease.

SECTION 5

Key Concepts in GWAS Design

Several important concepts help scientists design better GWAS studies and interpret their results accurately. Understanding these ideas will help you see both the power and the limitations of this technique.

This flowchart shows the six main stages of a GWAS study, from recruiting participants to following up on significant findings. The bottom panel highlights key principles at each stage.

Linkage Disequilibrium (LD)

Linkage disequilibrium (LD) is a concept that explains why GWAS does not need to test every single SNP in the genome. SNPs that are physically close together on a chromosome tend to be inherited together as a group (called a haplotype). If you know the identity of one SNP in a group, you can often predict the others. This means a GWAS chip testing 1 million SNPs can actually "cover" most of the genome's common variation, even though there are millions more SNPs out there.

Effect Size & Odds Ratio

When a GWAS identifies a significant SNP, scientists also measure the effect size — how much that SNP increases or decreases the risk of the disease. This is often reported as an odds ratio (OR). An OR of 1.0 means no effect. An OR of 1.3 means people with that allele have 30% higher odds of having the disease. Most GWAS hits have small effect sizes (OR between 1.1 and 1.5), which is why complex diseases involve many genes.

ODDS RATIO

OR = (odds of disease in exposed group) ÷ (odds of disease in unexposed group)

An OR > 1 means the allele increases disease risk. An OR < 1 means it is protective. An OR = 1 means no association.

SECTION 6

Worked Example — Interpreting a GWAS Result

Let's walk through a simplified example of how scientists interpret a single SNP from a GWAS study on Type 2 diabetes.

Is SNP rs7903146 Associated with Type 2 Diabetes?

1
Step 1 — Identify the Data

A GWAS study collected data from 5,000 people with Type 2 diabetes (cases) and 5,000 people without it (controls). At SNP rs7903146, each person can carry one of two alleles: C or T. The researchers counted how many T alleles appeared in each group.
2
Step 2 — Calculate Allele Frequencies

Each person has 2 copies of each chromosome, so the total alleles per group = 2 × 5,000 = 10,000. In the case group, the T allele appeared 3,500 times. In the control group, the T allele appeared 2,500 times.

Case frequency: f_case = 3,500 ÷ 10,000 = 0.35 (35%). Control frequency: f_control = 2,500 ÷ 10,000 = 0.25 (25%).
3
Step 3 — Compare the Frequencies

The T allele is 10 percentage points more common in the case group (35%) compared to the control group (25%). This suggests the T allele may be associated with higher risk of Type 2 diabetes. But is this difference large enough to be real?
4
Step 4 — Check the p-value

The statistical test produces a p-value of 1.2 × 10⁻¹⁰. We compare this to the genome-wide significance threshold of 5 × 10⁻⁸. Since 1.2 × 10⁻¹⁰ is much smaller than 5 × 10⁻⁸, this SNP passes the threshold.

This SNP is genome-wide significant! The association between rs7903146 (T allele) and Type 2 diabetes is very unlikely to be due to chance.
5
Step 5 — Interpret the Odds Ratio

The odds ratio is reported as OR = 1.37. This means that individuals carrying the T allele have about 37% higher odds of developing Type 2 diabetes compared to those carrying the C allele. This is a modest effect — one SNP does not doom you to getting the disease, but it slightly increases risk.

Final interpretation: SNP rs7903146 is significantly associated with Type 2 diabetes (p = 1.2 × 10⁻¹⁰, OR = 1.37). This SNP is located near the TCF7L2 gene, which is now one of the most well-studied diabetes risk genes.

SECTION 7

Strengths and Limitations of GWAS

Like every scientific tool, GWAS has important strengths that make it valuable and limitations that researchers need to keep in mind. Understanding both sides helps you think critically about genetic research in the news.

Strengths	Limitations
Scans the entire genome — no need to guess which genes to check	Finds associations, not causes — a significant SNP might not be the actual disease-causing variant
Can discover completely unexpected genes linked to disease	Most GWAS hits have small effect sizes, explaining only a tiny fraction of disease risk
Works well for common diseases like diabetes, heart disease, and asthma	Requires very large sample sizes (often tens of thousands of participants) to detect small effects
Results can guide drug development by identifying biological pathways	Primarily detects common variants; rare but important mutations may be missed
Hypothesis-free — scientists do not need prior knowledge about the biology	Many studies have lacked diversity — most early GWAS used participants of European ancestry, limiting generalizability

Comparing the strengths and limitations of GWAS

✦

KEY TAKEAWAY

Think of GWAS like a metal detector at the beach. It can tell you where to dig because something interesting is buried there, but it cannot tell you exactly what the object is — you still need to dig it up and examine it. GWAS points scientists to regions of the genome that deserve closer study, but further experiments are needed to understand the biology.

ℹ

The Missing Heritability Problem

For many traits, GWAS-identified SNPs only explain a small portion of the trait's heritability. For example, height is about 80% heritable, but the hundreds of SNPs found by GWAS together explain less than half of that. The rest of the genetic influence — called missing heritability — may come from rare variants, gene-gene interactions, or other factors that standard GWAS cannot easily detect.

SECTION 8

From GWAS to Advanced Genomics

GWAS was a groundbreaking first step, but modern genomics has developed more powerful techniques that build on GWAS discoveries. Understanding how GWAS connects to these advanced methods gives you a sense of where the field is heading.

