Abstract
Next-generation sequencing (NGS) technologies have achieved remarkable success in both biological research and clinical applications. However, in recent years, performance improvements have slowed due to fundamental limitations imposed by Poiseuille fluid dynamics in flow cells, which we overcome using Couette flow. Here we show NGS by roll-to-roll fluidics (r2r-fl), a cost-effective approach compatible with flexible biochip sizes. r2r-fl is a practical implementation of plane Couette flow, with up to 85-fold lower reagent consumption (US$0.16 per gigabase pair), rinsing times under 2 s and a reduction in paired-end 100-base pair sequencing turnaround from days to less than 12 h. The method maintains over 99.9% precision and 99.3% sensitivity of single nucleotide polymorphisms in the human genome, 99.9% mapping rate for Escherichia coli, and minimal nucleotide substitutions, deletions or insertions in the severe acute respiratory syndrome coronavirus 2 alpha strain. By lowering cost and time, r2r-fl enables rapid, scalable NGS for pathogen detection, cancer diagnostics and genetic disease profiling.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$32.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 print issues and online access
$259.00 per year
only $21.58 per issue
Buy this article
- Purchase on SpringerLink
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
Similar content being viewed by others
Data availability
The Illumina sequencing quality data used in this work for comparison are publicly available through BaseSpace (register in https://basespace.illumina.com/, and import runs of ‘NovaSeqX_WHGS_TruSeqPF_NA12878’, ‘COVIDSeq Test Demo - NovaSeq S4’ and ‘Nextera Repeat (600pM)- 2×151 DI’). Sequencing data and analysis are available via https://doi.org/10.57760/sciencedb.18851. Source data are provided with this paper.
References
Margulies, M. et al. Genome sequencing in microfabricated high-density picolitre reactors. Nature 437, 376–380 (2005).
Holt, R. A. & Jones, S. J. The new paradigm of flow cell sequencing. Genome Res. 18, 839–846 (2008).
Uhlen, M. & Quake, S. R. Sequential sequencing by synthesis and the next-generation sequencing revolution. Trends Biotechnol. 41, 1565–1572 (2023).
Bennett, H. M., Stephenson, W., Rose, C. M. & Darmanis, S. Single-cell proteomics enabled by next-generation sequencing or mass spectrometry. Nat. Methods 20, 363–374 (2023).
Gawad, C., Koh, W. & Quake, S. R. Single-cell genome sequencing: current state of the science. Nat. Rev. Genet. 17, 175–188 (2016).
Church, A. J. et al. Molecular profiling identifies targeted therapy opportunities in pediatric solid cancer. Nat. Med. 28, 1581–1589 (2022).
Bressan, D., Battistoni, G. & Hannon, G. J. J. S. The dawn of spatial omics. Science 381, eabq4964 (2023).
Lewis, S. M. et al. Spatial omics and multiplexed imaging to explore cancer biology. Nat. Methods 18, 997–1012 (2021).
Fuller, C. W. et al. The challenges of sequencing by synthesis. Nat. Biotechnol. 27, 1013–1023 (2009).
Balasubramanian, S. Sequencing nucleic acids: from chemistry to medicine. Chem. Commun. 47, 7281–7286 (2011).
Rodriguez, R. & Krishnan, Y. The chemistry of next-generation sequencing. Nat. Biotechnol. 41, 1709–1715 (2023).
Case, D. J., Liu, Y., Kiss, I. Z., Angilella, J. -R. & Motter, A. E. Braess’s paradox and programmable behaviour in microfluidic networks. Nature 574, 647–652 (2019).
Bruus, H. Theoretical Microfluidics (Oxford University Press, 2007).
Eisenstein, M. Innovative technologies crowd the short-read sequencing market. Nature 614, 798–800 (2023).
Mohammed, M. I., Haswell, S. & Gibson, I. Lab-on-a-chip or Chip-in-a-lab: challenges of Commercialization Lost in Translation. Procedia Technol. 20, 54–59 (2015).
