aiwiki.page
English
Science / Biology / genotyping

Genotyping

Genotyping determines genetic variants in biological samples using molecular assays, sequencing, and statistical inference for research, identification, breeding, and medicine.

37 keywords1 linked from7 not yet writtenWritten by AI
GenotypeDNAGenomeAllelePhenotypeBacteriaSingle-Nucleotid…Short tandem rep…Genotyping

Genotyping is the process of determining which genetic variants are present in an organism, individual, tissue, or other biological sample. It characterizes a genotype at one or more positions in DNA, ranging from a single clinically important variant to millions of sites distributed across a genome. The term encompasses both laboratory measurements and the computational procedures that convert those measurements into genetic assignments. Although genotyping often tests variants already known to exist, sequencing-based methods can discover previously unrecognized variants and determine their genotypes in the same analysis.

Genotyping connects molecular variation with inheritance, biological function, identity, and population history. Its applications include mapping inherited disease, investigating complex traits, identifying individuals and biological relationships, selecting crop and livestock breeding material, and characterizing infectious organisms. The meaning and reliability of a result depend on the variants examined, the biological properties of the sample, the measurement technology, and the statistical assumptions used to interpret the data. A genotype is therefore not simply a raw instrument reading, nor does it by itself establish an organism’s observable characteristics.

Biological meaning and scope

A genotype describes the genetic state at a defined locus or set of loci. In a typical diploid organism, most autosomal loci occur in two copies, one inherited from each parent. If a locus has two alternative alleles, conventionally called A and G, its possible diploid genotypes are AA, AG, and GG. AA and GG are homozygous, whereas AG is heterozygous. These labels describe molecular composition rather than biological superiority, normality, or disease status. The relationship between genotype and phenotype depends on gene function, environmental conditions, developmental history, and interactions with other genetic variants.

The two-copy model is not universal. Many bacteria are effectively haploid for their main chromosome, while numerous crop plants exhibit polyploidy, with more than two homologous chromosome sets. Sex chromosomes, organellar genomes, and regions affected by deletions or duplications require different copy-number assumptions. In a tetraploid organism, a biallelic locus can have five allele-dosage states, from zero to four copies of one allele. A method that distinguishes AA, AG, and GG in diploid samples may not reliably distinguish the corresponding dosage states in a polyploid.

Genotyping can target several classes of variation. A single-nucleotide polymorphism is a single-base variant observed within a population; single-nucleotide variant is the broader term, including rare changes. Small insertions and deletions alter a short stretch of sequence. Short tandem repeats differ in the number of adjacent repeat units, such as repeated copies of a four-base motif. Copy-number variation changes the number of copies of a genomic segment, while larger rearrangements can alter segment orientation or chromosomal position. Each class creates different measurement and interpretation problems.

A sample need not contain one uniform genotype. Mosaicism results when genetically distinct cell populations occur within an individual. Tumours may contain several evolving clones, while blood collected after a bone-marrow transplant can contain donor-derived genetic material. Microbial cultures, environmental samples, and forensic specimens may contain multiple organisms or contributors. In these settings, an observed allele fraction is not automatically equivalent to a conventional diploid genotype.

Genotyping also differs from related molecular measurements. Gene expression measures the production or abundance of RNA or other gene products, not simply inherited sequence state. Epigenetic assays examine features such as DNA methylation or chromatin organization. Sequencing can support all these activities, but the instrument does not determine the conceptual category: the biological question and analytical output do. In a genotyping experiment, the central output is an assignment of alleles, copy states, or genotype probabilities at specified genetic locations.

Historical development and marker selection

Genotyping developed from the study of inherited differences before direct DNA measurement was practical. Blood-group antigens and electrophoretically distinguishable protein variants served as genetic markers during much of the twentieth century. Such markers allowed investigators to follow inheritance, estimate population variation, and test biological relationships, but represented only a small and functionally biased fraction of the genome. DNA-based markers greatly expanded the number of accessible loci and made it possible to study variation outside protein-coding regions.

