Whole Exome Sequencing in Population Genetics: When WES Beats WGS

For Research Use Only. Not for use in diagnostic procedures or clinical decision-making.

Whole exome sequencing (WES) captures the ~1–2% of the human genome that codes for proteins — yet this small fraction harbors approximately 85% of known disease-associated variants. For population cohorts, that concentration of functional information translates into a practical advantage: WES delivers high-depth, analytically tractable data on the most interpretable part of the genome at roughly a quarter to a third of the per-sample cost of whole genome sequencing (WGS). A 2024 head-to-head comparison of 149,195 UK Biobank participants found that across 100 complex traits, WGS generated data on roughly five times more variants than WES plus imputation — but added only about 1% more robust genetic associations.

For population genetics studies focused on coding variant discovery, burden testing, and gene-based association, WES remains not just a budget compromise but often the statistically optimal design. This guide explains when WES is the right platform, when it is not, and how to design a population exome study that maximizes discovery power per dollar. It is written for human genetics researchers, animal cohort researchers, disease research teams, and biotech variant discovery teams evaluating sequencing platforms for projects from a few hundred to tens of thousands of samples.

Figure 1: At a fixed budget, WES accommodates 2–3× more samples than WGS — and for coding variant association studies, larger sample sizes consistently deliver more statistical power than broader genomic coverage per sample.

The WES Value Proposition

WES occupies a strategic middle ground in the sequencing landscape — deeper coverage than WGS at a given budget, broader functional annotation than genotyping arrays, and a mature analytical ecosystem that spans more than a decade of population-scale studies. Three structural advantages define its value for population cohorts.

Depth Over Breadth

A typical WES experiment sequences the exome to 100–200× mean coverage, compared to 30× for a standard WGS. This depth matters for rare variant calling. At 30×, a heterozygous site has approximately 15 supporting reads on average; at 100×, the same site has 50. The additional depth reduces genotype uncertainty at low-frequency variants and improves indel calling, which remains the most error-prone variant class in exome data. Belkadi et al. (2015) found that WGS detected approximately 650 high-quality coding SNVs per sample that WES missed — roughly 3% of all coding variants — primarily in regions with poor capture efficiency. For most population applications, that 3% gap is an acceptable trade-off for the cost savings.

Cost Efficiency at Scale

For a $100,000 sequencing budget, WES at $200–400 per sample processes roughly 250–500 samples; WGS at $600–1,000 per sample processes roughly 100–170 samples. Given that statistical power for rare variant association scales primarily with sample size — not variant count — the WES budget often produces more discoveries. Gaynor et al. (2024) demonstrated this empirically: tripling the sample size using WES+imputation yielded a fourfold increase in association signals across 100 complex traits, while switching from WES+imputation to WGS on the same samples added only ~1% more signals. For research teams planning population-scale exome studies, whole exome sequencing for population genetics provides end-to-end services from library preparation and capture through variant calling and annotation.

Table 1: WES vs WGS Cost and Coverage Comparison for Population Studies

When WES Outperforms WGS

WES is not a cheaper-but-worse WGS. For specific study designs, it is the statistically superior choice — because the resources saved on sequencing can be redirected to larger sample sizes.

Coding Variant Association Studies

If the research question is "which genes harbor rare coding variants associated with a trait," WES is purpose-built for the task. The exome captures approximately 95% of known Mendelian disease genes and the vast majority of clinically interpretable variants. For complex trait studies, gene-based burden tests (CMC, SKAT, SKAT-O) aggregate rare coding variants within each gene, and the statistical power of these tests depends primarily on the number of carriers — which increases with sample size, not with genomic coverage. Auer et al. (2016), drawing on the NHLBI Exome Sequencing Project experience across ~7,000 individuals, codified the core lesson: "prioritize sample size over sequencing depth or breadth for rare variant discovery."

