Ultima Genomics UG 100 platform vs. Illumina NovaSeq X Series whole-genome sequencing benchmarking

Published October 7, 2025

Key takeaways

An internal analysis evaluated the NovaSeq X Series against the Ultima Genomics UG 100 platform. This evaluation found that the NovaSeq X Series:

  • Measures accuracy of WGS against the full NIST v4.2.1 benchmark, while Ultima Genomics measures WGS accuracy using the UG 100 platform against a subset that excludes 4.2% of the genome
  • Results in 6× fewer SNV errors and 22× fewer indel errors than the UG 100 platform, when assessed against the full NIST v4.2.1 benchmark
  • Calls approximately 180,000 more SNVs and 270,000 more indels analyzing the full NIST v4.2.1 benchmark compared to the UG 100 platform analyzing only the Ultima Genomics "high confidence region"
  • Maintains high coverage and variant calling accuracy in repetitive genomic regions, including GC-rich sequences and homopolymers longer than 10 base pairs, compared to the UG 100 platform
  • Provides more insights into biologically relevant genes compared to the UG 100 platform

Ultima Genomics vs. Illumina sequencing platforms

Illumina has maintained a relentless commitment to innovating next-generation sequencing (NGS) capabilities and building future methods.1 The NovaSeq X Series enables data-intensive applications at production scale to empower scientists to make new discoveries. Over the past decade, emerging companies like Ultima Genomics have introduced NGS platforms. In 2024, Ultima Genomics launched the UG 100 sequencing platform with bold claims, including that it is capable of sequencing 20,000 human genomes per year with industry-leading accuracy in variant calling.2 To evaluate these performance and accuracy claims, Illumina conducted a comparative analysis of the NovaSeq X Series and the UG 100 platform. The results of this analysis demonstrate that the NovaSeq X Series delivers more accurate variant calling, provides a higher-quality, more comprehensive genome, and enables more insights into the molecular mechanisms of disease than the UG 100 platform.

Data sources for comparing Ultima Genomics to Illumina

We generated Illumina whole-genome sequencing (WGS) data on the NovaSeq X Plus System using the NovaSeq X Series 10B Reagent Kit, followed by DRAGEN v4.3 secondary analysis. Data included duplicates and downsampled to 35× coverage depth. We sourced Ultima Genomics WGS data from a publicly available data set released in a variant call format (VCF) that had been generated on the UG 100 platform at 40× coverage depth excluding duplicates, and analyzed using DeepVariant software by Ultima Genomics.

Higher variant calling accuracy with the NovaSeq X Series

The National Institute of Standards and Technology (NIST) v4.2.1 benchmark for the Genome in a Bottle (GIAB) HG002 reference genome is used to assess the accuracy and performance of WGS and variant calling analysis tools.3,4 The NIST v4.2.1 benchmark includes high-confidence genotype calls for single-nucleotide variants (SNVs), insertions and deletions (indels), and structural variants (SVs), along with challenging regions consisting of segmental duplications, low-mappability regions, or repetitive sequences that can include genes of biological interest.4,5  

Illumina measures the accuracy of WGS obtained with the NovaSeq X Series with DRAGEN secondary analysis against the full NIST v4.2.1 benchmark, including all genomic regions (Figure 1).6 Conversely, Ultima Genomics measures the accuracy of WGS with the UG 100 platform against a defined subset of the NIST v4.2.1 benchmark referred to as the UG “high-confidence region” (HCR). The UG HCR masks regions with low performance on the UG 100 platform, including homopolymers, repetitive sequences, and areas with low coverage (Figure 1).7 Importantly, the regions excluded from the UG HCR include 4.2% of NIST benchmark variants.

Figure 1: Ultima Genomics HCR masks 4.2% of the genome─WGS accuracy on the NovaSeq X Series is measured against the full NIST v4.2.1 benchmark, while the UG HCR excludes 4.2% of the genome from analysis.

We compared the variant calling performance of the NovaSeq X Series against the UG 100 platform. The UG 100 platform resulted in 6× more SNV errors and 22× more indel errors than the NovaSeq X Series when assessed against NIST v4.2.1 all benchmark regions (Figure 2). Variant calling errors are defined as the number of false positives (a variant is called that is not present in the reference genome) and false negatives (a variant present in the reference genome is not called). The UG HCR excludes ~450,000 variants (16% of genomic variants) from the full NIST v4.2.1 benchmark, resulting in up to 5% fewer SNV and 37% fewer indel calls on the UG 100 platform due to their exclusion from analysis of WGS (Figure 3). Compared to the NovaSeq X Series, the UG 100 platform results in 3% fewer SNV and 21% fewer indel calls when using the full NIST v4.2.1 benchmark (Figure 3).

