Single nucleotide polymorphisms, commonly called SNPs (pronounced “snips”), are the most abundant form of genetic variation in the human genome, numbering in the millions across any individual’s DNA. Each SNP is a spot where a single letter of the genetic code differs between people. Most do nothing noticeable, but collectively they influence everything from disease risk and drug metabolism to physical traits and evolutionary adaptation. Because SNPs are stable, common, and spread across the entire genome, they have become the workhorse marker of modern genetics, underpinning research from ancestry testing to cancer genomics.
What a SNP Actually Is
Your genome is a string of roughly three billion chemical “letters” (nucleotides), and at certain positions the letter that appears varies from person to person. If at least one percent of the population carries a different letter at a given position, that site is classified as a common SNP. Below that frequency threshold, the variant is typically called a rare variant, though the underlying biology is the same: one nucleotide swapped for another. The human genome holds somewhere around four to five million SNPs per individual when compared to a reference sequence, and catalogs like dbSNP have indexed well over a hundred million across all sampled populations.
SNPs can sit inside genes, between genes, or in regulatory regions that control when and how much a gene is active. A large-scale validation study found that a curated set of confirmed SNPs fell within ten kilobases of about two-thirds of all known and predicted genes, making them useful signposts for linking stretches of DNA to biological function and disease risk.1PubMed Central. Large-scale validation of single nucleotide polymorphisms in gene regions That density is why researchers can scan millions of SNPs in a single experiment and still get a reasonable picture of what’s going on across the genome.
When a Single Letter Change Alters a Protein
A SNP inside the protein-coding part of a gene can swap one amino acid for another in the resulting protein, a change known as a non-synonymous SNP. Sometimes the substitution is harmless because the new amino acid behaves similarly to the old one. Other times it destabilizes the protein, changes its shape, or wrecks its ability to do its job. Computational tools now predict these effects by analyzing how conserved a position is across species and how the amino acid change alters protein structure.2PubMed Central. Human non-synonymous SNPs: server and survey
A recent computational study of the human lipocalin-2 gene, which plays a role in immune defense, illustrates how this works in practice. Researchers identified six non-synonymous SNPs predicted to be harmful, and stability-prediction tools showed that most of them decreased the protein’s structural stability. All six fell in evolutionarily conserved regions, a strong hint that the affected amino acid positions are functionally important and that mutations there carry real consequences.3Egyptian Journal of Medical Human Genetics. Computational analysis of non-synonymous SNPs in the human LCN2 gene
The Bigger Story Lives Outside Genes
Most SNPs flagged in disease studies do not sit inside protein-coding sequences. They land in the vast non-coding stretches of the genome, particularly in regulatory elements like enhancers and promoters, where they can dial gene activity up or down without changing the protein itself. Researchers use a technique called expression quantitative trait locus (eQTL) mapping to connect these regulatory SNPs to the genes they influence. Large eQTL studies have revealed that gene regulation is highly tissue-specific: a SNP that tweaks gene expression in liver cells may do nothing in the brain, and vice versa.4Oxford Academic. Exploring the function of genetic variants in the non-coding genomic regions: approaches for identifying human regulatory variants affecting gene expression
One striking example comes from Alzheimer’s disease research. A study found that nearly 30 percent of non-coding Alzheimer’s-associated SNPs sit in enhancer regions, and the genes they regulate are involved in clearing amyloid plaques, transmitting signals between neurons, and orchestrating immune responses in the brain. Most of these enhancer SNPs and their target genes were located within the same large-scale structural domains of chromosomes, suggesting the regulation happens through the physical looping of DNA bringing distant elements into contact.5PubMed Central. Enhancer variants associated with Alzheimer’s disease affect gene expression via chromatin looping In endothelial cells, epigenetic profiling uncovered more than 3,000 regulatory elements whose activity is shaped by genetic variants, many of which disrupted binding sites for key transcription factors.6American Journal of Human Genetics. Regulatory Variants in Endothelial Cells Underlying Complex Disease Traits
SNPs can also alter patterns of DNA methylation, one of the chemical tags that cells use to silence or activate genes. A study of smoking-related methylation changes found that 70 out of 151 previously reported smoking-linked methylation sites were significantly influenced by nearby SNPs, with the genetic variants explaining between a fraction of a percent and roughly 19 percent of the methylation change at individual sites.7PubMed Central. The impact of methylation quantitative trait loci (mQTLs) on active smoking-related DNA methylation changes In other words, your genetic background can amplify or buffer the molecular marks left by environmental exposures like tobacco.