Schwarze, K. et al. The complete costs of genome sequencing: a microcosting study in cancer and rare diseases from a single center in the United Kingdom. Genet. Med. 22, 85–94 (2020).
Chai, J. H. et al. Cost-benefit analysis of introducing next-generation sequencing (metagenomic) pathogen testing in the setting of pyrexia of unknown origin. PLoS ONE 13, e0194648 (2018).
Schermelleh, L. et al. Super-resolution microscopy demystified. Nat. Cell Biol. 21, 72–84 (2019).
Heintzmann, R. & Huser, T. Super-resolution structured illumination microscopy. Chem. Rev. 117, 13890–13908 (2017).
Ma, W. et al. Gene sequencing reaction device, gene sequencing system, and gene sequencing reaction method. US11857973B2 (2024).
Beckett, N. & Caswell, N. Implementing barriers for controlled environments during sample processing and detection. US20210354126A1 (2022).
Barbee, K. et al. Methods for biological sample processing and analysis. US11499962B2 (2019).
Pennisi, E. $100 genome? New DNA sequencers could be a ‘game changer’ for biology, medicine. Science 376, 1257–1258 (2022).
Reynolds, O. IV. On the theory of lubrication and its application to Mr. Beauchamp tower’s experiments, including an experimental determination of the viscosity of olive oil. Philos. Trans. R. Soc. London 157–234 (1886).
Reichardt, H. Über die Umströmung zylindrischer Körper in einer geradlinigen Couetteströmung. Mitt. Max-Planck-Institut f. Strömungsforschung, Göttingen, No. 9 (1954).
Tillmark, N. & Alfredsson, P. H. Experiments on transition in plane Couette flow. J. Fluid Mech. 235, 89–102 (1992).
Krug, D., Lüthi, B., Seybold, H., Holzner, M. & Tsinober, A. 3D-PTV measurements in a plane Couette flow. Exp. Fluids 52, 1349–1360 (2012).
Piau, J., Bremond, M., Couette, J. & Piau, M. Maurice Couette, one of the founders of rheology. Rheol. Acta 33, 357–368 (1994).
Fardin, M., Perge, C. & Taberlet, N. ‘The hydrogen atom of fluid dynamics’—introduction to the Taylor–Couette flow for soft matter scientists. Soft Matter 10, 3523–3535 (2014).
Mardis, E. R. DNA sequencing technologies: 2006–2016. Nat. Protoc. 12, 213–218 (2017).
Ali, M. M. et al. Rolling circle amplification: a versatile tool for chemical biology, materials science and medicine. Chem. Soc. Rev. 43, 3324–3341 (2014).
Li, H. & Durbin, R. Fast and accurate long-read alignment with Burrows–Wheeler transform. Bioinformatics 26, 589–595 (2010).
Krusche, P. et al. Best practices for benchmarking germline small-variant calls in human genomes. Nat. Biotechnol. 37, 555–560 (2019).
Xiao, K. et al. Isolation of SARS-CoV-2-related coronavirus from Malayan pangolins. Nature 583, 286–289 (2020).
Foox, J. et al. Performance assessment of DNA sequencing platforms in the ABRF next-generation sequencing study. Nat. Biotechnol. 39, 1129–1140 (2021).
Zheng, W. et al. High-throughput, single-microbe genomics with strain resolution, applied to a human gut microbiome. Science 376, eabm1483 (2022).
Stoeckius, M. et al. Simultaneous epitope and transcriptome measurement in single cells. Nat. Methods 14, 865–868 (2017).
Chiu, C. Y. & Miller, S. A. Clinical metagenomics. Nat. Rev. Genet. 20, 341–355 (2019).
Zhu, N. et al. A novel coronavirus from patients with pneumonia in China, 2019. N. Engl. J. Med. 382, 727–733 (2020).
Nogrady, B. How cancer genomics is transforming diagnosis and treatment. Nature 579, S10–S11 (2020).
Saunders, C. J. et al. Rapid whole-genome sequencing for genetic disease diagnosis in neonatal intensive care units. Sci. Transl. Med. 4, 154ra135 (2012).
