AM‑DMF-Lib-prep: A clinical ready lib-prep-on-a-chip platform with active-matrix digital microfluidics

AM‑DMF-Lib-prep: A clinical ready lib-prep-on-a-chip platform with active-matrix digital microfluidics | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article AM‑DMF-Lib-prep: A clinical ready lib-prep-on-a-chip platform with active-matrix digital microfluidics Hanbin Ma, Yange Wang, Bingbing Zhang, Tong Wang, Yihan Yao, Yong Chang, and 8 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8889289/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Clinical adoption of metagenomic next‑generation sequencing (mNGS) is limited by costly, labor-intensive library preparation requiring substantial DNA input. Here, we present an active‑matrix digital microfluidics library preparation platform (AM‑DMF-Lib-prep) that overcomes these barriers by automating the “sample‑in, library‑out” workflow in an enclosed, digitalized lab-on-a-chip system. The platform accommodates a wide DNA input range (1–25 ng), producing high-quality libraries with exceptional reproducibility (inter-chip R² = 0.9998) and 25.6% lower variability than conventional manual methods. It achieves ~90% reagent reduction while enhancing detection sensitivity for low-abundance pathogens by 43.7–76.7% in microbial community standard spike-in sample. Validation across 94 diverse clinical specimens confirmed high concordance with reference diagnostics. Collectively, AM‑DMF-Lib-prep establishes a precise, cost‑effective, and scalable solution with demonstrated clinical readiness, directly addressing the pre‑analytical bottleneck in routine mNGS. Health sciences/Molecular medicine Physical sciences/Engineering/Biomedical engineering Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 INTRODUCTION Metagenomic next-generation sequencing (mNGS) has emerged as a powerful, unbiased pathogen detection approach and is increasingly adopted in clinical diagnostics, outbreak surveillance and source tracing, microbiome discovery, as well as environmental microbiome surveillance 1 – 4 . However, its widespread adoption is hindered by substantial bottlenecks in library preparation, which contributes substantially to cost, hands-on time, and technical variability, particularly for low-biomass samples 5 . Current mNGS library preparation relies on complex and labor-intensive manual protocols that are prone to operator‑dependent variability, cross‑contamination, and high plastic consumption/waste generation. While robotic liquid‑handling systems can improve throughput and standardization 6 , their high costs, large footprint, and ongoing consumable needs constrain scalable clinical deployment. Thus, a streamlined, cost-effective and clinic-ready solution that ensures standardization and minimizes contamination is urgently needed. As a compelling alternative, Digital Microfluidics (DMF) is an emerging liquid-handling technology that enables the precise manipulation of discrete droplets, ranging from picolitres to microliters, on an array of microelectrodes, typically using the principle of electrowetting on dielectric 7 . The clinical translation potential of DMF is immense, stemming from its capacity for end-to-end automation and miniaturization, which minimizes sample loss and contamination. Active-Matrix Digital Microfluidics (AM-DMF), which leverages ubiquitous thin-film transistor (TFT) backplane technology 8 , represents a crucial paradigm shift in lab-on-a-chip systems by resolving the inherent scalability and pixel-density limitations of conventional passive-matrix devices 9 – 12 . AM-DMF provides a robust, versatile, and programmable foundation for next-generation in vitro diagnostics and Point-of-Care Testing 13 – 17 . Despite the promise of microfluidic and digital microfluidic systems, there is currently no fully integrated and clinically validated platform that enables mNGS library preparation from diverse clinical samples with robust analytical performance and cost-effectiveness 18 – 20 . Building on this, we developed a fully integrated AM-DMF library preparation platform that executes the entire workflow on a single device. AM-DMF-Lib-prep achieves ~ 90% reduction in reagent consumption, 25.6% lower intra‑assay variability, and 43.5–76.7% improvement in the detection of low‑abundance pathogens in microbial community standard spike-in sample, while maintaining excellent concordance with conventional full‑volume method. Validation with 94 diverse clinical specimens further confirmed high agreement with reference mNGS reports. Taken together, this work establishes AM‑DMF library preparation as a precise, cost‑effective, and scalable solution that is ready for clinical application, directly