Begin With the Biological Target
Choosing an assay is not a contest to find the most comprehensive technology. It is a decision about whether the measurements an assay can provide match the study’s question. A study of inherited variants needs DNA measurements; a study of transcript abundance needs RNA measurements. Even among DNA assays, the right choice depends on whether the targets are already known, which variant classes matter, and how broadly the study needs to look.
The four options considered here answer different questions. A genotyping array measures alleles at a selected set of DNA loci. An expression array measures signals from probes designed to represent selected transcripts. Exome sequencing targets protein-coding regions and some nearby sequence. Whole-genome sequencing (WGS) samples DNA across the genome, including noncoding regions. No option guarantees that every biologically relevant target is measured accurately: probe design, capture, read mapping, and the properties of the target sequence can all limit detection.
A Practical Decision Framework
Start by writing the study question in terms of the measurement it requires. “Which common inherited variants are associated with a trait?” calls for DNA genotypes. “Which transcripts change in abundance after treatment?” calls for RNA-expression measurements. “Which rare coding variants might explain a phenotype?” points toward sequencing targeted at coding regions or more broadly. A method that measures the wrong molecule cannot answer the question, even if it generates many data points.
Next, list the expected targets. Arrays are most directly suited to prespecified targets represented by appropriate probes. A genotyping array is a natural candidate when the study concerns many previously identified, usually common variants and a designed panel covers those loci. An expression array may suit a study of known transcripts when probe representation and the expected range of expression are adequate. Sequencing is more appropriate when the study needs to detect variants beyond a fixed list of array targets. Exome sequencing focuses that search on coding regions; WGS broadens it to noncoding regions as well.
Then ask which relevant targets could be missed. Genotyping-array coverage is determined by the variants included on the array, and coverage is uneven: an unrepresented variant is not directly measured. Expression-array results depend on the probes, their specificity, and their signal range; cross-hybridization and saturation can compromise interpretation. Exome capture is uneven and does not cover all coding bases equally, while much regulatory and other noncoding sequence lies outside its target. WGS offers broader genomic coverage, but repetitive or difficult-to-map regions, some structural changes, and low-quality evidence can still limit detection.
Finally, consider scale and resources. A lower per-sample cost can matter greatly in a large cohort, while a more comprehensive assay may be justified if the study’s main aim depends on targets the cheaper assay cannot observe. Costs, sample requirements, and available laboratory and analytic capabilities vary; use study-specific estimates rather than assuming a fixed price or performance ranking.
Specify DNA genotype, RNA abundance, or another target the proposed assay can actually measure.
List whether they are known or exploratory and whether the relevant variants are common, coding, noncoding, or another class.
Compare the target list with the assay’s probes, capture regions, or genome-wide scope, and identify likely technical limitations.
Compare expected information, sample scale, cost, and operational constraints; choose the least costly feasible option that can answer the question adequately.
How to Compare Broader Coverage With Added Cost
When two feasible assays differ in breadth, the decision can be framed as a comparison of expected benefit and added cost. Suppose assay B covers targets omitted by assay A. Let \(q\) be the probability, judged from the study’s scientific context, that the extra coverage will reveal information that changes the study’s conclusion, and let \(V\) represent the value of that information. If the additional total cost is \(\Delta C\), the expected incremental benefit is \(qV\). On this simplified scale, the broader assay is preferred when that expected benefit exceeds its additional cost:
This is a decision aid, not a universal statistical test: \(q\) and \(V\) must be justified for the study, and a broader assay must still be technically suitable and affordable. The logic follows directly from expected benefit: the added information is valuable only in the circumstances where it matters, so its expected value is its value in those circumstances multiplied by their probability. If the extra targets are central to the question, \(q\) may be high; if they are unlikely to affect the aim, paying for broader coverage may not be warranted.
Worked Scenarios
Worked Example: A Large Study of Known Common Variants
A hypothetical association study plans to recruit 4,000 participants. Its primary question concerns 100,000 previously identified common variants. A genotyping array has probes for the planned loci, and the study’s laboratory confirms that the relevant samples and DNA are suitable. For budgeting, assume hypothetical per-sample quotes of $55 for the array, $320 for exome sequencing, and $950 for WGS. Expression is not the target.