Figure 2: Significantly more errors in variant calling with Ultima Genomics─The UG 100 platform produces 6× and 22× more errors in variant calling for SNVs and indels, respectively, compared to the NovaSeq X Series, assessed against NIST v4.2.1 all benchmark regions. Errors are defined as the combined number of false positives and false negatives.

Figure 3: Ultima Genomics HCR masks key variants─Limiting analysis of WGS on the UG 100 platform to the UG HCR subset of the NIST v4.2.1 benchmark results in fewer variant calls for SNVs and indels compared to the NovaSeq X Series.

We also evaluated the performance of both systems in challenging GC-rich regions. The results show that relative genome coverage with the UG 100 platform dropped significantly in mid-to-high GC-rich regions, compared to the NovaSeq X Series (Figure 4). The lack of coverage in these regions could exclude genes with known associations with disease from analysis and interpretation.

Figure 4: Loss of coverage in GC-rich regions with Ultima Genomics─The UG 100 platform loses coverage in repetitive, GC-rich regions of the genome compared to the NovaSeq X Series.

Increased genome coverage with the NovaSeq X Series

Illumina sequencing on the NovaSeq X Series delivers a comprehensive, high-quality genome that accurately calls 99.94% of SNVs,* 97% of copy number variants (CNVs), 88% of SVs,ǂ 95.2% of short tandem repeats (STRs),§ at least 13 targeted gene callers for medically relevant genes with 98% concordance, and human leukocyte antigen (HLA) typing with 94% concordance.**

* Compared against HG002 NIST v4.2.1 with rtg vcfeval.
Compared against HG002 NIST v0.6 (hg38) with wittyer v0.4.1.
ǂ Compared against HG002 NIST T2T Q100 v1.1_v0.019 (hg38) with Truvari v.4.2.2.
§ 28 STR loci included in evaluation across 359 samples (157 samples with expansions in at least one locus).
Concordance metrics against orthogonal technology sourced from published technical papers.
** Compared against T1K over 3202 diverse WGS samples from The 1000 Genomes Project.

 

In contrast, Ultima Genomics masks performance deficits by assessing the accuracy of WGS with the UG 100 relying on the UG HCR, which excludes regions of the NIST v4.2.1 benchmark that exhibit unreliable performance and confidence using Ultima Genomics sequencing technology. The extensive number of variants (about 450,000) not called by limiting analysis to the UG HCR (Figure 3) underscores the uncertainty about what data could be missed.

 

The regions excluded by the UG HCR amount to 4.2% of the genome, including 2.3% of the exome, and 1.0% of ClinVar variants. These excluded regions also account for 5.1% of genomic CNVs and 4.7% of ClinVar CNVs. Together, pathogenic variants in 793 genes are excluded from the UG HCR, limiting insight into the associated diseases. Furthermore, while Ultima Genomics claims high variant calling accuracy for SNVs and indels while sequencing through homopolymers, internal analysis showed that indel accuracy with the UG 100 platform decreased significantly with homopolymers longer than 10 base pairs, compared to the NovaSeq X Series (Figure 5). Indeed, the UG HCR excludes homopolymer regions longer than 12 base pairs.7 Given that published studies show that homopolymer repeat length may modulate the expression of nearby genes by altering nucleosome positioning,8 excluding these regions from Ultima Genomics analysis could result in valuable, biologically relevant insights being missed.

 

Figure 5: Reduced indel accuracy with Ultima Genomics─Indel accuracy with the UG 100 platform drops significantly with homopolymers longer than 10 base pairs, compared to the NovaSeq X Series.

More biologically relevant insights with the NovaSeq X Series

Genomic regions excluded from the UG HCR include functionally important loci in disease-related genes and STR-rich genes tied to immune and neurological traits (Table 1). One example is the B3GALT6 gene, which encodes an enzyme crucial to synthesis of glycosaminoglycans (GAGs). Variants in B3GALT6 have been linked to Ehlers-Danlos syndrome.9 Another example is the fragile X messenger ribonucleoprotein 1 (FMR1) gene, which is crucial to normal brain development and reproductive function. Alterations in FMR1 cause fragile X syndrome.10 Both B3GALT6 and FMR1 have GC-rich sequences that resulted in loss of coverage with the UG 100 platform, which likely excluded pathogenic and likely pathogenic variants in these genes (Figure 6A, 6B). Additionally, variants in the BRCA1 tumor suppressor gene have a well-established link to predisposing an individual toward breast cancer development.11,12 However, 1.2% of pathogenic BRCA1 variants are excluded from the UG HCR, and sequencing with the UG 100 platform resulted in significantly more indel calling errors in the BRCA1 gene compared to the NovaSeq X Series (Figure 6C).