addressing key barriers to the translation and scaling of mNGS‑based infectious disease diagnostics. RESULTS Fully Integrated Workflow of mNGS Library Preparation on an Active-Matrix Digital chip Library preparation is transformed into a sealed, automated “sample-in, library-out” process on an AM-DMF chip. This transformation is achieved through programmable manipulation of discrete droplets via an electrode array coupled with integrated thermal control. The system executes all steps including dispensing, merging, splitting, and transport, which fundamentally eliminates manual pipetting, inter-tube transfers, and the associated risk of human error and cross-contamination. In the initial step, a droplet containing input DNA is merged on AM-DMF with the fragmentation master mix. and rapidly mixed by electrode-driven droplet motion to ensure homogeneous reaction conditions prior to incubation. Using an integrated thermoelectric (TEC) heat exchanger, the platform executes precise thermal control with the following program: 30°C for 5 min, 72°C for 5 min, then hold at 4°C. Subsequently, adapter ligation is performed by merging the fragmented DNA with droplets containing ligation mix and adapters, followed by incubation at 20°C for 15 min. Throughout these steps, the AM‑DMF platform actively maintains non‑reaction phases at 4°C, preserving enzyme activity and nucleic acid integrity between thermal steps. This demonstrates that standard library preparation biochemistry can be faithfully translated to the droplet-based AM-DMF format without compromising thermal control or mixing efficiency. For each purification step, droplets containing solid phase reversible immobilization (SPRI) beads are merged with the reaction droplet for DNA binding. With beads immobilized by an external magnet, the supernatant is first removed and routed to waste region via droplet actuation. The bead-bound DNA is then washed by sequentially introducing and removing droplets of 80% ethanol. For elution, nuclease-free water is introduced and incubated at room temperature for 5 minutes to release DNA from the beads. Finally, with the beads again immobilized by the magnet, the eluate droplet containing the purified DNA is transferred away from the beads to a clean region for subsequent steps. This integrated droplet–bead handling strategy provides precise control over wash and elution conditions, significantly reduces sample loss, and standardizes clean-up across all reactions. Following purification, the eluted DNA is merged with PCR reagent and primers directly on the AM-DMF chip. The platform then executes thermal cycling protocol (95°C for 3 min; 8–10 cycles of 95°C for 20 s, 60°C for 15 s, 72°C for 30 s; final extension at 72°C for 1 min), leveraging its precise temperature control. A final SPRI-based cleanup using the same droplet–magnet workflow yields a ready-to-sequence mNGS library directly from the AM-DMF chip. Collectively, this droplet-based AM-DMF workflow consolidates all reagent mixing, bead washes, and elution into a single, software-controlled process. By eliminating manual handling between steps, AM-DMF-Lib-prep minimizes sample loss and technical variability, providing a robust, automated, and highly reproducible route for mNGS library preparation. (Fig. 1 ) Optimization and Performance Benchmarking of Miniaturized On‑AM‑DMF mNGS Library Preparation The technical feasibility of the miniaturized AM‑DMF workflow was validated by successfully preparing libraries from 25 ng and 5 ng of human gDNA, while employing a ten-fold reduction in reaction volume. All critical steps, including bead handling, were reliably executed at this scale, confirming the platform's operational viability (Supplementary Fig. 1a‑b; Supplementary video 1). Following feasibility confirmation, key reaction parameters were optimized. Using a 5 ng input, we determined that 8 PCR cycles provided an optimal balance, yielding significantly higher library concentrations than 5‑7 cycles while avoiding the over‑amplification artifacts observed with ≥ 10 cycles (Supplementary Fig. 2a‑c). The condition of 8 PCR cycles was therefore selected for subsequent low‑input library preparations. For higher DNA inputs in later comparisons, 6-cycles was adopted to achieve sufficient yield while minimizing PCR‑induced duplicates and bias, in line with standard library preparation principles. Then, a rigorous, head-to-head benchmark against the manual protocol was conducted using a defined mock microbial community standard spike-in sample (ZymoBIOMICS™ Microbial Community Standard D6300 spiked into 10⁵ cells/mL Jurkat cells) across a 1–25 ng DNA input range, to evaluate the performance of the automated, miniaturized on‑AM‑DMF