Feature	Standard GWAS	Advanced Approaches
Variants detected	Common SNPs (frequency > 1%)	Whole-genome sequencing finds rare variants too
Sample size	Thousands to hundreds of thousands	Biobanks now include millions (e.g., UK Biobank)
Prediction	Identifies individual risk SNPs	Polygenic risk scores (PRS) combine thousands of SNPs into one risk number
Causation	Shows association only	Mendelian randomization uses GWAS data to test causal relationships
Diversity	Historically European-focused	New efforts actively include diverse populations worldwide

How advanced approaches extend GWAS capabilities

One of the most exciting developments is the polygenic risk score (PRS). Instead of looking at one SNP at a time, a PRS adds up the effects of thousands of SNPs to give a single number estimating your overall genetic risk for a disease. Imagine getting a "genetic weather forecast" that tells you whether your genes make it slightly more or less likely that you will develop a particular condition. While PRS is not yet accurate enough for most clinical decisions, it is an active area of research that could shape personalized medicine — tailoring healthcare to your individual genome.

💡

Looking Ahead

As more diverse populations are included in genomic studies and technology costs continue to drop, GWAS and its descendants will become even more powerful. Future courses in genomics will explore techniques like whole-genome sequencing, CRISPR functional studies, and multi-omics integration that turn GWAS "hits" into actual medical treatments.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

In your own words, explain what a GWAS does and why it compares a case group to a control group. Why is it important to have both groups?

Show answer

A GWAS scans across the entire genome to find SNPs (single nucleotide polymorphisms) that are more common in people with a disease (case group) than in people without it (control group). Both groups are needed because without a comparison, you would not know whether a particular allele frequency is unusual. The control group acts as a baseline, showing what 'normal' allele frequencies look like. Any SNP that is significantly more common in cases than controls becomes a candidate for association with the disease.

Question 2: BASIC CALCULATION

In a GWAS study, 8,000 people are in the case group and each person contributes 2 alleles at a given SNP. If the risk allele (A) is found 4,800 times in the case group, what is the allele frequency of A in the case group?

Show answer

Total alleles in the case group = 2 × 8,000 = 16,000. The allele frequency of A = 4,800 ÷ 16,000 = 0.30, or 30%. This means 30% of the chromosomes sampled from the case group carry the A allele. This frequency would then be compared to the allele frequency in the control group to see if there is a meaningful difference.

Question 3: INTERMEDIATE

A researcher tests 800,000 SNPs in a GWAS. One SNP produces a p-value of 3 × 10⁻⁶. Does this SNP meet the genome-wide significance threshold? Explain why the threshold is set so low.

Show answer

No, this SNP does NOT meet genome-wide significance. The threshold is p < 5 × 10⁻⁸, and 3 × 10⁻⁶ is larger than 5 × 10⁻⁸ (since 10⁻⁶ is 100 times bigger than 10⁻⁸). The threshold is set so low because of the multiple testing problem: when you test 800,000 SNPs, some will appear significant purely by chance. The Bonferroni correction divides the standard significance level (0.05) by the number of tests (approximately 1 million for common GWAS), yielding 5 × 10⁻⁸. This prevents false positives.

Question 4: APPLIED

A news headline reads: 'Scientists Find Gene for Obesity!' The article describes a GWAS that found a SNP with an odds ratio of 1.15 and p = 2 × 10⁻¹². As a scientifically informed reader, how would you evaluate this headline? Discuss at least two reasons why the headline might be misleading.

Show answer

The headline is misleading for at least two reasons. First, GWAS finds associations, not causation — the SNP might not directly cause obesity but could be near a gene that does, or it could be in linkage disequilibrium with the true causal variant. The headline should say 'linked to' or 'associated with' rather than 'gene for.' Second, the odds ratio of 1.15 is very small, meaning this SNP increases obesity risk by only about 15%. Obesity is a complex trait influenced by hundreds of genetic variants plus diet, exercise, and environment. No single SNP qualifies as 'the gene for obesity.' Additionally, the study might have been conducted in one population, so the result may not apply equally to all ethnic groups.

Question 5: CRITICAL THINKING

Most early GWAS studies were conducted with participants of European ancestry. Explain why this is a problem and predict what might happen when the same SNPs are studied in populations from different parts of the world. Consider concepts like allele frequency, linkage disequilibrium, and effect size in your answer.

Show answer

Conducting GWAS mainly in European populations creates several problems. First, allele frequencies differ across populations due to different evolutionary histories, migration, and genetic drift. A SNP that is common in Europeans might be rare or absent in African or East Asian populations, meaning risk predictions based on European GWAS could be inaccurate for other groups. Second, linkage disequilibrium patterns vary across populations — LD blocks tend to be shorter in African populations because of greater genetic diversity. This means the 'tag SNP' identified in a European GWAS may not be in LD with the causal variant in other populations, so the association may not replicate. Third, effect sizes could differ across populations due to gene-environment interactions. The prediction is that some GWAS findings will replicate across populations (suggesting a biologically real effect), while others will not. This is exactly what researchers have observed. Expanding GWAS to diverse populations is essential for equitable and accurate genomic medicine.

SUMMARY

Lesson Summary

Summary

A genome-wide association study (GWAS) is a powerful method for scanning the entire genome to find SNPs (single nucleotide polymorphisms) associated with diseases or traits. The study compares allele frequencies between a case group and a control group, using strict statistical thresholds (p < 5 × 10⁻⁸) to prevent false positives. Results are displayed on a Manhattan plot, where tall peaks indicate SNPs with strong associations.

Key concepts include linkage disequilibrium (nearby SNPs inherited together), odds ratios (measuring effect size), and the critical distinction that GWAS finds associations, not causes. While GWAS has limitations — including small effect sizes and historical lack of population diversity — it remains the foundation of modern genomics and has paved the way for advanced tools like polygenic risk scores and personalized medicine.