Sequencing for rare and genetic disease research: comparing exome, genome, and trio sequencing

Clinical Research | Educational Series

Key Takeaways 

• Trio WGS/WES outperforms singleton sequencing because parental data helps identify de novo variants and resolve inheritance patterns. 

• Even with trio WGS, more than half of suspected monogenic rare disease families remain unsolved. 

• Genome benchmarking studies show long reads add resolution in low-mappability and structural regions but underperform homopolymers remain a challenge across platforms, and that's where Avidite Base Chemistry™ (ABC™) chemistry shows an order-of-magnitude lower error rate than every other platform tested. 

• Reagent cost per WGS trio ranges from roughly $300 to several thousand dollars depending on platform and depth strategy; exome sequencing costs far less per sample in sequencing reagents but incur additional capture-reagent and lab costs. 

Rare and genetic disease research increasingly relies on whole exome sequencing (WES) or whole genome sequencing (WGS), most powerfully performed as a parent-child trio. Trio designs let researchers directly distinguish inherited variants from de novo (newly arising) mutations, which lifts diagnostic yield above single-sample testing. Published estimates put trio-WES yield around 51% versus roughly 34% for singleton WES, with WGS pushing higher still for some phenotypes (Lai et al., 2024; Albuquerque et al., 2025). Even so, more than half of families with a suspected monogenic rare disease remain unsolved after standard short-read WGS (Nurchis et al., 2023).

A major reason: de novo mutations are disproportionately generated in and around the genome’s hardest-to-sequence regions—homopolymers, tandem repeats, and segmental duplications—which legacy short-read chemistries have historically struggled to read accurately. Long-read sequencing adds measurable diagnostic yield in these regions, but at a high-cost premium. Newer short-read chemistries, including Avidite Base Chemistry (ABC) from Element, can close part of that gap in specific difficult regions at short-read economics. This post walks through the biology, the benchmarking evidence, and the cost and throughput tradeoffs behind these choices.

What is a rare or genetic disease?

The definition of “rare disease” varies by region: the U.S. FDA defines it as a condition affecting fewer than 200,000 people nationally, while the European Union defines rare as affecting fewer than 1 in 2,000 people. Between 6,000 and 8,000 distinct rare diseases have been described, and an estimated 80% have an identifiable genetic origin (NIH GARD; Global Genes). Individually rare, these conditions are collectively common: conservative estimates put 263–446 million people worldwide living with a rare disease — roughly 3.5–5.9% of the global population (Wakap et al., 2019).

Most genetic rare diseases present early in life, often as developmental delay, congenital anomalies, neurodevelopmental or neuropsychiatric conditions, or progressive degenerative disease. Because thousands of genes can each cause a rare, overlapping phenotype, it is difficult to narrow number of genes that could be involved, which is why sequencing-based methods, rather than single-gene or panel assays, have become the practical approach for research into uncovering more of the genetic drivers of these diseases.

Best practices in rare disease sequencing: exome, genome, and trio design

Exome sequencing captures the protein-coding portion of the genome (roughly 1–2% of total DNA) using hybridization-capture reagents (from vendors such as SOPHiA Genetics, Twist Bioscience, or IDT) to enrich those regions before sequencing. It remains the most cost-efficient way to comprehensively survey coding variants. Whole genome sequencing is a more comprehensive method, covering the entire genome to also include introns, regulatory sequence, and structurally complex regions. WGS costs more per sample but delivers more even coverage (no capture bias), and is substantially better positioned to detect structural variants, copy-number changes, and repeat expansions that fall outside coding exons.

Sequencing a proband together with both biological parents, a trio, is widely considered best practice when parental samples are available. Trio sequencing lets researchers directly filter for de novo variants (present in the child, absent in both parents) and resolve compound heterozygous and other inheritance patterns without leaning on population allele-frequency databases alone. Variants of unknown significance inherited from a healthy parent can also be eliminated as non-disease causing, simplifying interpretation.

In practice, “trio” does not always mean identical sequencing depth for all three family members. Many cost-conscious designs sequence the proband deeply and the parents at reduced depth, since the goal for the parental samples is often simply to confirm or exclude the presence of a candidate variant rather than to call every variant genome-wide with full sensitivity.

Ongoing challenges: why solve rates remain incomplete

While NGS sequencing is a powerful tool for understanding genetic disease, published estimates indicate that more than half of families with a suspected monogenic rare disease remain unsolved after WGS-based analysis (Nurchis et al., 2023).