workflow (Fig. 2 a). AM-DMF-Lib-prep demonstrated robust performance across mid-to-high DNA inputs (5, 12, and 25 ng; 6 PCR cycles). It consistently produced libraries with concentrations > 5 ng/µL (Fig. 2 b), near‑identical fragment size distributions (Fig. 2 c), and microbial profiles highly correlated with those from manual 25 ng libraries (Fig. 2 d). This robustness extended to the low-input limits of 1 ng and 3 ng, where libraries maintained sufficient yield (Fig. 2 e), expected fragment sizes (Fig. 2 f), and strongly correlated pathogen profiles with manual 5 ng libraries (Fig. 2 g). To precisely quantify performance advantages, we conducted head-to-head comparisons at key input levels. At 25 ng, on‑AM‑DMF libraries achieved comparable yield and ligation efficiency to manual libraries but exhibited markedly lower PCR duplication rates, indicating superior library complexity (Fig. 2 h). Metagenomic sequencing confirmed the absence of systematic bias, with all pathogens quantitatively detected (Fig. 2 i), and revealed a specific 43.7–76.7% enhancement in detecting low‑abundance fungi ( Cryptococcus neoformans , Saccharomyces cerevisiae ), while recovery of GC‑extreme bacteria remained comparable ( Pseudomonas aeruginosa , Listeria monocytogenes ) (Fig. 2 j). The same trend was observed at 5 ng input, where the on‑AM‑DMF workflow again yielded higher library complexity (Fig. 2 k) and provided 65.3% and 51.9% greater detection of low‑abundance fungi, with unbiased community profiling (Figs. 2 l, m). Collectively, these results demonstrate that the miniaturized on-AM‑DMF lib-prep platform supports robust and unbiased mNGS library preparation across a broad input range (1–25 ng), achieving equivalent or superior performance compared to a conventional manual protocol in terms of library complexity, quantitative accuracy, and sensitivity for low‑abundance targets. Cross‑contamination Assessment of the AM‑DMF Library‑preparation Platform Minimizing inter‑sample cross‑contamination is paramount for the reliability of high-throughput mNGS library preparation. To rigorously evaluate this risk on the AM‑DMF lib-prep platform, a two‑phase assessment was conducted: first, a test for carry-over using defined bacterial standards alongside no‑template controls (NTCs); second, an assessment using clinical samples to determine whether parallel on‑chip processing altered sample‑specific pathogen profiles (Fig. 3 a). In the direct carry-over test, Klebsiella pneumoniae and Staphylococcus aureus (each at a clinically relevant high load of 1,000 copies/mL in a 10⁵ cells/mL Jurkat cells) were processed adjacent to NTCs on the same AM‑DMF chip (Fig. 3 b). Target pathogens were robustly detected exclusively in their designated wells, with no detectable reads in adjacent NTCs, confirming the absence of measurable carry‑over between neighboring reaction sites. Principal component analysis (PCA) of the pathogen profiles showed clear separation into distinct clusters according to biological origin, with NTCs forming an isolated cluster, demonstrating that parallel on‑chip processing does not blur biological signatures (Fig. 3 c). Building on this, a quantitative, profile-based assessment was then performed using clinical samples. The pairwise Jaccard index—a stringent metric of compositional similarity—was significantly lower for samples processed on‑AM‑DMF than for manually prepared ones (Fig. 3 d). This provides direct statistical evidence that the compartmentalized, droplet‑based workflow more effectively maintains sample identity and minimizes cross-sample contamination compared to conventional open-tube processing. Further analysis revealed additional library quality advantages of the AM-DMF-Lib-prep platform. Libraries prepared on AM‑DMF exhibited significantly lower PCR duplication rates and a higher fraction of uniquely mapped reads (Fig. 3 e), indicating superior library complexity and more efficient conversion of input DNA into unique, informative sequencing fragments. Although both methods achieved equally high Q30 scores, all key quality metrics showed substantially tighter distributions for AM‑DMF libraries, underscoring the enhanced process control, reproducibility, and operational consistency of the automated AM-DMF workflow (Fig. 3 e). In summary, the sealed, droplet-based design of the AM‑DMF lib-prep platform not only maintains sample‑specific pathogen signatures without introducing detectable cross‑contamination, but also consistently generates higher‑complexity libraries with reduced technical variability. These results thoroughly validate the platform's reliability for clinical mNGS applications requiring parallel sample processing. Analytical Performance of