Sazonovs, A. et al. Large-scale sequencing identifies multiple genes and rare variants associated with Crohn’s disease susceptibility. Nat. Genet. 54, 1275–1283 (2022).
Drmanac, R. et al. Human genome sequencing using unchained base reads on self-assembling DNA nanoarrays. Science 327, 78–81 (2010).
Li, Z. et al. DNB-based on-chip motif finding: a high-throughput method to profile different types of protein-DNA interactions. Sci. Adv. 6, eabb3350 (2020).
Aksamentov, I., Roemer, C., Hodcroft, E. B. & Neher, R. A. Nextclade: clade assignment, mutation calling and quality control for viral genomes. J. Open Source Softw. 6, 3773 (2021).
Chen, Y. et al. SOAPnuke: a MapReduce acceleration-supported software for integrated quality control and preprocessing of high-throughput sequencing data. Gigascience 7, gix120 (2018).
Jung, Y. & Han, D. BWA-MEME: BWA-MEM emulated with a machine learning approach. Bioinfomratics 38, 2404–2413 (2021).
Danecek, P. et al. Twelve years of SAMtools and BCFtools. Gigascience 10, giab008 (2021).
‘Picard Toolkit.’ GitHub https://broadinstitute.github.io/picard (2019).
McKenna, A. et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 20, 1297–1303 (2010).
Abyzov, A., Urban, A. E., Snyder, M. & Gerstein, M. CNVnator: an approach to discover, genotype, and characterize typical and atypical CNVs from family and population genome sequencing. Genome Res. 21, 974–984 (2011).
Chen, K. et al. BreakDancer: an algorithm for high-resolution mapping of genomic structural variation. Nat. Methods 6, 677–681 (2009).
Acknowledgements
This work is supported by the National Key R&D Program of China (2020YFA0210800, led by H.Y., and 2020YFA0709900, led by Q.C.); BGI research and MGI Tech Co.; the China Postdoctoral Science Foundation (2019M650215) awarded to Y.Q.; China National Gene Bank; the National Natural Science Foundation of China (22027805, 22334004 and 22421002, led by H.Y., and 62134003 and U2441224, led by Q.C.); and the Natural Science Foundation of Fujian Province (2022J06008, led by Q.C.).
We thank C. Xing and S. Lv for discussion and suggestion in engineering issues; M. Gong, J. Yang and X. Huang for instruction and guidance in biochemical processes; M. He, M. Cao, Z. Zhen and Y. Huang for discussion and calibration of imaging systems; H. Zhang for heating subsystem setup; Z. Niu for liquid dynamic discussion; W. Li and D. Nie for biochip preparation and loading; H. Xu for thickness determination; J. Wang for discussion of model comparison; and J. Liu, H. Jiang and F. Mu for their funding support of this project. The people mentioned above are from MGI Tech Shenzhen.
We thank L. Shao for bioinformatic analysis, M. Li for writing ‘Data quality’ in the Methods, M. Ni and D. Wei for providing DNBSEQ-TX-related resources, as well as X. Xu and M. Shen for their funding support of this project. The people mentioned above are from BGI Research Shenzhen.
Author information
Authors and Affiliations
Contributions
Y.Q. initiated the project, conceived the idea and led the development team. S.K. wrote several draft versions of the paper and cover letters. Y.L. monitored and debugged the first prototype. S.Y., J.L., K.C., J.J., J.Z., Q.C., J.L. and H.C. contributed to building the second prototype on different modules. F.W. helped with a CFD simulation figure. W.L., X.W. and D.L. programmed and tested the control system for the second prototype. Z.D. analyzed the data for NA12878, E.coli and COVID-19 samples. Z.Y. helped with data collection for characterization and figure drawing. W.Z. helped with construction of the whole system including first and second prototype. Y.H. and X.P. helped with material characterization about chemistry. T.M.H. and H.Y. helped with paper writing, making figures and videos. S.Z. helped with imaging on DNBSEQ-T10. R.M. oversaw electrical engineering and coding of the project. Z.L. oversaw the mechanical engineering of the project. X.Y. and X.H helped with biochemical experiments. Z.Y. and Y.Z. helped with the data analysis. Z.N. helped with fluid engineering of the prototypes. H.L. and T.M.H. helped with CFD, C.X. helped with DNB loading of the biochips and X.Z helped review and refine the wording of the paper. Y.Q., S.K., Q.C. and T.M.H. discussed the paper. The people in BGI Shenzhen have conducted the project, designed and tested two iterations of prototypes, with assistance of people in MGI. X.Y. and Y.G. were interns of BGI Shenzhen. X.L., X.Q. and X.Z. helped with paper writing and potential application discussion. All authors discussed the results.