An important early DNA method was restriction fragment length polymorphism analysis. Restriction enzymes cut DNA at particular sequences; variants that create or remove a recognition site can alter the lengths of resulting fragments. Fragments were separated and detected using sequence-specific probes. In 1980, David Botstein and colleagues described how a sufficiently dense map of such markers could support human genetic mapping. In 1985, Alec Jeffreys and colleagues reported highly variable DNA regions that became foundational to DNA fingerprinting.

The development of the polymerase chain reaction during the 1980s transformed genotyping by allowing selected DNA segments to be amplified from small starting quantities. PCR-based microsatellite assays subsequently became major tools in linkage mapping and forensic identification. During the 1990s and 2000s, high-density SNP assays and arrays made it practical to genotype hundreds of thousands, and later millions, of positions in large cohorts. Massively parallel sequencing then extended genotyping to broader sequence contexts and enabled simultaneous discovery and measurement of variation.

Marker selection depends on the purpose of the study. A highly variable repeat can be informative for distinguishing individuals because many alleles may occur at one locus. A biallelic SNP generally provides less identification information per locus, but SNPs are abundant and can be assayed at large scale. A disease-focused test may examine one established pathogenic variant, whereas a population survey may require markers distributed widely across chromosomes.

Some markers are chosen because they correlate with nearby variants through linkage disequilibrium, the nonrandom association of alleles at different loci. A selected marker can therefore provide information about an unmeasured region without being biologically responsible for an associated trait. However, correlations differ among populations and change with demographic and recombination history. A marker panel informative in one population may capture substantially less variation in another.

The variants represented in a panel also reflect how they were discovered. Arrays designed using common variants from a restricted set of populations may underrepresent rare or geographically localized variation. This ascertainment bias affects comparisons of genetic diversity, demographic reconstruction, and association analysis. Marker choice is consequently part of the scientific design, not merely a preliminary purchasing decision.

Targeted molecular assays

Targeted genotyping assays interrogate specified loci rather than surveying all available sequence. They are particularly useful when the relevant variants are known in advance, sample throughput is high, or a small panel provides sufficient information. Their molecular specificity usually depends on differences in primer binding, probe hybridization, enzyme activity, or fragment length. Each strategy translates a sequence difference into a measurable signal.

Allele-specific PCR exploits the sensitivity of DNA synthesis to the match between a primer and its template, especially near the primer’s extending end. Separate reactions or distinguishable products indicate which allele is present. Assay performance depends on primer design, reaction conditions, and the surrounding sequence. An additional variant beneath a primer can reduce amplification of one chromosome, producing allelic dropout and potentially making a heterozygous sample appear homozygous.

Probe-based assays often use fluorescent reporters. In a common format, two probes recognize alternative alleles at the same site and carry different fluorescent labels. During amplification, probe cleavage separates a reporter from a quencher, generating a signal. Samples can then be classified according to the relative intensities of the two reporters. Strong signal from one reporter suggests one homozygous state, strong signal from the other suggests the alternative state, and signal from both supports heterozygosity. Actual classification requires controls and validated separation of the resulting signal clusters.

Restriction-based assays amplify a region and digest the product with an enzyme whose recognition site differs between alleles. Gel electrophoresis or another fragment-sizing method reveals the resulting pattern. These assays can be inexpensive and transparent, but incomplete digestion can resemble a different genotype. A nearby sequence change may also modify the expected pattern. Consequently, the relationship between fragment sizes and genotype must be established for the particular locus.

Repeat genotyping commonly uses fluorescently labelled PCR products separated by capillary electrophoresis. Fragment length is converted into an allele designation using a size standard and, in forensic applications, an allelic ladder. PCR can generate stutter products, usually shorter by one repeat unit, that complicate interpretation of minor alleles and mixed samples. Very long repeats may amplify poorly, and equal fragment lengths do not always imply identical underlying sequences.

Other targeted approaches include single-base extension followed by mass spectrometry, ligation-based assays, high-resolution melting analysis, and digital PCR. Melting analysis detects differences in the thermal behaviour of amplified DNA, although identifying the exact variant may require additional testing. Digital PCR partitions a sample into many small reactions and can quantify allele proportions under an appropriate statistical model. No targeted method automatically detects every variant in its target region: most answer a specifically designed molecular question.