the AM-DMF Library Preparation Platform Rigorous analytical validation is critical to establish the diagnostic reliability of any library preparation method within an mNGS workflow. Here, we evaluated the fully integrated AM‑DMF lib-prep platform across key metrics for library construction: analytical sensitivity (limit of detection), specificity, interference tolerance, and reproducibility. The limit of detection (LoD) was determined using a panel of five clinically relevant pathogens ( Staphylococcus aureus, Klebsiella pneumoniae, Legionella pneumophila, Cryptococcus neoformans , and Adenovirus ) across a concentration range of 50–1000 copies/mL spiked into 10⁵ cells/mL Jurkat cells. The experimentally determined LoDs were 50, 100, 125, 500, and 1000 copies/mL, respectively. The on-AM-DMF libraries achieved the 100 copies/mL benchmark for all bacterial and fungal targets, and four of the five pathogens were detectable at or below 100 copies/mL. A strong linear correlation was observed between the input microbial titer and sequencing reads (RPM) across the concentration range tested, confirming that the AM-DMF Lib-prep platform preserves the quantitative relationship between input abundance and sequencing readout over clinically relevant concentrations (Fig. 4 a). Specificity was assessed using defined mixtures (1:3, 1:1 and 3:1) of closely related species ( Staphylococcus aureus and Staphylococcus epidermidis ). Libraries generated on-AM-DMF accurately recapitulated the expected abundance ratios in all mixtures, demonstrating on-AM-DMF libraries maintain high specificity without significant competitive bias (Fig. 4 b). Furthermore, the workflow proved robust against common assay interferents. Although the addition of gentamicin (0.1 mg/mL) or mucin (0.5 mg/mL) caused modest fluctuations in absolute RPM values, all target pathogens remained reliably detectable, and the original microbial community composition faithfully reproduced, confirming resilience to inhibitors (Fig. 4 c). A core advantage of the automated AM‑DMF workflow is its superior technical reproducibility. The intra‑method coefficient of variation (CV) for pathogen RPMs across technical replicates was 25.6% lower for on‑AM‑DMF preparations compared to manual processing, indicating substantially reduced technical variability (Fig. 4 d). In addition, excellent inter-chip consistency was demonstrated, as CVs across different AM‑DMF chip batches showed no significant difference. This was further supported by strong correlations of pathogen RPMs both between AM‑DMF and manual libraries and across independent AM‑DMF chips (Fig. 4 e). In summary, the AM‑DMF library preparation platform constitutes a clinic-ready solution, demonstrating excellent analytical sensitivity, linear quantitative response, and robustness against common interferents. Most importantly, it delivers substantially improved precision and reproducibility—both within and across chips—compared to manual methods. These performance characteristics strongly support its suitability for standardized clinical mNGS testing. Validation of the AM-DMF Library Preparation Platform Using Diverse Clinical Specimens To evaluate the performance of the AM‑DMF library preparation platform in a clinical context, a retrospective analysis was conducted using 94 patient specimens with previously established mNGS-based diagnoses. The cohort comprised 84 clinically positive cases and 10 negative cases. The specimens encompassed a range of types, including whole blood (n = 30), bronchoalveolar lavage fluid (BALF; n = 30), sterile body fluids (n = 19), urine (n = 4), pus (n = 4), and tissue (n = 7) (Fig. 5 a-b). Among the positive cases, 44.1% were single-pathogen infections and 55.9% were mixed infections (Fig. 5 c-d). Sequencing analysis of on-AM‑DMF libraries correctly identified 82 of the 84 positive cases and all 10 negative cases, achieving 97.7% overall concordance with the reference clinical reports. This corresponds to a positive percent agreement (PPA) of 97.4% and a negative percent agreement (NPA) of 100% (Fig. 5 e). Quantitative comparison of pathogen readout (RPM) showed strong agreement between AM‑DMF‑prepared and manually prepared libraries: Deming regression demonstrated a highly linear correlation (slope = 1.474, p < 0.0001; Fig. 5 f), and Bland–Altman analysis confirmed no significant systematic bias (Fig. 5 g). For low‑abundance pathogens (RPM ≤ 1), detection rates obtained from AM‑DMF libraries were comparable to those from conventional libraries ( p = 0.6941; Fig. 5 h). The two discordant cases (false negatives) involved samples with extremely low pathogen loads, high host nucleic acid background, and a history of freeze‑thaw