Corresponding authors
Ethics declarations
Competing interests
BGI Research and MGI Co. are developing r2r-fl technology for NGS sequencers. Several authors are employees of the involved institutions: Y.Q., J.L., Y.L., K.C., J.L., J.Z., H.C., F.W., W.L., D.L., X.Y., X.W., S.Y., X. H., C.X., W.Z. and Y.G. are employees of BGI Research, Shenzhen. S.Z., H.L., Z.D., R.M., Z.L. and Z.N. are employees of MGI Co. Patents have been applied for based on this technology. The other authors declare no competing interests.
Peer review
Peer review information
Nature Methods thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editor: Lei Tang, in collaboration with the Nature Methods team.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 Structure, flows, and meshing of pressure-driven microfluidic flow cells.
a, Photos of flow cells by Illumina: the high-throughput four-channel S4 flow cell for NovaSeq 6000, the medium-throughput four-channel flow cell for NextSeq 550, and the small-throughput single-channel flow cell for MiSeq. BGI’s fastest flow cell, DNBSEQ-T7 is also shown (cover glass breakage occurs above 90 kPa at the zones indicated with white arrows). Images courtesy of Illumina, Inc and MGI Tech Co. b, Structure (not to scale) of one lane for a typical flow cell (DNBSEQ-G400), its side view (c) and frontal view (d) of the flow cell with inlet and outlet tubing. The inset shows the gap height. e, The velocity field for the Poiseuille flow in the biochip chamber, which is approaching 0 when it is close to the walls. f-h, Meshing of the pressure-driven microfluidic flow cell for computational fluid dynamics (CFD) calculations. Top view (f) and side view (g) of the mesh for a flow cell biochip and its entrance tubing (h) with 30-cm length and 0.8-mm diameter.
Extended Data Fig. 2 Importance of contact angle of PET belt and biochip.
a, Plasma treatment generates hydrophilic moieties. b, Raman spectra before and after plasma treatment showing an increased O-H vibration. c, List of surface tensions of the different reagents and wash buffers. d, Hand-drawn straight lines using an Arcotest pen with 36 mN/m ink on the biochip, which immediately beads up and indicates poor wetting, while on the PET belt lines with 36, 38 and 42 mN/m ink remain continuous, whereas those with higher surface energies, 46 and 50 mN/m, bead up. The biochip surface is fabricated with a matrix of hydrophilic nanowells surrounded by hydrophobic HMDS (hexamethyldisilane), resulting in an overall poor-wetting surface. e, Contact angle measurements of water and wash buffer 2 on the treated biochip and PET surfaces.
Extended Data Fig. 3 Assessment of height fluctuations and potential damage to the biochip by the moving PET belt.
a, Experimental setup for determining vertical fluctuations of PET belt using Keyence SI-T80 by detecting air layer thickness between the PET belt and the vacuum box. b, Height fluctuations of the PET belt moving at 0.32 m/s by measurement of air layer thickness between the PET belt and the vacuum box. c–e, three sensors (A, B, Gap) show a constant 10 µm of liquid thickness before/after the biochip, and 20 µm on the biochip. f-g, side-view image of 10 µm liquid film on the PET belt by optical microscopy. h, Thickness measurement of the liquid film in the width direction using a Keyence SI-T80. i, Photograph of a DNA sequencing biochip (7×7 cm2) with a handle attachment at the top. j, k, Fluorescent images of the biochip where the gap heights to the moving PET belt are 10 μm and 0 μm, respectively. Note the scratched (red circled) regions for 0 µm.