The reasons are varied:

• Some causal variants sit in genomic regions that short reads cannot reliably call 

• Some are non-coding or regulatory variants of still-uncertain functional effect

• Some require long-range phasing or structural/repeat characterization that short, unphased reads cannot resolve

• Some diseases involve genes not yet linked to a known phenotype

• Some causal variants are mosaic, present in only a subset of cells, and easily missed at standard sequencing depth.

Understanding why these variants are hard to find starts with understanding how de novo mutations arise in the first place.

Why de novo mutations are hard to find

A de novo variant is a new mutation present in a child, but absent from either parent’s genome, arising either in a parent’s germ cells or very early in embryonic development. As the male germline undergoes continuous mitotic division throughout life to produce sperm, it is estimated to contribute 70–80% of de novo variants on average, with offspring inheriting 1.3-1.5 more mutations for every additional year of paternal age (Shoag et al., 2025). Most de novo single-nucleotide variants originate as copying errors during DNA replication in the germline.

However, point mutations are only part of the picture. Replication slippage, where the replication machinery stutters while copying repetitive DNA, is a major source of small insertions, deletions, and repeat expansions or contractions at short tandem repeats (STRs) and homopolymers. Similarly, misalignment during recombination at segmental duplications (large, near-identical repeated blocks of genome) is a major driver of larger structural variants and copy-number changes. Finally, a meaningful share of de novo variants also arise post-zygotically, after fertilization, producing mosaicism that can be tissue-restricted and potentially absent from a blood sample.

The practical consequence is that the molecular processes that generate some types of disease-causing de novo variants are mechanistically concentrated in and around the genomic features, homopolymers, STRs, segmental duplications, GC-rich and low-complexity sequence, that are intrinsically hardest for sequencing technologies to read accurately.

A platform’s performance in these “difficult” regions is therefore relevant to its ability to comprehensively detect mutations that may be linked to rare disease.

What the genome benchmarking studies show

The Genome in a Bottle (GIAB) Consortium’s genomic stratifications resource formalizes this idea, defining BED-file “stratifications”, including homopolymers, tandem repeats, segmental duplications, GC extremes, low-mappability regions across GRCh37, GRCh38, and the T2T-CHM13 reference, and is now used industry-wide to benchmark sequencing platform and variant-caller performance by genomic context (Dwarshuis, et al., 2024).

A 2025 complete diploid human genome benchmark, built on a fully telomere-to-telomere-resolved HG002 reference, added 701 Mb of autosomal and 217 Mb of sex-chromosome sequence (about 15% of the genome) that was invisible to prior benchmarks, and directly compares sequencing platforms across it. The study found that Illumina short reads carry roughly five times the transversion error rate of other platforms, while PacBio HiFi and Oxford Nanopore long reads carry indel error rates roughly two orders of magnitude higher than short reads. In homopolymers specifically, the study reported that Element reads showed an order of magnitude lower error rate than every other platform tested—short- or long-read (Hansen, Zook, Phillippy et al., 2025).

A related preprint from the SMaHT Network, focused on detecting sub-1% frequency somatic variants, found a similarly stratified pattern: error rates cluster tightly across platforms in “easy” genomic regions but diverge sharply in difficult ones. For homopolymer indels specifically, the study reported error rates of 0.14% on AVITI™ and 0.62% on NovaSeq, versus 15–18% for the long-read platforms tested (Jang et al., 2025).

Long reads still have an important role to play, despite higher indel error rates. Their ability to span an entire repeat expansion or segmental duplication in a single read gives them an advantage for discovery of structural and repeat-driven variants that short reads cannot span. Published estimates suggest PacBio HiFi WGS added 7–17% incremental yield of potential causative variants in cohorts that had already been through inconclusive short-read WGS or WES. (Del Gobbo and Boycott, 2025).

However, despite the added yield from long reads, it remains more cost effective to default to short read WES or WGS as a first-pass method, with long-read follow-up reserved for unresolved cases or where a structural or repeat-expansion mechanisms are suspected. As detailed in the cost comparison below, HiFi and nanopore WGS reagent costs currently run several-fold higher per genome than high-throughput short-read platforms.

Other considerations: platform cost, multiplexing, and turnaround time

Sequencing cost is not a single number—it depends on target read depth, multiplexing level, and whether parents are sequenced at full or reduced depth. The estimates below are reagent-only, list-price approximations based on each vendor’s published specifications and statements as of mid-2026; actual costs vary with contract terms, volume, and lab-specific workflow, and exclude library preparation, exome capture reagents, and informatics.