The array’s estimated assay cost is \(4{,}000 \times \$55 = \$220{,}000\). The corresponding exome estimate is \(4{,}000 \times \$320 = \$1{,}280{,}000\), and the WGS estimate is \(4{,}000 \times \$950 = \$3{,}800{,}000\). These are planning figures, not universal market prices.
For this question, the array directly targets the planned common variants and has the lowest quoted cost. Exome sequencing and WGS would generate broader sequence data, but that extra breadth is not required by the stated primary aim. Thus, given the stated probe fit and budget assumptions, the genotyping array is the best match. This conclusion would change if important expected variants were absent from the array or if discovering unanticipated rare variants became a central aim.
Worked Example: Measuring Expression of Known Transcripts
A hypothetical experiment compares RNA from treated and untreated cell samples across 120 samples. The primary endpoint is relative abundance for 40 named transcripts, all represented by suitable expression-array probes. Assume a hypothetical expression-array quote of $90 per sample and a sequencing-based expression option quoted at $250 per sample. The question is not about inherited DNA variants.
The expression-array estimate is \(120 \times \$90 = \$10{,}800\). The sequencing-based expression estimate is \(120 \times \$250 = \$30{,}000\), a difference of \(\$30{,}000-\$10{,}800=\$19{,}200\). Since the stated target is RNA abundance for a prespecified set of represented transcripts, the expression array is a reasonable choice under these assumptions.
The choice is conditional on probe performance: the laboratory should confirm that the probes distinguish the intended transcripts and that expected signals are not outside the array’s useful range. If novel transcripts, unrepresented targets, or fine distinctions between isoforms are central, this fixed-probe design may not meet the aim. Exome and WGS assays target DNA and, as described here, do not directly measure the study’s RNA-abundance endpoint.
Worked Example: Prioritizing Rare Coding Variants
A hypothetical study investigates a condition for which researchers expect the relevant inherited variants to be rare changes in protein-coding genes. The candidate list contains 30 genes. Assume the planned exome design includes these genes, but laboratory specifications indicate that some target bases may have inadequate capture or read evidence. The study’s resources support either an array or exome sequencing, but not WGS.
A genotyping array would measure only its selected loci. Unless the rare variants of interest were specifically represented, it would not provide a direct measurement of them. Exome sequencing is better matched because its intended target includes the coding regions in the candidate genes, where the study expects the variants to occur.
The choice is not equivalent to assuming perfect coverage. If a hypothetical review of the design estimates that 27 of the 30 genes meet the study’s prespecified target-coverage requirement, the fraction meeting that requirement is \(27/30=0.90\), or 90%. The remaining 3 genes are a material limitation to report and address—for example, by checking whether an alternative assay is feasible or revising the study scope. Exome sequencing is the strongest match among the stated affordable options, but it cannot support a claim that all 30 genes will be assessed equally well.
Worked Example: A Suspected Noncoding Cause
A hypothetical disease study has evidence that relevant variants may lie in both coding regions and regulatory regions outside them. Its planning list contains 18 plausible targets: 12 coding, 3 near coding-region splice boundaries, and 3 regulatory noncoding regions. Assume that the selected exome design covers the 12 coding targets and the 3 splice-boundary targets adequately, but does not capture the 3 regulatory regions. Assume also that the WGS design can assess all 18 targets, subject to ordinary mapping and evidence limitations.
Under these assumptions, the exome covers \(12+3=15\) of the 18 listed targets, giving \(15/18=0.8333\), or about 83.3%. It omits \(3/18=0.1667\), or about 16.7%, of the targets that are central to the hypothesis. WGS is therefore the better match if the regulatory targets are important enough to justify its additional cost and the study can support the required analysis.
The arithmetic describes the hypothetical design, not a guarantee of detection. In particular, “within the WGS scope” does not mean that every variant at a target will be detected or interpreted successfully. If the noncoding targets were merely speculative and resources were very limited, an exome could still be a defensible choice, but the resulting study would need to acknowledge that its design is less informative about those targets.