Table 1: Biologically relevant genes excluded from the UG HCR

Figure 6: Loss of coverage with Ultima Genomics─Sequencing with the UG 100 platform results in loss of coverage in disease-relevant genes with high GC content, such as (A) B3GALT6 and (B) FMR1, and (C) significantly more indel calling errors in the BRCA1 gene compared to the NovaSeq X Series.

Summary

After evaluating Ultima Genomics and Illumina NGS platforms, the results demonstrate the superior performance of the NovaSeq X Series compared to the UG 100 platform for WGS. The NovaSeq X Series delivers higher accuracy and provides comprehensive coverage across the genome, including challenging regions. The performance of the UG 100 platform is inflated by the Ultima Genomics "high-confidence region," which excludes 4.2% of the genome from analysis where the platform’s performance is poor. The NovaSeq X Series generates the accuracy and genome coverage necessary to deliver biological insights into biologically relevant genes, whereas the UG 100 platform fails to accurately sequence genes with known associations to disease, such as B3GALT6 and BRCA1.

As established in this article, Ultima Genomics excludes various genomic regions from analysis to claim exceptional coverage and accuracy. The capability of the UG 100 platform to discover novel variants or even detect known variants in relevant genes is limited. In contrast, Illumina is a trusted global leader with 27 years of expertise and continues to provide comprehensive support and best-in-class product consistency, setting the standard for NGS solutions. The NovaSeq X Series and DRAGEN secondary analysis deliver accuracy and quality for comprehensive WGS at scale, without the compromises or tradeoffs required by Ultima Genomics technology.

Related links

NovaSeq X Series

DRAGEN secondary analysis

Whole-genome sequencing

Illumina SBS chemistry

Have a question about WGS?

Have questions about how Illumina platforms compare to Ultima Genomics? We can help you determine the total cost of ownership and recommend the best solution for your setup.

