Exome Sequencing

Exome sequencing—also known as whole-exome sequencing (WES)—is a targeted DNA sequencing strategy that selectively captures and decodes the protein-coding regions (exons, collectively termed the exome) across an organism's genome. Developed in 2009 by Sarah B. Ng, Jay Shendure, Michael J. Bamshad, and Debbie A. Nickerson at the University of Washington, exome sequencing demonstrated that hybridization capture coupled with massively parallel sequencing could identify the causative genes for Mendelian disorders—beginning with the 2010 discovery of DHODH mutations in Miller syndrome—without linkage pedigrees or whole genome sequencing.

In the human genome (haploid length 3.055 gigabases), the exome encompasses approximately 180,000 exons distributed across ~20,000 protein-coding genes, spanning 30 to 45 megabases (Mb) of sequence. Although this coding footprint constitutes only 1.0% to 1.5% of the nuclear genome, it harbors an estimated 85% of known high-penetrance disease-causing variants in Mendelian single-gene disorders alongside most actionable somatic driver mutations in oncology. By restricting sequencing reads to the 1.5% coding fraction, WES achieves high depth of coverage (80× to 150× in germline diagnostics; 500× to 1,000× in somatic tumor-normal profiling) at roughly one-quarter the sequencing throughput required for a 30× whole genome.

Genomic Shearing, Pre-Capture Library Construction, and Biotinylated Baits

Clinical exome sequencing relies on in-solution hybridization capture (which superseded early solid-phase microarray capture). First, 50 to 500 ng of genomic DNA is mechanically sheared via acoustic cavitation or enzymatically fragmented into 150 to 250 bp double-stranded fragments. Following end-repair, 3′-adenylation, and ligation of dual-indexed sequencing adapters, a low-cycle pre-capture polymerase chain reaction (4–6 cycles) generates an adapter-ligated shotgun genomic library.

The denatured library is then incubated in solution for 2 to 16 hours at 65 °C with a synthetic pool of biotinylated oligonucleotide capture probes ("baits") designed to tile every annotated exon in the RefSeq, Consensus CDS (CCDS), and GENCODE databases, plus ±10 to 50 bp of flanking intronic sequence to capture canonical splice donor (GT) and acceptor (AG) sites. To prevent repetitive elements (such as Alu or LINE repeats) within library inserts from cross-hybridizing non-specifically, excess Cot-1 human placental DNA and adapter-specific blocking oligonucleotides are included in the hybridization buffer.

Comparison of Commercial Exome Capture Platforms and Clinical Modalities

Streptavidin Paramagnetic Bead Pulldown and Post-Capture Sequencing

Following hybridization, streptavidin-coated paramagnetic beads are added to the reaction. Because the non-covalent interaction between biotin and streptavidin exhibits one of the highest affinities in biochemistry (K_d ≈ 10⁻¹⁴ to 10⁻¹⁵ mol/L), the biotinylated RNA:DNA or DNA:DNA duplexes bind tightly to the magnetic beads. Applying an external magnet immobilizes the target-bound beads against the vessel wall while the ~98.5% unhybridized non-coding genomic background is removed through high-temperature, low-salt washes.

The captured exonic fragments are eluted by alkaline or thermal denaturation, amplified by 8 to 10 cycles of post-capture PCR, and sequenced via paired-end 2 × 100 bp or 2 × 150 bp reads to generate 8 to 15 gigabases of sequence per germline sample, ensuring that >95% to 98% of target coding bases achieve ≥20× coverage.

Variant Yield, Population Frequency Filtering, and Trio-Based Discovery

A single human exome sequenced at 100× mean depth yields 20,000 to 25,000 coding and splice-region variants relative to the GRCh38 reference assembly, including roughly 10,000 to 11,000 synonymous SNVs, 9,000 to 10,000 missense SNVs, 300 to 500 coding indels, and 100 to 200 putative loss-of-function variants (nonsense stop-gains, canonical splice-site disruptions, and frameshift indels).

To isolate the single causal mutation responsible for a rare Mendelian disorder from 25,000 background variants, bioinformatics pipelines apply sequential filters: (1) removing common polymorphisms whose minor allele frequency (MAF) exceeds 0.1% to 1.0% in population reference cohorts (gnomAD >730,000 exomes); (2) prioritizing nonsynonymous and splice-altering variants using in silico conservation and pathogenicity predictors (CADD, REVEL, SpliceAI, AlphaMissense); and (3) performing proband-parent trio WES (sequencing the affected child alongside both biological parents). Trio phasing identifies de novo mutations (only 1 to 2 coding de novo variants occur per human exome per generation), homozygous autosomal recessive variants, and compound heterozygous trans-allelic mutations, achieving a 25% to 40% diagnostic yield across developmental delay and congenital malformation cohorts.