High-density genotyping arrays

A DNA microarray measures interactions between sample DNA and many immobilized probes. Genotyping arrays use these interactions, often combined with enzymatic extension reactions, to distinguish alleles at predetermined positions. Thousands to millions of markers can be processed in parallel. Their major advantage is economical, reproducible measurement of a fixed set of variants across large numbers of samples, making them important in population studies, biobanks, and breeding programmes.

Although laboratory chemistries differ, a typical workflow includes DNA preparation, amplification or other processing, fragmentation, hybridization, and fluorescent detection. At each marker, the assay produces intensity measurements related to the alternative alleles. A genotype-calling algorithm groups samples into clusters corresponding to the expected genotype states. For a well-behaved diploid biallelic marker, three clusters represent the two homozygotes and the heterozygote.

These clusters are empirical distributions rather than perfectly separated categories. Weak signals, contamination, nearby sequence variants, and technical differences between processing batches can shift their position or increase overlap. Rare alleles create a particular challenge because few samples may define the heterozygous cluster and the rarer homozygous cluster may be absent altogether. An algorithm can assign a confident-looking call even when the biological cluster structure is poorly supported.

Array intensity data can also reveal genomic dosage. Two widely used summaries are B-allele frequency, which represents relative allelic signal, and a normalized total-intensity measure often expressed as a log R ratio. A deletion may reduce total intensity and remove the usual heterozygous band; a duplication can produce additional allele-balance patterns. Such evidence can support detection of copy-number changes, chromosome abnormalities, or mosaic alterations, but sensitivity depends on marker density, segment length, cell fraction, and assay noise.

Because arrays test selected sites, they do not constitute comprehensive sequencing. A sample classified as carrying the reference genotype at all tested positions can still contain untested variants nearby. Array probes may also behave unexpectedly when a deletion removes their binding region or when closely related genomic sequences cross-hybridize. Marker annotations require careful handling because chromosome coordinates, reference assemblies, and strand conventions can differ between datasets.

Arrays remain valuable even where sequencing is available because their cost and data-processing requirements can be favourable for very large cohorts. Their usefulness is often extended by statistical inference of untyped variants. Nevertheless, a dense array does not measure every inferred variant directly, and the distinction between observed and computationally estimated genotypes is essential when interpreting results.

Sequencing-based genotyping

DNA sequencing determines nucleotide order and can therefore identify both known and previously unrecognized variation. Genotyping is a subsequent interpretive step: sequence observations are evaluated to determine which alleles are present and in what copy state. Targeted sequencing, exome sequencing, whole-genome sequencing, and reduced-representation approaches differ in genomic coverage, cost, and the classes of variants they can measure effectively.

In a common short-read workflow, DNA is fragmented, adapters are attached, and many fragments are sequenced in parallel. Reads are aligned to a reference genome, after which software identifies positions or regions where the observed data differ from that reference. The procedure must distinguish real biological variation from sequencing errors, alignment mistakes, and laboratory artifacts. Some variant callers reconstruct local sequence haplotypes rather than considering each position independently, improving interpretation near insertions and deletions.

Read depth is the number of reads covering a position, but depth alone does not determine reliability. Base-quality estimates, mapping quality, strand representation, read placement, and the presence of repeated sequences all matter. A mean genome-wide depth of 30 reads does not imply that every base has 30 independent, high-quality observations. Coverage is uneven, and some genomic regions remain difficult to map or amplify.

For a straightforward diploid heterozygous site, approximately half the informative reads are expected to support each allele. Random sampling, reference bias, contamination, copy-number changes, and biological mosaicism can alter this balance. In cancer, the observed variant allele fraction also depends on tumour purity and the copy numbers and clonal proportions within the specimen. It cannot generally be interpreted as a direct measure of the fraction of cells carrying a mutation without additional assumptions.

Short reads have difficulty resolving some repeat expansions, highly similar duplicated genes, and complex rearrangements. Long-read sequencing can span larger repeats and connect variants along individual DNA molecules, improving structural interpretation and phasing. Its performance still depends on platform characteristics, coverage, error correction, and analytical methods. Some loci require specialized analysis rather than a general-purpose variant caller.

