Comparing Genotyping Arrays, Exomes, and Genome Sequencing

Three Ways to Observe Genetic Variation

Genotyping arrays, exome sequencing, and whole-genome sequencing can all produce information about genetic variation from the same DNA sample, but they do not measure the same things. An array tests a predefined set of positions using probes. Exome sequencing enriches and sequences selected regions, usually protein-coding exons. Whole-genome sequencing samples DNA across the genome. These differences in target and resolution affect which variants can be observed and what kinds of errors can obscure or distort the evidence.

A useful comparison keeps the biological material fixed and asks three separate questions: Which positions was the assay designed to measure? What is the finest level of variation it can resolve there? What can make the evidence incomplete or misleading? A broader target is not automatically more accurate at every position, and a high-resolution measurement does not guarantee that the relevant position was successfully observed.

Targets and Resolution

A genotyping array has probes designed for selected variants, often common single-nucleotide polymorphisms. At each targeted position, the assay measures hybridization intensities associated with different alleles; genotype clusters are used to infer a genotype. The measured set is therefore largely predetermined by the array design. An array can provide many genotype measurements efficiently, but it does not ordinarily sequence every base around each marker or discover arbitrary variants elsewhere in the genome.

Exome sequencing enriches regions selected by a capture design, commonly exons and sometimes nearby sequence. Sequencing reads can reveal single-base changes and some small insertions or deletions within regions that are sufficiently captured, sequenced, and mapped. The target is much broader than a typical marker list but remains a selected subset of the genome. Capture boundaries and designs differ, so “exome” does not mean that every coding base is measured equally well.

Whole-genome sequencing (WGS) sequences reads from across the genome rather than deliberately restricting enrichment to exons. It can therefore provide evidence in coding and noncoding regions, including positions not represented on an array or targeted by an exome design. With short reads, however, some positions remain difficult to assess, and read depth is not uniform. A genome-wide target is not equivalent to complete, error-free observation of every base.

The comparison assumes the same underlying DNA sample, with its true genetic variants fixed, and interpretable assay outputs from each technology. In practice, DNA quality, library preparation, array design, capture design, sequencing depth, and analysis procedures can differ. Thus, the table describes typical measurement scope, not a guarantee that every platform or implementation behaves identically.

Why Sequencing Evidence Varies Across Positions

Sequencing evidence depends partly on how many reads cover a position. Even when reads are sampled from a true heterozygous site, random sampling can yield different numbers of reference and alternate alleles. A simple model makes this point. Suppose each read independently samples either allele with probability one-half, mapping and base calls are correct, and the position is truly heterozygous. If \(X\) is the number of alternate-allele reads among \(d\) reads, then \(X\sim\mathrm{Binomial}(d,0.5)\).

This calculation describes sampling alone. It does not include sequencing error, mapping bias, capture bias, or the thresholds a caller uses. Those additional processes can make real evidence less balanced or less reliable than the simplified model suggests.

Characteristic Errors and Their Consequences

Array errors arise in part from the probe-based measurement. A variant in or near a probe-binding region can disrupt hybridization and make an allele appear weak or absent. Similar sequences elsewhere in the genome can contribute cross-hybridization. Intensity clusters may be poorly separated, so genotype assignment can be uncertain or wrong. Array ascertainment is also central: variants absent from the design are not ordinary failed calls; they were not directly targeted in the first place.

For exome sequencing, capture efficiency can differ across regions. A target may have low depth, and nearby or boundary positions may be inconsistently captured. Sequence composition, including GC content, can contribute to uneven coverage. Reads can also contain sequencing errors, map ambiguously, or show unequal support for the two alleles. Consequently, an exome can have broad coding scope while still providing weak evidence at particular exons or bases.

WGS avoids the exome's deliberate restriction to captured coding targets, but it retains many sequencing and mapping challenges. Read depth varies; sequencing errors can create false variant evidence; and reads from repetitive or highly similar regions may map ambiguously. Mapping can also favor the reference allele, leaving fewer reads that support a real alternate allele. Some forms of genomic variation are difficult to resolve from short reads even when the region has substantial read depth.

A low-confidence or absent call should not be treated as interchangeable with a reliable reference genotype. Likewise, observing a few reads that differ from the reference does not by itself establish a true variant. A platform's characteristic error mechanisms help explain why evidence may be missing or misleading, but the observed result at a particular position must still be evaluated on its own terms.

How to Read a Cross-Assay Comparison

A fair comparison must make clear which positions are included. Comparing array calls at its selected markers with exome calls across coding regions and WGS calls across the genome mixes target scope with assay performance. One assay may appear to have fewer calls simply because more of the positions under consideration were outside its intended target. A comparison on a shared, adequately defined set of positions answers a different question from a comparison of total genomic breadth.

It is also important to separate three ideas:

• Target inclusion: Was the position included in the assay's design or intended target?
• Call availability: Was there enough usable signal or read evidence to report a genotype?
• Call accuracy: Does the reported genotype match the sample's true genotype?

These quantities need not move together. A position can be targeted but have inadequate evidence; a call can be available but incorrect; and an unrepresented variant can be outside an array's direct measurement scope rather than a technical failure at a targeted marker.

Common Mistake and Practical Limitations

Common mistake: “WGS covers the whole genome, so it must give a complete and correct genotype at every position; if an array and WGS disagree, the WGS call is automatically right.” This reasoning confuses target breadth with call certainty. WGS can provide evidence beyond array markers, but coverage, mapping, and sequencing errors remain. An array can also be affected by probe design or cluster ambiguity. A disagreement identifies a position to examine; by itself, it does not reveal which call is correct.

The right interpretation is limited to the evidence and target definition available. Array measurements are constrained by their predefined markers; exome measurements by capture design and successful sequencing of targeted regions; WGS measurements by read coverage and the ability to distinguish and map sequence. None of the three produces perfect evidence at every potentially relevant site. The comparison describes what each assay measures and how its evidence can fail; it does not, on its own, determine which assay is appropriate for a particular study.