Figures above are estimates derived from publicly available vendor specifications, pricing statements, and Element-published case data as of mid-2026 (Illumina, PacBio, Oxford Nanopore, and Element Biosciences product and pricing pages; Genetic Disease – 2022 Cost-Effective Trio Sequencing case study). They reflect sequencing reagents only and will vary with contract pricing, sample volume, and depth strategy.

Real-world factors also play into cost per sample. High-multiplexing on larger flow cells generally lowers per-sample reagent cost, but only if a lab can reliably accumulate enough samples to fill a run in a timely manner. Otherwise, the effective cost per sample rises to cover unused capacity, or turnaround time stretches while samples queue to fill a batch. High-throughput platforms (NovaSeq X Plus, VITARI) suit large, centralized labs that can consistently fill high-capacity flow cells. Benchtop platforms with mid-throughput or more flexible flow cells let lower-volume rare disease programs run a trio or a handful of exomes on their own schedule rather than waiting to fill a large flow cell, trading a per-sample cost premium for shorter real-world turnaround. That said, the AVITI24 system is noted for cost-effectiveness among mid-throughput systems, with a list price, reagent-only cost per genome that approximates that of the higher throughput NovaSeq 6000.

How researchers are using Element for genetic disease research

Accuracy where it matters most: ABC and UltraQ

ABC is designed to deliver high per-base accuracy on PCR-free libraries—a baseline of 99.99% accuracy (Q40)—and the UltraQ™ workflow pushes further by directly addressing two significant sources of error brought in through library prep. An on-instrument USER enzyme treatment digests deaminated DNA fragments before they can generate erroneous data, and “dark cycling” the first cycles of read 2 skips the error-prone end-repair region of each library fragment. On a human T2T reference genome, this produced more than 70% Q50 data (99.999% accuracy) without masking any difficult regions, to Element’s knowledge, the first time any sequencing platform reported that level of accuracy at scale (Kruglyak et al., 2024). Higher raw accuracy reduces noise and improves confidence in de novo mutation calls, which by definition cannot be confirmed with population-frequency support or cross-validated against parental sequences.

Evidence in homopolymers and repeats

A 2025 Nature study sequenced a four-generation, 28-member CEPH pedigree using five sequencing technologies—PacBio HiFi, Illumina, Element AVITI, ultralong Oxford Nanopore, and Strand-seq—to characterize human de novo mutation rates directly. Across a random sample of 1,000 homozygous homopolymer loci, the study found that 99.5% of Element reads perfectly supported the genotyped allele size, compared with 93.5% of Illumina reads, confirming an accuracy advantage in a genomic context relevant to de novo tandem-repeat mutation discovery. The same study estimated 98–206 de novo mutations per generational transmission and a short tandem repeat mutation rate of roughly 5.5 × 10⁻⁶ per locus per haplotype per generation (Porubsky, Eichler et al, 2025).

A related bioRxiv study assembling telomere-to-telomere genomes across a three-generation pedigree found that Element reads contributed 6,231 corrective edits to the reference assembly, more than double the 2,558 contributed by HiFi, with 95% of Element-sourced edits landing in microsatellite and STR regions (Cechova, Miga et al, 2025).

Taken together with the diploid genome benchmark and SMaHT Network findings described above, ABC sequencing demonstrates an improved ability to accurately call an important subset of hard-to-sequence regions, homopolymers and tandem-repeats, where de novo mutations concentrate.

Real-world adoption

Inherited retinal degeneration gene-hunting

A 2024 study sequenced 10 pedigrees (12 samples) affected by inherited retinal degeneration on AVITI, using both PCR-free and PCR-based whole genome sequencing workflows. Causal variants were identified in 5 of 10 pedigrees, a 50% yield consistent with published literature for this disease category, across the genes ABCA4, CERKL, MAK, PEX6, and RDH12. In several pedigrees, sequencing just the affected individual and confirming segregation in relatives resolved the entire family without sequencing everyone, an efficient and cost-effective study design (Biswas et al., 2024).

Early-onset psychosis in an underserved population

At ASHG 2024, Catherine Brownstein (Boston Children’s Hospital/Harvard Medical School) presented whole genome sequencing of 867 children and adolescents and their families on Element AVITI Systems, recruited from a pediatric psychiatric hospital in Mexico City, a population historically underrepresented in genomic research. Analysis of the first 24 genomes found that 9 (38%) carried a variant with an established neuropsychiatric association, across genes and findings including CUL3, TAOK1, TRIO, an NRXN1 deletion, a FAR1 deletion, LZTR1, CPOX, an Xp22.31 deletion, and mosaic XXY —an early demonstration that WGS can surface actionable Mendelian findings even in a category, early-onset psychosis, not traditionally understood as a genetic disease first.