Reduced-representation sequencing, including methods often called genotyping-by-sequencing, samples reproducible subsets of a genome, frequently using restriction enzymes to select fragments. It is widely used in ecological and agricultural research, particularly where a mature array resource is unavailable. Variable coverage and missing data are important limitations. At low depth, likelihood-based analysis may preserve uncertainty more accurately than forcing every sample into a definite genotype category.

Genotype inference, phasing, and imputation

Genotypes are inferred from evidence rather than observed as abstract labels. A sequencing experiment produces reads, and an array produces intensities. A statistical model then evaluates how compatible those measurements are with alternative genetic states. For observed data DD and a possible genotype GG, Bayesian inference expresses this relationship as

P(G∣D)∝P(D∣G)P(G).P(G\mid D)\propto P(D\mid G)P(G).

Here, P(D∣G)P(D\mid G) is the genotype likelihood and P(G)P(G) is a prior probability. The posterior probability P(G∣D)P(G\mid D) combines the two. Distinguishing likelihoods from posteriors matters because population assumptions can influence the latter, especially when the measurement evidence is weak.

For a biallelic diploid locus, the possible counts of an alternative allele are 0, 1, and 2. Instead of reporting only the most probable state, an analysis can retain all three probabilities. The expected allele dosage is

E[G∣D]=P(G=1∣D)+2P(G=2∣D).E[G\mid D]=P(G=1\mid D)+2P(G=2\mid D).

A dosage of 0.8 is not a literal fractional chromosome count. It represents uncertainty averaged over possible integer genotypes. Association studies often use dosages because converting uncertain data into hard calls discards information and can introduce avoidable error.

Phasing determines which alleles lie together on the same chromosome, forming a haplotype. A person heterozygous at two sites may carry both alternative alleles on one chromosome or one on each chromosome. Ordinary unphased genotypes do not distinguish these arrangements. Family inheritance, reads spanning multiple sites, and population haplotype patterns can provide phase information. Phasing is particularly important when interpreting whether two variants affecting the same gene occur on opposite copies or together on one copy.

Genotype imputation estimates unmeasured genotypes using observed markers and a reference panel of haplotypes. The method identifies reference haplotypes that resemble portions of the study sample and predicts alleles at positions not directly measured. It greatly increases the number of variants available for association testing and permits partial harmonization of studies that used different arrays.

Imputation accuracy depends on the density and quality of measured markers, reference-panel size, ancestry representation, allele frequency, and local haplotype structure. Rare variants absent from the reference cannot be recovered merely by increasing computational effort. A high-quality common-variant imputation result also does not establish equivalent performance for every rare or structurally complex variant.

Population-based inference relies on patterns shaped by meiosis, recombination, demographic history, and relatedness. Models can be inappropriate when applied without adjustment to polyploid organisms, strongly inbred samples, admixed populations, or mixtures. The appropriate output may therefore be a genotype probability, dosage distribution, phased sequence, or explicit no-call rather than a single categorical answer.

Quality control and analytical reliability

Reliable genotyping requires both specimen-level and variant-level quality control. Sample collection, storage, extraction, and identification affect all subsequent steps. Degraded DNA, chemical inhibitors, contamination, and sample swaps can create errors even when the genotyping chemistry itself performs as designed. Chain-of-custody requirements are particularly important where identity, legal evidence, or longitudinal sample tracking is involved.

Common sample metrics include call rate, heterozygosity, relatedness, contamination estimates, and agreement with expected chromosome patterns. Call rate is the proportion of attempted loci receiving an accepted genotype, but a high call rate does not guarantee correctness. Excess heterozygosity can indicate contamination or technical artifacts; unusually low heterozygosity can reflect inbreeding, biological population structure, or assay failure. Interpretation therefore requires biological context rather than a universal numerical cutoff.

Variant-level checks examine missingness, signal-cluster quality, allele balance, read support, and consistency between replicate measurements. In family data, Mendelian inconsistencies can reveal errors: for example, a child’s called genotype may be incompatible with the parents’ calls under ordinary inheritance. Such discrepancies can also reflect a new mutation, a structural variant, an incorrect relationship assumption, or an unusual biological event, so they are not automatically evidence that the child’s call is wrong.