Extended Data Fig. 4 Workflow for DNA nanoballs (DNBs) loading onto nanowells of biochip and DNA sequencing.
a, Negatively charged DNBs from solution selectively adhere onto positively charged attachment points on the biochip, forming the DNBs array for sequencing. b, A picture of a biochip (7 cm × 7 cm) containing an array of positively charged nanowells and negatively charged DNBs (diameter ~220 nm, same as the nanowell). c, Atomic force microscopic study of the biochip surface before and after loading. Left panels show top view and sampling positions, and right panels show height profiles. After loading and dehydration, the DNBs’ height shrinks to 5 nm and are protected by the donut shape of the nanowell edges. d, Depictions of a layer of wash buffers flushing unbonded DNBs from the biochip surface, resulting in a completely loaded biochip as shown in (e). f, Depiction of a self-termination reaction for sequencing that single DNA strand extension by one dATP. See also Supporting Video 2.
Extended Data Fig. 5 COMSOL simulations investigating the initiation of pCF on a completely dry biochip surface.
a, A 10 µm layer of water (reagent) with a thicker leading edge approaches the biochip. b, Upon contact, wetting occurs. c, The belt continuously shears the water layer to the right, leading to detachment. d, A thin film is formed, extending well beyond the biochip. e, PCF is properly established; f, A steady-state condition develops, characterized by the attachment and detachment of the water phase.
Extended Data Fig. 6 Velocity field and computational fluid dynamics (CFD) simulations of reagent replacement in roll-to-roll fluidics.
a, Front view of the r2r-fl cell, where α and β are contact angles of reagent to the PET belt and biochip, respectively. It shows that surface tension holds the liquid inside the chamber thus side walls are not necessary. b, Side view velocity field in the chamber c-h, Both pressure and width of the concentration front increase with increasing flow velocities. The pre-coating reagent thickness and gap height are 10 μm and 20 μm, respectively. The chamber length is 7 cm. The average flow velocities are \({V}_{a}=\,\)0.015, 0.15, and 0.35 m/s in (c) and (d), (e) and (f), (g) and (h), respectively.
Extended Data Fig. 7 Second generation roll-to-roll fluidics setup with larger biochip area.
The biochip area in this device is 1.35 × 1.2 m, and can be loaded with > 100 biochips (7×7 cm2). The robotic arm has just loaded a biochip. The main text is focussed on the manipulation of a single biochip as shown in Fig. 1, to ensure better understanding.
Extended Data Fig. 8 Comparation of data quality of E. coli sequencing achieved with r2r-fl prototype and Illumina’s HiSeq 4000 flow cell.
a, Heatmap of Q30 for the whole biochip, where each green dot indicates a field of view of microscope with high data quality (Q30 > 80% before filtration), and the red regions around the edges indicate low quality. A selected region used for image registration is enclosed in the blue circle. b, Zoomed-in fluorescent image from channel A, where each bright spot indicates a DNB with newly added adenine nucleotides in the current bioreaction cycle. In the arrows’ position, there are track lines that with less DNBs divides different units of square region and help registration and alignment of images from different cycles. c, Q30 heatmap of the flow cell (HiSeq 4000). d, Fluorescent intensity plots for the A, T, C, and G channels for the r2r-fl cell and the flow cell, respectively. Comparison of working status for each cycle of the r2r-fl cell and HiSeq 4000 flow cell, including lag (e), run on (f), base distribution (g) and filtered Q30 (h), and estimated error rate (i). Flow cell data sourced from basespace.illumina.com.
