How Genotyping Arrays Measure Alleles

From a DNA Variant to an Array Signal

A genotyping array is designed to measure a specified set of DNA variants. At each selected position, the array uses probes—short DNA sequences attached at known locations—to distinguish the alternative alleles. After DNA from a sample is prepared and exposed to the probes, fluorescent measurements provide evidence about which allele or alleles are present. The central idea is that a signal is generated at a probe associated with a particular allele; the signal is not a direct photograph or count of the DNA molecules.

This distinction matters because the array measures fluorescence at many probe locations, while the biological quantity of interest is the sample's genotype at a locus. The measurement depends on the DNA binding or reacting with the probe, the efficiency of that reaction, the fluorescence chemistry, and the instrument. Interpreting the result therefore requires knowing what each probe was designed to detect.

How the Measurement Is Made

A typical genotyping-array workflow begins with DNA from a sample. The DNA is prepared so that target regions can interact with array probes. Depending on the array chemistry, the signal may arise because DNA hybridizes preferentially to an allele-specific probe, or because a probe helps incorporate a labeled nucleotide that matches the allele at the selected position. These are different implementations of the same broad measurement principle: the array has a defined address for each target and records fluorescence associated with allele-specific chemistry there.

The array is then scanned. For a locus with alleles C and T, the data may include one signal associated with the C-targeting probe and another associated with the T-targeting probe. The two signals need not use two different fluorescent dyes: some designs distinguish probes by their array locations or other labeling and detection arrangements. What matters for interpretation is the mapping between each measurement and its intended allele.

For a diploid sample, the genotype at a biallelic locus is commonly one of three possibilities: CC, CT, or TT. A CC sample has two C copies and no T copies at that locus; a CT sample has one copy of each; and a TT sample has two T copies. In an idealized assay, the C-associated signal tends to be stronger when more C copies are present, and likewise for T. Real signals are affected by assay and measurement variation, so this tendency is not a guarantee that signal strength equals copy number.

A Simple Signal Model

A useful way to describe the measurement is to separate background fluorescence from the part associated with allele-specific target. Let \(S_C\) and \(S_T\) be the measured signals for the C- and T-associated probes. Let \(q_C\) and \(q_T\) be the numbers of corresponding allele copies in a diploid sample, so \(q_C+q_T=2\). A simplified model is

Here \(b_C\) and \(b_T\) represent background signal, \(g_C\) and \(g_T\) represent probe-specific response per target copy, and \(\varepsilon_C\) and \(\varepsilon_T\) collect random measurement variation and other unmodeled effects. This is a conceptual model, not a universal calibration formula. In particular, the two probes may have different response strengths, so \(g_C\) need not equal \(g_T\).

The model clarifies why raw fluorescence is not automatically an allele count. Even if two samples have the same \(q_C\), their measured \(S_C\) values can differ because of background, probe response, DNA preparation, or measurement variation. In an idealized setting where background is removed, the responses are equal, and errors are negligible, the two signals are proportional to the corresponding copy numbers.

To see this, set \(b_C=b_T=0\), \(g_C=g_T=g\), and both errors to zero. Then \(S_C=gq_C\) and \(S_T=gq_T\). For a heterozygote, \(q_C=q_T=1\), so \(S_C=S_T=g\). For CC, \(q_C=2\) and \(q_T=0\), so \(S_C=2g\) and \(S_T=0\). This reasoning explains the qualitative pattern, but the ideal conditions are not guaranteed in actual measurements.

Assumptions, Diagnostics, and Failure Modes

Interpreting allele-specific fluorescence depends on several assumptions. The probe locations must be correctly mapped to the intended locus and allele; the DNA preparation must leave target sequence available to the assay; and the measured response must be sufficiently specific that a probe signal mainly reflects its intended allele. Signal detection must also be within a useful measurement range. If the instrument or chemistry produces a saturated signal, a larger amount of target may not produce a proportionally larger measurement.

A practical first check is whether the expected allele-specific signals can be distinguished from background. Other warning signs include unusually weak signals for both probes, inconsistent measurements across technical replicates, or signal behavior that changes markedly with sample preparation. Such observations can indicate poor target binding, failed chemistry, contamination, or an instrument problem. They do not identify the cause on their own. A signal should be treated as a measurement requiring quality assessment, not as an unquestionable biological fact.

Probe specificity is another key condition. A probe may bind imperfectly to a non-target sequence, or a nearby sequence difference may affect binding or extension. In that case, the observed fluorescence can depart from the simple model. These possibilities are reasons to check probe and assay performance; they should not be resolved by assuming that every signal difference reflects a difference in allele copies.

What Arrays Can and Cannot Measure

A genotyping array is a targeted assay: it measures variants selected in advance and represented by probes on that array. This makes it suitable when the study needs measurements at those selected sites. It does not, from the two signals at one designed locus alone, reveal every unrepresented variant nearby or elsewhere in the genome.

Direct sequencing is a different measurement approach: rather than relying on a preselected allele-specific array probe at each site, it measures DNA sequence information across the regions assayed. Sequencing may therefore provide evidence about variants not represented on a particular array, but the methods have different laboratory and data-analysis requirements. Neither method makes an imperfect measurement infallible. The choice depends on the study question, the variants of interest, and the required scope of measurement.

For this tutorial’s purpose, the essential interpretation is modest: the array measures fluorescence associated with probes designed for selected alleles. The signals can support an inference about which allele targets are present, but the signal-to-genotype interpretation depends on assay-specific performance.

Common Mistake

• Wrong reasoning: “The C probe has twice the fluorescence of the T probe, so the sample has twice as many C copies as T copies.”
• Correction: Probe response, background, and measurement scale can differ by allele. Fluorescence is not a copy count unless an appropriate assay model and calibration justify that interpretation. First identify the allele assigned to each probe, assess signal quality, and use the assay’s validated interpretation procedure.