Speak to a specialist

References

  1. Illumina. 25 greatest impacts in 25 years: Illumina and the evolution of genomics. illumina.com/company/news-center/feature-articles/25-greatest-impacts-in-25-years--a-look-back-at-illumina-and-the.html. Published April 3, 2023. Accessed July 22, 2025.
  2. Business Wire. Ultima announces UG 100 and reveals disruptive cost and accuracy profile to enable the era of the $100 genome and beyond. businesswire.com/news/home/20240206310974/en/Ultima-Announces-UG-100-and-Reveals-Disruptive-Cost-and-Accuracy-Profile-to-Enable-the-Era-of-the-%24100-Genome-and-Beyond. Published February 6, 2024. Accessed July 22, 2025.
  3. Daniels CA, Abdulkadir A, Cleveland MH, et al. A robust benchmark for detecting low-frequency variants in the HG002 Genome In A Bottle NIST reference material. Preprint. bioRxiv. 2024;2024.12.02.625685. Published 2024 Dec 5. doi:10.1101/2024.12.02.625685
  4. Wagner J, Olson ND, Harris L, et al. Benchmarking challenging small variants with linked and long readsCell Genom. 2022;2(5):100128. doi:10.1016/j.xgen.2022.100128
  5. Wagner J, Olson ND, Harris L, et al. Curated variation benchmarks for challenging medically relevant autosomal genesNat Biotechnol. 2022;40(5):672-680. doi:10.1038/s41587-021-01158-1
  6. Illumina. Unlocking the full potential of Illumina genomes: The journey to enhanced variant calling quality with DRAGEN informatics and high-quality sequencing. https://www.illumina.com/science/genomics-research/articles/CMRG_hg38.html. Accessed July 17, 2025.
  7. Ultima Genomics. UG 100 Sequencing Platform specification sheet. Published 2025. Accessed July 17, 2025.
  8. Fotsing SF, Margoliash J, Wang C, et al. The impact of short tandem repeat variation on gene expressionNat Genet. 2019;51(11):1652-1659. doi:10.1038/s41588-019-0521-9
  9. Van Damme T, Pang X, Guillemyn B, et al. Biallelic B3GALT6 mutations cause spondylodysplastic Ehlers-Danlos syndromeHum Mol Genet. 2018;27(20):3475-3487. doi:10.1093/hmg/ddy234
  10. Grigsby J. The fragile X mental retardation 1 gene (FMR1): historical perspective, phenotypes, mechanism, pathology, and epidemiologyClin Neuropsychol. 2016;30(6):815-833. doi:10.1080/13854046.2016.1184652
  11. Semmler L, Reiter-Brennan C, Klein A. BRCA1 and Breast Cancer: a Review of the Underlying Mechanisms Resulting in the Tissue-Specific Tumorigenesis in Mutation CarriersJ Breast Cancer. 2019;22(1):1-14. doi:10.4048/jbc.2019.22.e6
  12. Fu X, Tan W, Song Q, Pei H, Li J. BRCA1 and Breast Cancer: Molecular Mechanisms and Therapeutic StrategiesFront Cell Dev Biol. 2022;10:813457. Published 2022 Mar 1. doi:10.3389/fcell.2022.813457
  13. Schulte J, Littleton JT. The biological function of the Huntingtin protein and its relevance to Huntington's Disease pathologyCurr Trends Neurol. 2011;5:65-78.
  14. Bishara J, Keens TG, Perez IA. The genetics of congenital central hypoventilation syndrome: clinical implicationsAppl Clin Genet. 2018;11:135-144. Published 2018 Nov 15. doi:10.2147/TACG.S140629
  15. Coffin SL, Durham MA, Nitschke L, et al. Disruption of the ATXN1-CIC complex reveals the role of additional nuclear ATXN1 interactors in spinocerebellar ataxia type 1Neuron. 2023;111(4):481-492.e8. doi:10.1016/j.neuron.2022.11.016
  16. Goswami R, Bello AI, Bean J, et al. The Molecular Basis of Spinocerebellar Ataxia Type 7Front Neurosci. 2022;16:818757. Published 2022 Mar 24. doi:10.3389/fnins.2022.818757
  17. Leardini D, Messelodi D, Muratore E, et al. Role of CBL Mutations in Cancer and Non-Malignant PhenotypeCancers (Basel). 2022;14(3):839. Published 2022 Feb 8. doi:10.3390/cancers14030839
  18. Winnepenninckx B, Debacker K, Ramsay J, et al. CGG-repeat expansion in the DIP2B gene is associated with the fragile site FRA12A on chromosome 12q13.1Am J Hum Genet. 2007;80(2):221-231. doi:10.1086/510800
  19. Barratt KS, Arkell RM. ZIC2 in HoloprosencephalyAdv Exp Med Biol. 2018;1046:269-299. doi:10.1007/978-981-10-7311-3_14
  20. Malerba A, Klein P, Lu-Nguyen N, et al. Established PABPN1 intranuclear inclusions in OPMD muscle can be efficiently reversed by AAV-mediated knockdown and replacement of mutant expanded PABPN1Hum Mol Genet. 2019;28(19):3301-3308. doi:10.1093/hmg/ddz167
  21. Tazelaar GHP, Dekker AM, van Vugt JJFA, et al. Association of NIPA1 repeat expansions with amyotrophic lateral sclerosis in a large international cohortNeurobiol Aging. 2019;74:234.e9-234.e15. doi:10.1016/j.neurobiolaging.2018.09.012
  22. Bavassano C, Eigentler A, Stanika R, et al. Bicistronic CACNA1A Gene Expression in Neurons Derived from Spinocerebellar Ataxia Type 6 Patient-Induced Pluripotent Stem CellsStem Cells Dev. 2017;26(22):1612-1625. doi:10.1089/scd.2017.0085
  23. Deng J, Yu J, Li P, et al. Expansion of GGC Repeat in GIPC1 Is Associated with Oculopharyngodistal MyopathyAm J Hum Genet. 2020;106(6):793-804. doi:10.1016/j.ajhg.2020.04.011
  24. van Kuilenburg ABP, Tarailo-Graovac M, Richmond PA, et al. Glutaminase Deficiency Caused by Short Tandem Repeat Expansion in GLSN Engl J Med. 2019;380(15):1433-1441. doi:10.1056/NEJMoa1806627
  25. Kühnel T, Leitão E, Lunzer R, et al. Repeat Expansions with Small TTTCA Insertions in MARCHF6 Cause Familial Myoclonus without EpilepsyMov Disord. 2025;40(7):1401-1408. doi:10.1002/mds.30192
  26. Goodman FR, Bacchelli C, Brady AF, et al. Novel HOXA13 mutations and the phenotypic spectrum of hand-foot-genital syndromeAm J Hum Genet. 2000;67(1):197-202. doi:10.1086/302961
  27. Kumutpongpanich T, Ogasawara M, Ozaki A, et al. Clinicopathologic Features of Oculopharyngodistal Myopathy With LRP12 CGG Repeat Expansions Compared With Other Oculopharyngodistal Myopathy SubtypesJAMA Neurol. 2021;78(7):853-863. doi:10.1001/jamaneurol.2021.1509
  28. Yu H, Wu J, Cong J, et al. Congenital insensitivity to pain associated with PRDM12 mutation: Two case reports and a literature reviewFront Genet. 2023;14:1139161. Published 2023 Mar 20. doi:10.3389/fgene.2023.1139161