Budget-Constrained Cohorts

When per-sample cost is the binding constraint — a common situation for investigator-initiated studies, pilot projects, and multi-phenotype cohorts — WES delivers more statistical power per dollar for coding-focused questions. The cost gap between WES and WGS has narrowed, but WES retains a 2–3× per-sample advantage. For cohorts above 1,000 samples, that difference translates into hundreds of additional samples, which directly improves power for gene-based tests. A 2025 study by Wainschtein et al. using WGS from 347,630 UK Biobank participants found that rare variants accounted for approximately 22% of variant-based heritability on average — but 79% of that rare variant signal came from outside the exome. This finding underscores both WES's value (it captures the coding fraction efficiently) and its fundamental limitation.

Established Analytical Toolkit

WES analysis pipelines have benefited from over a decade of development by large consortia. The NHLBI ESP, ExAC/gnomAD, and UK Biobank exome projects have produced validated workflows for variant calling, quality control, annotation, and association testing. Tools for gene-based rare variant testing — SKAT, SKAT-O, burden tests, VT — were developed and calibrated on exome data. Reference population databases (gnomAD v4.1 includes exome data from over 800,000 individuals) provide allele frequency estimates specifically for coding regions at depths comparable to research WES. For teams planning population-scale variant analysis, variant calling services provide joint-genotyping and quality control pipelines optimized for exome cohort data.

Where WGS Wins

Understanding WES's limitations is as important as understanding its strengths. There are study designs where choosing WES means choosing to miss the answer.

Non-Coding and Regulatory Variants

WES is blind to approximately 98–99% of the genome. Promoter variants, enhancer mutations, long non-coding RNA loci, and deep intronic splice-altering variants are all invisible to exome capture. For traits where genome-wide association studies have identified predominantly non-coding signals — autoimmune diseases, many complex metabolic traits, and behavioral phenotypes — GWAS loci often map outside the exome, and WES alone cannot interrogate them. Gaynor et al. (2024) confirmed that WGS-exclusive associations were concentrated in non-coding regions with lower linkage disequilibrium to exonic variants.

Structural Variants and CNVs

Exome capture is fundamentally unsuited to detecting structural variants. While some large CNVs leave a footprint in exome read depth that can be detected by specialized tools (XHMM, CoNIFER, ExomeDepth), sensitivity for CNVs under ~50 kb is poor. Balanced rearrangements (inversions, translocations) are completely invisible. For population studies where structural variation is a primary focus, CNV analysis from WGS data is the appropriate platform.

Multi-Ancestry Populations

Exome capture probe design has historically been optimized against European reference genomes. Population-specific coding variants — particularly in African ancestry populations, which harbor the highest levels of human genetic diversity — may fall in regions with reduced capture efficiency. Auer et al. (2016) noted that African American participants in the ESP showed substantially more population-specific coding variation than European Americans, underscoring the importance of ancestry-aware probe design and reference panel selection. For multi-ancestry cohort studies, WGS or WES with ancestry-informed capture designs provides more equitable variant discovery across populations. Studies that require analysis of population stratification and ancestry inference should incorporate population structure analysis to account for these differences in both study design and association testing.

Designing a Population WES Study

Sample Size and Power

The most common design error in population WES studies is under-powering. Kiezun et al. (2012) demonstrated that for rare variant associations at effect sizes typical of complex traits, over 10,000 samples may be required to achieve exome-wide significance for most genes. Auer et al. (2016) provided practical recommendations from the ESP: use permutation-based empirical significance thresholds rather than rigid Bonferroni correction; stratify or weight variants by functional annotation (nonsense > splice-site > damaging missense, excluding synonymous variants); and prioritize extreme-trait sampling when the parent cohort is large enough to support it. For quantitative traits, selecting individuals beyond the 5th and 95th percentiles can dramatically increase power at a given sample size.

Capture Kit Selection

Belova et al. (2025) compared four exome capture kits — Agilent SureSelect v8, Roche KAPA HyperExome, Vazyme VAHTS, and Nanodigmbio NEXome — on a standardized human sample at 50 million reads. All four kits delivered strong performance (>95% of targets covered at ≥20×), but they differed in coverage uniformity. The fold-80 penalty — the additional sequencing needed to bring 80% of targets to the mean coverage — ranged from 1.43 (Roche, most uniform) to 1.58 (Nanodigmbio). For population studies, coverage uniformity matters more than maximum on-target rate: uneven coverage means some genes are systematically under-sequenced across all samples, creating gene-specific false-negative rates that no amount of sample size can overcome.