Genome-Wide Association Studies and the Search for Disease Variants
The technology that transformed SNP research is the genome-wide association study, or GWAS. In a typical GWAS, researchers genotype hundreds of thousands of SNPs across a large group of people, then compare allele frequencies between those with a disease and those without. Any SNP whose frequency differs more than expected by chance gets flagged as associated with the trait. Since the first large GWAS appeared in the mid-2000s, thousands of studies have linked tens of thousands of SNPs to conditions ranging from heart disease and diabetes to schizophrenia and autoimmune disorders.
One challenge is that SNPs in close physical proximity tend to be inherited together in blocks, so the SNP that lights up in a GWAS is often just a proxy for the true causal variant sitting nearby. Haplotype block methods help deal with this by grouping correlated SNPs and testing blocks rather than individual markers, which can improve the ability to detect small effects and reduce false leads.8PubMed Central. Haplotype block partitioning as a tool for dimensionality reduction in SNP association studies
Missing Heritability and Rare Variants
For most complex traits, the common SNPs identified by GWAS explain only a fraction of the heritability estimated from family studies. For height, for instance, standard GWAS chips capture roughly a third to two-thirds of the heritable variation. A whole-genome sequencing study of over 25,000 people estimated the total heritability of height at about 0.68, with rare, low-frequency variants that are poorly tagged by standard chips accounting for a meaningful share of the gap. These rare variants were enriched in protein-altering positions and in genomic regions where they are not well-correlated with surrounding common SNPs, consistent with natural selection keeping harmful variants at low frequencies.9PubMed Central. Assessing the contribution of rare variants to complex trait heritability from whole genome sequence data The implication: as sequencing costs drop and sample sizes grow, a large chunk of the so-called “missing heritability” should come into view.
Polygenic Risk Scores and Their Limits
One of the most hyped applications of SNP data is the polygenic risk score, or PRS. The idea is straightforward: take all the GWAS-identified SNPs for a disease, weight each one by the size of its effect, and sum them up to get a single number representing an individual’s genetic risk. In principle, this lets you stratify people from low to high risk for conditions like heart disease, breast cancer, or type 2 diabetes.
In practice, the predictive power of current scores is limited. Research has shown that useful levels of individual prediction require training data from sample sizes an order of magnitude larger than what was available even recently, and that published scores have more statistical utility for confirming associations in research than for making clinical predictions about a given person.10PubMed Central. Power and Accuracy of Polygenic Scores As biobank-scale datasets continue to grow, this gap is narrowing, but we are not yet at the point where a PRS for most conditions can rival traditional risk factors like blood pressure or family history for individual-level prediction.
The Diversity Problem
The single biggest limitation of SNP-based medicine right now is that the underlying data come overwhelmingly from people of European ancestry. A review of GWAS for major non-communicable diseases found that about 91 percent of study participants were of European descent, with East Asians a distant second at roughly five percent.11Human Molecular Genetics. Ethnic, gender and other sociodemographic biases in genome-wide association studies for the most burdensome non-communicable diseases: 2005–2022 This matters because allele frequencies, linkage patterns, and the environmental context of gene-trait relationships all differ across populations. A polygenic risk score trained on European samples can be several times more accurate for European-descent individuals than for African, South Asian, or Indigenous populations.12PubMed Central. Clinical use of current polygenic risk scores may exacerbate health disparities
This is not a minor statistical wrinkle. If PRS-based screening were deployed clinically today, it would systematically benefit the populations that are already best served by healthcare and offer less value to those who are most underserved.13PubMed Central. Importance of Including Non-European Populations in Large Human Genetic Studies to Enhance Precision Medicine Encouragingly, even relatively modest-sized studies in underrepresented populations show gains in transferability, so the path forward is clear even if progress has been slow.