Extended Data Fig. 9 Comparation of data quality of SARS-CoV-2 (the Alpha strain) sequencing achieved with our r2r-fl prototype and the NovaSeq 6000 flow cell.
a, Heatmap of Q30 for the whole biochip, where each green dot indicates a field of view (FOV) of microscope with high data quality (Q30 > 80% before filtration), and the red regions around the edges indicate poor quality for base calling algorithm. A selected region used for image registration is enclosed in a dash-line circle. b, A zoomed-in fluorescent image from A channel, where each bright spot indicates a DNB that adding Adenine nucleotides in current cycle of bioreaction. Arrows indicating track lines that with less DNBs divides different units of square region and help registration and alignment of images from different cycles. c, Q30 heatmap of the flow cell (Novaseq 6000). d-h, Comparison of the fluorescent intensity, lag, run on, base distribution, filtered average Q30 for the first 50 cycles for the flow cell and r2r-fl cell are shown in (d), (e), (f), (g), and (h), respectively. Flow cell data sourced from basespace.illumina.com.
Extended Data Fig. 10 Phylogeny of the reference SARS-Cov-2 positive sample using the r2r-fl platform.
The star indicates the genome sequence result produced from our platform, which shows exactly no nucleotide and amino acid deletion, insertion or substitutions compared to literature data of the SARS-CoV-2 wild-type reference genome (Genbank: MN908947). Please refer to Supplementary Table 12 for data.
Supplementary information
Supplementary Information
Supplementary Figs. 1–3, Discussion and Tables 1–12.
Supplementary Video 1
Computational fluid dynamics simulations of plane Couette flow in r2r-fl and a Poiseuille flow in a flow cell. The first part illustrates Finite Element Methods simulation of wetting and detachment of water onto the biochip. A thicker leading edge (due to the PET belt acceleration) approaches the biochip. Wetting and capillary action draws the liquid into the gap. The shearing action of the belt keeps pulling the liquid to the right side. See Extended Data Fig. 5 for annotated snapshots with additional details. The second part illustrates the reagent replacement for r2r-fl, showing the relative concentration of dNTP in the gap between the PET belt and the biochip surface. The diffusion coefficient is 10−9 m2s−1, the gap height between the PET belt and biochip is 20 μm, the average flow velocity is 0.16 m s−1and the length of the biochip is 7 cm. Full replacement (concentration > 99.9%) is observed at approximately 1 s. In the third part of the video, the reagent replacement process of a traditional single-lane flow cell with tubing (0.8-mm inner diameter and 30-cm length) is shown, where full replacement is observed at approximately 30 s. In the fourth and fifth part of the video, they show experiments of reagent sourcing using r2r-fl and a flow cell, respectively.
Supplementary Video 2
Experimental workflow of DNA sequencing by r2f-fl. The first part illustrates the working process of the system. The second and third parts demonstrate the biochemical processes for library preparation and DNB loading, and a BGI version of a DNA sequencing cycle for SBS bioreactions, respectively. The fourth part is a video showing the biochip transfer from buffer storage to the imaging system using a robotic arm after the SBS bioreaction.
Source data
Source Data Fig. 1
Combined source data for Fig. 1.
Source Data Fig. 2
Combined source data for Fig. 2.
Source Data Fig. 3
Combined source data for Fig. 3.
Source Data Fig. 4
Combined source data for Fig. 4.
Source Data Extended Data Fig. 2/Table 2
Combined source data for Extended Data Fig. 2.
Source Data Extended Data Fig. 3/Table 3
Combined source data for Extended Data Fig. 3.
Source Data Extended Data Fig. 4/Table 4
Combined source data for Extended Data Fig. 4.
Source Data Extended Data Fig. 8/Table 8
Combined source data for Extended Data Figs. 8 and 9.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Qin, Y., Koehler, S.A., Ling, Y. et al. Fast, cost-effective and flexible DNA sequencing by roll-to-roll fluidics. Nat Methods 22, 1662–1668 (2025). https://doi.org/10.1038/s41592-025-02730-2
Received:
Accepted:
Published:
Issue date:
DOI: https://doi.org/10.1038/s41592-025-02730-2