Comparing the Main Options
| Assay | Best fit | Important limitation to check |
|---|---|---|
| Genotyping array | Many selected DNA variants, often common and already known | Unrepresented variants are not directly measured; coverage depends on array design and probe performance |
| Expression array | RNA abundance for represented, known transcripts | Probe specificity and signal range can limit measurement; unrepresented transcripts are not assessed |
| Exome sequencing | Variants in targeted coding regions and some nearby sequence | Capture is uneven; much noncoding sequence is outside the target |
| Whole-genome sequencing | Broad search across coding and noncoding genomic regions | Greater breadth does not remove mapping, evidence, interpretation, or resource limitations |
The comparison is about fit, not a claim that one method dominates on every dimension. An array may be more efficient for a large study of selected common variants; sequencing may be essential for rare or unlisted variants. An expression array measures RNA, whereas the other three choices described here measure DNA. Even the broadest DNA assay cannot substitute for an expression measurement when the outcome of interest is transcript abundance.
Diagnostics, Failure Modes, and Common Mistakes
Before committing to a design, ask whether the assay’s targets actually overlap the study’s biological targets. For arrays, inspect the probe or locus design rather than relying on the assay label. For exome sequencing, inspect the capture design and expected performance for the genes of interest. For WGS, check whether the relevant regions are likely to be assessed reliably. For expression arrays, verify transcript representation, probe specificity, and whether the expected expression range is compatible with the technology.
- Choosing by breadth alone. “WGS covers more, so it must be best” ignores whether that extra coverage matters, whether it is technically usable, and whether the study can afford it. Match breadth to the question.
- Choosing a DNA assay for an RNA question. DNA variant measurements do not directly provide transcript-abundance measurements. Select an assay that measures the study’s endpoint.
- Treating a designed target as a guaranteed measurement. A locus on an array or within a capture region may still be poorly measured. Confirm assay specifications for the actual targets.
- Assuming an exome is a complete coding survey. Capture performance varies, and some targets may be weak or absent. Review target-level limitations before describing the intended coverage.
- Equating genome-wide scope with complete detection. WGS is broader, not flawless. Difficult sequence and downstream interpretation can still limit what the study learns.
A useful design review records the question, analyte, target classes, targets expected to be measurable, important blind spots, and the reason the chosen assay is adequate. If a key target cannot be measured reliably, the design should be revised, the target scope narrowed explicitly, or the limitation treated as a substantive constraint—not hidden by describing the assay as comprehensive.
Check Your Understanding
For each question, identify the best-matched assay from the four options and explain the decisive consideration. Solutions follow the questions.
- A large cohort study aims to test a list of common inherited variants, all represented on a suitable genotyping array. Which assay is the most direct match?
- A study asks whether treatment changes the abundance of 25 known transcripts represented by reliable probes. What is the central reason to consider an expression array rather than a DNA assay?
- A rare-disease study expects variants in coding regions of a set of genes, and its resources do not support WGS. Which sequencing option is the better match, and what limitation must be checked?
- A study’s main hypothesis depends on variants in regulatory regions outside coding sequence. Why may an exome be an inadequate choice even if it covers coding genes well?
- True or false: WGS guarantees that every variant in every genomic region will be detected. Correct the statement.
Solutions. (1) A genotyping array is the direct match because the targets are known common DNA variants represented on the array; its design and probe suitability still need confirmation. (2) An expression array measures RNA abundance for represented transcripts, while the DNA assays do not directly measure that endpoint. (3) Exome sequencing is the better match because the expected variants are coding; check capture performance for the specific genes and targets rather than assuming uniform coverage. (4) An exome’s target is concentrated on coding and nearby regions, so it may omit the regulatory targets central to the hypothesis; WGS is the broader DNA option if feasible. (5) False. WGS samples broadly across the genome, but mapping difficulty, evidence quality, and interpretation can still prevent detection or confident assessment of some variants.