SNPs and Drug Response
Pharmacogenomics is one area where SNPs already influence clinical decisions. Variants in the cytochrome P450 enzyme family, which metabolizes a large share of prescription drugs, can make you a fast or slow metabolizer of specific medications. A survey of P450 variants found 34 alleles with at least a one percent frequency in the European population that affect enzyme activity. The CYP2D6 gene alone carried 11 such alleles, with the most common reaching frequencies above 20 percent. On the other end, the CYP3A5*3C allele, which decreases enzyme activity, was present in more than 80 percent of that population.14PLoS ONE. Polymorphic Cytochrome P450 Enzymes (CYPs) and Their Role in Personalized Therapy These frequencies vary substantially across ethnic groups, which is one reason the same drug dose can work well for one person and cause side effects in another.15PubMed. Human cytochrome P450 and personalized medicine
Beyond metabolism speed, certain genetic variants trigger dangerous immune-mediated drug reactions. Alleles in the human leukocyte antigen (HLA) system have been linked to severe hypersensitivity reactions to specific drugs.16PubMed Central. HLA-associated adverse drug reactions – scoping review The HLA-B*57:01 allele, for example, is strongly associated with hypersensitivity to the HIV drug abacavir. Pre-prescription genetic testing for this allele is now standard practice in many countries, a clear win for SNP-guided medicine. Allele prevalence varies by region: in Greater Middle East populations, HLA-B*57:01 ranges from about one to three percent, while HLA-DPB1*03:01, linked to aspirin-induced asthma, is found in 10 to 14 percent.17PubMed. HLA pharmacogenetic markers of drug hypersensitivity from the perspective of the populations of the Greater Middle East
SNPs as Windows into Human Evolution
Because SNPs accumulate over generations and their frequencies shift under natural selection and genetic drift, they serve as a historical record of human migration and adaptation. One of the most dramatic examples involves the EPAS1 gene, which helps regulate the body’s response to low oxygen. In Tibetan populations, one SNP in EPAS1 shows a frequency 78 percent higher than in Han Chinese lowlanders, the fastest allele frequency shift documented at any human gene. That SNP is associated with lower red blood cell counts at high altitude, helping Tibetans avoid the dangerously thickened blood that afflicts many lowlanders who move to high elevations.18PubMed Central. Sequencing of 50 human exomes reveals adaptation to high altitude Sherpas in Nepal carry the same pattern of EPAS1 alleles as Tibetans, clearly separating high-altitude populations from lowland groups worldwide.19PLoS ONE. Genetic Variants in EPAS1 Contribute to Adaptation to High-Altitude Hypoxia in Sherpas
Intriguingly, analysis of local ancestry in Tibetans suggests the adaptive variants at EPAS1 and another altitude gene, EGLN1, are enriched for ancestry tracing back to an archaic high-altitude population, meaning Tibetans may have acquired some of their hypoxia tolerance through ancient admixture rather than evolving it independently.20Nature Communications. Admixture facilitates genetic adaptations to high altitude in Tibet
SNPs also help quantify how much archaic DNA modern humans carry. A catalog of Neanderthal ancestry segments across more than 300 ancient and present-day genomes showed that the vast majority of Neanderthal gene flow traces to a single extended period of interbreeding roughly 50,500 to 43,500 years ago, and that most natural selection on those inherited Neanderthal variants, both positive and negative, happened rapidly afterward.21PubMed Central. Neanderthal ancestry through time: Insights from genomes of ancient and present-day humans Some ancient genomes from the Americas carry unexpected patterns, including greater Denisovan than Neanderthal ancestry in individuals from Uruguay and Panama, and a strong genetic signal linking an ancient Panamanian genome to Australasian populations, complicating simple models of New World migration.22PubMed Central. Genomic evidence for ancient human migration routes along South America’s Atlantic coast
SNPs in the Courtroom
Forensic scientists have begun using panels of SNPs to predict physical appearance and geographic ancestry from crime-scene DNA, an approach called forensic DNA phenotyping. A review of the field counted dozens of SNP panels developed for inferring biogeographical ancestry and externally visible characteristics like eye, hair, and skin color.23PubMed Central. Forensic DNA phenotyping: a review on SNP panels, genotyping techniques, and prediction models The accuracy of these predictions varies by trait and color category. For eye color, validated models achieve area-under-the-curve values between 0.74 and 0.99; for hair color the range is 0.64 to 0.94; and for skin color, 0.72 to 0.99.24PubMed Central. The Use of Forensic DNA Phenotyping in Predicting Appearance and Biogeographic Ancestry Predicting blue versus brown eyes, for example, is far more reliable than distinguishing blond from red hair. These tools are used in real investigations in several countries, though the ethical and legal frameworks around DNA phenotyping remain actively debated.