Departure from Hardy–Weinberg equilibrium is often used as a diagnostic statistic. If two alleles have frequencies pp and q=1−pq=1-p, the equilibrium genotype proportions are

p2,2pq,q2.p^2,\qquad 2pq,\qquad q^2.

Large departures can indicate genotyping problems, but they can also arise from population mixture, inbreeding, selection, or study ascertainment. Testing a pooled sample containing genetically differentiated populations can itself create a heterozygote deficit. Disease-associated loci in affected samples may also depart from equilibrium for biological reasons.

Batch effects occur when processing differences align with study groups. If cases and controls are genotyped in different laboratories, on different platforms, or during different periods, systematic measurement differences can mimic genetic associations. Randomized sample placement, technical replicates, shared controls, and careful harmonization help distinguish biological effects from laboratory structure. Confirmation by a genuinely different measurement method can be especially informative because repeating the same chemistry may reproduce the same systematic error.

Reliability is also representation-dependent. A result requires a defined genomic coordinate system, reference and alternative alleles, ploidy assumptions, and an explicit treatment of missing or uncertain data. Equivalent insertions or deletions can sometimes be written in different ways within repeated sequence, complicating comparisons. Overall concordance percentages can conceal poor performance on rare heterozygotes or difficult variant classes; evaluation by genotype and variant type is therefore more informative than one aggregate accuracy figure.

Research, breeding, and population applications

Genotyping provides the observations used to connect genetic variation with inheritance and traits. In family-based mapping, markers are followed through pedigrees to locate chromosomal regions that co-segregate with a phenotype. This strategy was central to identifying many genes underlying Mendelian disorders. Its resolution depends on informative family structure and recombination events, rather than simply on the number of genotyped markers.

In a genome-wide association study, researchers test large numbers of variants for statistical relationships with a trait. The approach became widely established in human genetics during the mid-2000s as dense arrays and large cohorts became available. Because many tests are performed, stringent statistical thresholds and independent replication are important. An associated marker often tags a region through linkage disequilibrium; it does not by itself identify the causal variant, the affected gene, or the biological mechanism.

Population structure is a major analytical consideration. If ancestry is correlated with both allele frequencies and the measured trait, apparent associations can arise without a direct genetic effect. Statistical approaches account for ancestry patterns and relatedness, but study design and population representation remain fundamental. Combining genotyping with functional experiments, expression data, and molecular annotation helps distinguish statistical localization from biological explanation.

In population genetics, genotype data support estimates of genetic diversity, relatedness, migration, admixture, and demographic history. These analyses depend on assumptions about mutation, recombination, sampling, and evolutionary processes. Present-day genetic similarity does not map perfectly onto language, nationality, or social identity. Historical inference is strengthened when genetic evidence is interpreted alongside archaeological, geographical, and documentary information rather than treated as a complete record on its own.

Plant and animal breeding use genotyping for marker-assisted selection, pedigree verification, and genomic prediction. Marker-assisted selection tracks particular loci with substantial known effects. Genomic selection instead combines effects estimated across many markers to predict breeding values. In 2001, Theo Meuwissen, Ben Hayes, and Mike Goddard described a foundational framework for predicting total genetic value using dense genome-wide markers. Prediction performance depends on the training population, trait architecture, environment, and relationship between training and selection candidates.

Genotyping also supports conservation and organism identification. It can estimate relatedness in captive populations, detect hybridization, distinguish stocks, and measure loss of genetic diversity. In microbiology, sequence-based typing distinguishes strains and helps investigate transmission. Closely matching microbial genotypes provide evidence of relatedness but do not independently establish a direct transmission event, its direction, or the precise route by which exposure occurred.

Clinical testing and biological identification

Clinical genotyping examines variants relevant to diagnosis, inherited risk, treatment response, or compatibility between donors and recipients. A targeted assay may test a small set of established variants, whereas sequencing may evaluate many genes or an entire genome. Analytical validity—the ability to measure the intended genotype accurately—is distinct from clinical validity, the relationship between that genotype and a health-related outcome. A technically correct result may still have limited clinical meaning.