Table 2: Exome Capture Kit Performance Comparison (Belova et al. 2025, 50M reads)

Sequencing Depth

For population WES, 100× is the standard target depth — but "100×" means average coverage across the target, and the fold-80 penalty means some regions receive substantially less. With a fold-80 penalty of 1.5, achieving 20× at 95% of targets requires a mean coverage of approximately 100× (20 × 1.5 / 0.3 ≈ 100×). Below 80× mean coverage, genotype quality degrades for low-frequency heterozygous variants, which are exactly the variants of interest in rare variant studies. The incremental cost of moving from 80× to 100× is modest and buys insurance against the genotype errors that are most damaging to rare variant association tests.

Quality Control Essentials

Auer et al. (2016) outlined QC fundamentals that remain current: jointly call variants across all samples rather than calling samples individually and merging; exclude samples with extreme heterozygosity, low concordance with genotyping array data, or evidence of contamination; filter variants by genotype quality, depth, allele balance for heterozygotes, and Hardy-Weinberg equilibrium in controls; and balance cases and controls across sequencing batches, capture kits, and flow cells. Including intentional sample duplicates — at a marginal cost of roughly 0.74% additional samples — provides direct estimates of variant calling consistency and should be standard practice in population WES studies above 500 samples.

Figure 2: Statistical power for gene-based rare variant association tests scales primarily with sample size, not genomic coverage — the fundamental reason that WES on larger cohorts can outperform WGS on smaller ones for coding variant discovery.

WES in the Drug Target Context

WES data feeds directly into variant-to-gene pipelines that underpin drug target identification. Rare coding variants that disrupt a gene and protect against disease — so-called "human knockout" models — provide some of the strongest genetic evidence for therapeutic targeting. PCSK9 loss-of-function variants, discovered through population sequencing, led to the development of PCSK9 inhibitors for hypercholesterolemia. This gene-to-phenotype logic is the organizing principle behind large-scale exome sequencing efforts in biobanks: sequence the exomes of hundreds of thousands of individuals, identify genes where rare protein-truncating variants associate with clinically relevant phenotypes, and prioritize those genes for drug development. For research teams working from GWAS summary statistics toward target hypotheses, GWAS variant prioritization for drug targets provides the analytical framework for moving from associated locus to candidate gene to therapeutic hypothesis.

For projects integrating exome data with transcriptomic and epigenomic measurements, multi-omics QTL integration extends the WES signal from coding variant to expression QTL, methylation QTL, and protein QTL colocalization — building a multi-layer evidence chain that strengthens target confidence beyond what coding annotation alone can provide.

Choosing Between WES and WGS

The platform decision reduces to four questions: What variant types matter? What is the budget? How many samples? What populations?

Table 3: Platform Selection Decision Framework for Population Sequencing

For whole-genome sequencing as an alternative platform, whole genome re-sequencing for population genetics provides 30–60× coverage with comprehensive variant detection across coding and non-coding regions.

Figure 3: The platform selection decision reduces to four questions — variant types of interest, budget, sample size, and population diversity — with WES favored for coding-focused, budget-constrained, large-N studies and WGS favored for non-coding, structural variant, and future-proofed population resource projects.

Frequently Asked Questions

For Research Use Only. Not for use in diagnostic procedures or clinical decision-making.

  Related Services

Whole Exome Sequencing for Population Genetics

Whole Genome Re-sequencing for Population Genetics

Reduced-Representation Sequencing for Population Genetics

Population Structure Analysis Service

DNA Methylation Microarray for population genetics

Population Dynamics Analysis

Population Structure & Evolution Analysis

Population Evolution Analysis Service

  Download

Population Genetics and Sequencing Solutions

Whole Exome Sequencing (WES)

HLA Typing