Technical Limitations Relative to Whole Genome Sequencing

Despite its cost efficiency, exome sequencing exhibits four structural and biochemical constraints compared to PCR-free whole genome sequencing:

1. GC-Bias and Capture Dropout: Hybridization efficiency drops at both extremes of base composition (<25% GC and >75% GC), causing systematic coverage dropout across GC-rich CpG-island first exons unless mitigated by balanced double-stranded DNA probe designs.
2. Non-Coding Blind Spots: WES misses distal intronic splice mutations (>50 bp into an intron), promoter and enhancer mutations, and non-coding RNAs.
3. Structural Variant and CNV Noise: Because hybridization capture efficiency and PCR amplification vary exon-by-exon, read-depth CNV callers (ExomeDepth, XHMM, CNVkit) have reduced sensitivity for single-exon deletions/duplications and cannot detect balanced inversions or translocations whose breakpoints lie in non-coding introns.
4. Pseudogene Cross-Hybridization: Capture probes homologous to functional genes (such as SMN1, PKD1, or CYP2D6) simultaneously pull down high-identity non-coding pseudogenes (SMN2), confounding short-read alignment.

Why does Agilent SureSelect use 120-nt single-stranded RNA baits whereas Twist Bioscience and IDT use double-stranded or single-stranded DNA probes?

RNA:DNA heteroduplexes exhibit a higher thermodynamic melting temperature (T_m) and greater duplex stability than DNA:DNA homoduplexes of identical length and sequence, allowing 120-nt biotinylated RNA baits (transcribed in vitro from synthetic oligonucleotide libraries) to tightly capture divergent or polymorphism-containing target fragments. Conversely, silicon-synthesized double-stranded DNA (dsDNA) probes (Twist Bioscience) capture both the sense and antisense strands of denatured genomic DNA fragments independently and avoid transcriptional representation bias during probe manufacturing, achieving more uniform coverage across GC-rich exons with less over-sequencing of AT-rich regions.

Why does Trio Exome Sequencing (proband + mother + father) achieve a higher diagnostic rate than singleton (proband-only) WES?

A singleton exome contains 300 to 600 rare coding missense or loss-of-function variants (MAF < 0.1% in gnomAD), making it difficult to distinguish a novel dominant disease mutation from benign private family variants inherited from an unaffected parent. By sequencing the affected proband and both unaffected biological parents simultaneously (trio WES), Mendelian inheritance phasing restricts the search for dominant disorders to newly arisen de novo coding mutations—of which a human child carries an average of only 1.0 to 1.5 across the 35-Mb exome—and confirms whether two recessive variants in the same gene reside in trans (compound heterozygous, one from each parent) rather than in cis on the same chromosome.

How is Exome Sequencing used in cancer immunotherapy to calculate Tumor Mutational Burden (TMB) and predict neoantigens?

In clinical oncology, paired tumor-normal exome sequencing compares somatic DNA extracted from a tumor biopsy (sequenced to 500×–1,000× depth to capture subclonal populations down to 2%–5% variant allele fraction) against germline DNA from the patient's peripheral blood leukocytes (100×–150× depth). Subtracting germline polymorphisms reveals the catalog of tumor-specific somatic nonsynonymous mutations. Tumor Mutational Burden (TMB) is defined as the number of somatic coding mutations per megabase of captured exome (≥10 mut/Mb indicates responsiveness to PD-1/CTLA-4 immune checkpoint blockade), while combining somatic missense calls with patient HLA class I genotypes predicts tumor-specific neoantigen peptides for personalized mRNA cancer vaccines.

1. Ng, S. B., Turner, E. H., Robertson, P. D., Flygare, S. D., Bigham, A. W., Lee, C., Shaffer, T., Wong, M., Bhattacharjee, A., Eichler, E. E., Bamshad, M., Nickerson, D. A., & Shendure, J. (2009). "Targeted Capture and Massively Parallel Sequencing of 12 Human Exomes." Nature, 461(7261), 272–276. https://doi.org/10.1038/nature08250
2. Ng, S. B., Buckingham, K. J., Lee, C., Bigham, A. W., Tabor, H. K., Dent, K. M., Huff, C. D., Shannon, P. T., Jabs, E. W., Nickerson, D. A., Shendure, J., & Bamshad, M. J. (2010). "Exome Sequencing Identifies the Cause of a Mendelian Disorder." Nature Genetics, 42(1), 30–35. https://doi.org/10.1038/ng.499
3. Gnirke, A., Melnikov, A., Maguire, J., Rogov, P., LeProust, E. M., Brockman, W., Fennell, T., Giannoukos, G., Fisher, S., Russ, C., Gabriel, S., Jaffe, D. B., Lander, E. S., & Nusbaum, C. (2009). "Solution Hybrid Selection with Ultra-Long Oligonucleotides for Massively Parallel Targeted Sequencing." Nature Biotechnology, 27(2), 182–189. https://doi.org/10.1038/nbt.1523
4. Karczewski, K. J., Francioli, L. C., Tiao, G., Cummings, B. B., Alföldi, J., Wang, Q., Collins, R. L., Laricchia, K. M., Ganna, A., Birnbaum, D. P., et al. (2020). "The Mutational Constraint Spectrum Quantified from Variation in 141,456 Humans." Nature, 581(7809), 434–443. https://doi.org/10.1038/s41586-020-2308-7