How Reliable Are Consumer SNP Chips
Millions of people have spit into a tube for direct-to-consumer genetic testing, which relies on SNP genotyping arrays (chips) rather than full genome sequencing. For common variants, these chips are generally reliable. But their performance falls off sharply for very rare variants, the kind most likely to signal a serious medical condition. An analysis using UK Biobank sequencing data as ground truth found that for very rare variants, only about 16 percent of heterozygous calls on one widely used chip were confirmed by sequencing, and only about nine percent on another chip.25bioRxiv. Assessing the analytical validity of SNP-chips for detecting very rare pathogenic variants: implications for direct-to-consumer genetic testing The practical takeaway: if a consumer test flags you as carrying a rare pathogenic variant, that result absolutely needs confirmation through clinical-grade sequencing before you or your doctor act on it.
Predicting Which Variants Are Harmful
With millions of SNPs in any genome and most of them poorly characterized, computational tools that predict whether a given variant is likely to cause disease are increasingly important. Machine-learning approaches trained on databases of known pathogenic and benign variants, such as the ClinVar repository, now handle multiple variant types (not just single-letter swaps but also small insertions and deletions) and show strong performance on independent test sets.26PubMed Central. MAGPIE: accurate pathogenic prediction for multiple variant types using machine learning approach These scores are not definitive diagnoses, but they help clinical geneticists prioritize which variants in a patient’s genome deserve closer attention, filtering thousands of candidates down to a manageable shortlist.
Editing Single-Letter Mutations
If a disease is caused by a single SNP, the logical question is whether you can fix it. CRISPR-based gene editing has been moving in that direction. Base editors can chemically convert one DNA letter to another at a specified position without cutting both strands of the double helix, reducing the risk of unwanted insertions or deletions. Researchers have demonstrated that engineering the connection between the deaminase and Cas domains of these editors can achieve high precision for targeted single-nucleotide changes.27Nature Communications. Engineering of high-precision base editors for site-specific single nucleotide replacement
A persistent challenge has been specificity: how do you edit the disease-causing allele without also modifying the normal copy sitting on the other chromosome? A recent approach addressed this by designing guide RNAs with intentional mismatches that selectively reduce binding to the normal sequence while still engaging the mutant one. The researchers demonstrated the strategy on a cancer-specific single-nucleotide mutation in the TERT promoter, a driver mutation common in glioblastomas, successfully editing the mutant allele while leaving the normal sequence intact.28bioRxiv. High-fidelity CRISPR genome editing of single-nucleotide mutation with near-complementary guide RNA via enhanced target binding kinetics This kind of allele-specific editing is still in early development, but the ability to discriminate between two sequences differing by a single letter is a meaningful step toward SNP-level precision in gene therapy.
SNPs in Agriculture
The same principles that make SNPs useful in human medicine apply to crop and livestock breeding. Genomic selection uses dense SNP data to predict the breeding value of individual plants or animals before they mature, dramatically shortening the breeding cycle. Over the past two decades, this approach has significantly accelerated genetic gains in plant breeding programs for traits like yield, drought tolerance, and disease resistance.29PubMed. Genomic selection in plant breeding: Key factors shaping two decades of progress Livestock breeders adopted it first, partly because individual animals are worth more and the reduction in generation time is larger, but large-scale plant breeding programs are catching up as genotyping costs drop and statistical models improve their handling of gene-by-environment interactions.30Plant Communications. Enhancing Crop Genetic Gain in the Era of Genomic Selection In a world pressing for higher food production under climate stress, the ability to scan a seedling’s SNP profile and predict its adult performance months or years early is not trivial.