Interpretation depends on the disease mechanism and inheritance pattern. The same type of sequence change can have different consequences in different genes, and two variants in one individual may need to be phased to establish their relationship. Some genetic disorders involve repeat expansions, large rearrangements, or dosage changes that are poorly detected by a test optimized for single-base substitutions. A negative result therefore applies to the variants and mechanisms the assay can assess, not automatically to every possible genetic explanation.

Pharmacogenetics studies how genetic variation affects responses to medicines. Variants in drug-metabolizing enzymes, transporters, and immune-related genes can be associated with altered drug exposure or adverse reactions. For example, CYP2D6 contains deletions, duplications, and hybrid structures that complicate allele assignment. A small SNP panel may not establish the full genotype, and translating an assigned genotype into a functional category requires a separate interpretation framework. Genotype alone does not capture all influences on drug response.

Typing of the major histocompatibility complex, particularly human leukocyte antigen genes, is important in transplantation. These loci are exceptionally polymorphic, and closely related alleles can differ at clinically significant positions. Typing resolution varies: an assay may distinguish broad allele groups or identify much more specific sequence differences. The distinction between an unambiguous allele assignment and a set of possible alleles is therefore consequential.

DNA profiling uses selected genetic markers to compare biological samples or assess relationships. Autosomal short tandem repeats remain widely used in forensic systems because they are highly variable and support standardized comparison. Sex-chromosome markers and mitochondrial sequences can add information in particular circumstances, though they follow different inheritance patterns and generally provide different levels of individual discrimination.

Forensic interpretation evaluates competing propositions rather than treating a match as a direct statement of guilt or identity. A likelihood ratio can compare the probability of the observed genetic evidence under alternative explanations. Population allele frequencies, relatedness, degradation, contamination, dropout, and mixtures influence that calculation. A very strong genetic association between a person and a sample does not by itself establish when the material was deposited, how it arrived, or what activity produced it.

Data standards, privacy, and interpretive limits

Genotyping results need standardized representation to remain interpretable across laboratories and over time. Sequence-based calls are commonly exchanged in Variant Call Format, which can record genomic position, alleles, genotype assignments, quality measures, and other annotations. Reference-genome version and software provenance are essential because coordinates and even local sequence representation can change. Array datasets additionally require attention to probe identity, strand orientation, and manufacturer-specific allele conventions.

Reference choice influences measurement. Reads containing an alternative allele may align less readily to a linear reference carrying another allele, producing reference bias. Highly variable or structurally complex regions can be especially affected. A pangenome represents multiple sequence paths or assemblies and can reduce some limitations of relying on one reference sequence. However, it also creates new requirements for coordinate systems, allele naming, and comparison with existing datasets.

Genetic data raise distinctive privacy concerns because they are persistent, highly identifying, and informative about biological relatives. Removing names does not necessarily make a rich genotype dataset anonymous. Genotypes can reveal relationships among samples and, when combined with other information, contribute to re-identification. Data governance therefore concerns access, permitted uses, security, retention, and the sharing of results, not merely whether an identifying label is attached.

Consent and interpretation are especially important when analyses reveal information beyond the original purpose. A research dataset may contain evidence of unexpected relatedness, chromosome variation, or medically relevant variants. Whether such findings are examined or returned depends on the study framework, consent process, validation requirements, and applicable rules. Research-grade measurements and clinically validated results are not interchangeable simply because they concern the same genomic position.

Ancestry inference and trait prediction illustrate the limits of interpretation. An ancestry estimate is conditional on the reference populations, marker panel, and statistical model used. Its categories describe similarity to selected comparison groups rather than discrete, immutable biological divisions. Genetic prediction of complex traits likewise depends on the population and environment in which predictive effects were estimated. Performance can decline when models are transferred to groups poorly represented in their development data.

Single-cell genotyping, increasingly accurate long reads, and integrated analysis of sequence and molecular state extend the range of detectable variation. They also expose additional sources of uncertainty: single-cell amplification can create dropout and artifacts, while tumour and microbial mixtures require models of heterogeneous populations. The appropriate analytical unit may be an individual chromosome, a cell lineage, a clone, or a mixture rather than a whole organism. Explicitly specifying that unit remains necessary even when measurement resolution increases.