Frequently asked questions

What sequencing approach has the highest yield for rare genetic disease?

Trio sequencing — sequencing a proband together with both biological parents — consistently outperforms singleton sequencing because it lets researchers directly identify de novo variants and resolve inheritance patterns. Whole genome sequencing generally adds further yield over exome sequencing alone.

Why do so many rare disease cases remain unsolved after whole genome sequencing?

A large share of undiscovered causal variants sit in genomic regions — homopolymers, tandem repeats, segmental duplications — that short-read sequencing and standard reference genomes have historically struggled to read and map accurately.

Do I need long-read sequencing to discover a rare genetic variant?

Short-read trio WGS or WES remains the practical, scalable first-line approach for most rare disease research. Long-read sequencing adds meaningful incremental diagnostic yield — particularly for structural variants and repeat expansions — in cases that remain unsolved after short-read sequencing, but at a substantially higher reagent cost per sample.

How much does trio sequencing cost?

Reagent cost per WGS trio ranges from roughly $300 on Element’s pre-order VITARI™ platform to $500–600 on Illumina NovaSeq X Plus, $1,260-1680 on Element AVITI24 or PacBio Revio, and $1,000–2,000 or more on Oxford Nanopore PromethION, depending on the depth strategy used for parental samples and current list pricing. These figures cover sequencing reagents only.

What is a de novo mutation and why does it matter for rare disease?

A de novo mutation is a new genetic change present in a child but absent from both parents, typically arising from a replication or repair error in a parent’s germ cells or very early in embryonic development. De novo mutations are a leading cause of severe, early-onset genetic disease because trio sequencing is the most direct way to identify them.

Explore the data and sources

Albuquerue et al., “Diagnostic Yield of Genome Sequencing Versus Exome Sequencing in Pediatric Patients With Rare Phenotypes: A Systematic Review and Meta-Analysis,” AJMG Part A, 2025

Biswas, Villanueva, Krajacich, Kruglyak, Ayyagari et al., “Avidity sequencing of whole genomes from retinal degeneration pedigrees identifies causal variants,” PLOS ONE, 2024

Cechova, Miga et al., “Complete genomes of a multi-generational pedigree to expand studies of genetic and epigenetic inheritance,” bioRxiv preprint, 2025

Del Gobbo and Boycott, “The additional diagnostic yield of long-read sequencing in undiagnosed rare diseases”, Genome Research, 2025

Dwarshuis, Olson, Wagner, Zook et al., “The GIAB genomic stratifications resource for human reference genomes,” Nature Communications, 2024

Hansen, Zook, Phillippy et al., “A complete diploid human genome benchmark for personalized genomics,” bioRxiv preprint, 2025

Jang, Gardiner, Abyzov et al., “Accurate detection of sub-1% frequency somatic mutations by whole genome sequencing,” SMaHT Network preprint, 2025

Kruglyak, Altomare, Ambroso et al., UltraQ chemistry preprint, bioRxiv, 2024

Lai, et al., “The application of whole-exome sequencing in the early diagnosis of rare genetic diseases in children: a study from Southeastern China,” Frontiers in Pediatrics, 2024

Nurchis et al., “Whole genome sequencing diagnostic yield for paediatric patients with suspected genetic disorders: systematic review, meta-analysis, and GRADE assessment”, Arch Public Health, 2023

Porubsky, Dashnow, Sasani, Eichler et al., “Human de novo mutation rates from a four-generation pedigree reference,” Nature, 2025

Shoag, et al. “Direct measurement of the male germline mutation rate in individuals using sequential sperm samples” Nature Communications, 2025

Wakap et al., “Estimating cumulative point prevalence of rare diseases: analysis of the Orphanet database”, European Journal of Human Genetics, 2019.

NIH Genetic and Rare Diseases Information Center (GARD)

Element Biosciences, AVITI24 System and Specifications

Element Biosciences, VITARI System

The Element Biosciences AVITI, AVITI24, and VITARI platforms, and the UltraQ workflow described above, are intended for research use only and are not intended for use in diagnostic procedures. Cost, throughput, and multiplexing figures are based on publicly available vendor specifications and pricing statements as of mid-2026 and reflect sequencing reagents only unless otherwise noted. VITARI is a pre-commercial, pre-order product; performance and pricing claims are forward-looking and subject to change prior to commercial launch. Preprints cited above have not yet completed peer review. Third-party product and platform names are trademarks of their respective owners.