cycles, reflecting the inherent stochastic sampling limits of mNGS in extreme low-biomass scenarios. Performance under limited DNA input was further evaluated using a subset of clinical samples. For ten positive and two negative samples, all target pathogens were accurately detected from libraries prepared at both 3 ng and 5 ng input, with no significant degradation in library quality (Fig. 5 i, Supplementary Fig. 3a, b). Deming regression analysis of these libraries yielded slopes close to 1 (1.133 at 3 ng, 1.036 at 5 ng) (Fig. 5 j), and Bland‑Altman plots showed > 95% of data points within the limits of agreement, confirming strong quantitative concordance even with scarce input material (Supplementary Fig. 3c). Collectively, these findings establish that the AM‑DMF library preparation platform delivers high pre‑analytical accuracy, producing libraries whose downstream sequencing results show strong quantitative concordance with those from standard clinical workflows across a wide range of challenging specimen types. Taken together, AM-DMF-Lib-prep’s reliability, robustness, and validated clinical concordance demonstrate its readiness for clinical deployment. DISCUSSION The translation of mNGS from a powerful research tool into a routine, reliable clinical assay has been persistently constrained by the pre-analytical library preparation phase. Conventional methods force a trade-off between performance, cost, labor, and standardization. Here, we present an AM‑DMF platform that offers a transformative approach to these challenges by consolidating the entire “sample-in, library-out” workflow. This work demonstrates how AM‑DMF addresses the tripartite challenge of automation, performance, and cost in a unified manner, potentially changing how clinical mNGS is implemented. A primary advance of this platform is its capacity to enable end-to-end standardization. By performing all steps within a sealed, programmable droplet system, it dramatically reduces the inter-operator and inter-batch variability inherent in manual protocols. The negligible cross-contamination and high inter-chip reproducibility (R² = 0.9998) stem directly from this integrated design, creating a level of operational robustness that is critical for deploying consistent mNGS testing across diverse laboratories—a key step towards establishing universally comparable diagnostic criteria. The analytical performance translates directly into diagnostic utility. The negligible cross-contamination and excellent inter-chip reproducibility (R² = 0.9998) ensure high test specificity and consistent run-to-run performance, mitigating risks of false positives and technical variability. Crucially, the 43.7%–76.7% increase in sensitivity for low-abundance pathogens addresses a core diagnostic gap. This enhancement, validated down to 3 ng input, is particularly relevant for improving diagnostic yield in paucibacillary samples like cerebrospinal fluid, tissue biopsies, or specimens from antibiotic-treated patients. From a practical implementation standpoint, the platform offers compelling operational advantages. The ~ 90% reduction in reagent consumption significantly lowers the direct cost per test. Combined with minimal hands‑on time and reduced rates of assay failure or repeat testing due to contamination, these factors enhance the financial and operational feasibility of integrating mNGS into clinical laboratories. The robust performance across a wide input range, reliably down to 3 ng, alleviates a major constraint of conventional protocols for precious, low‑biomass samples. While library generation from 1 ng inputs was feasible in this study, the analytical robustness and clinical sensitivity at this extreme low limit require further validation. Nevertheless, the ability to process samples consistently at lower input amounts could expand access to mNGS testing in settings where specimen volume is inherently limited, such as pediatrics or neurology. This study has limitations. Clinical validation was retrospective and single-centered; prospective, multi-center trials assessing impact on clinical decision-making and patient outcomes are the essential next step. While benchmarked against a robust manual protocol, comparisons with other automation systems will further define its competitive landscape. The current chip throughput, while suitable for batch testing, requires scaling up for high-demand laboratories. Taken together, the AM‑DMF library preparation platform represents more than a technical advance; it is an enabling infrastructure for clinical metagenomics. By concurrently solving the intertwined problems of manual complexity, performance variability, and high cost, it provides a scalable, precise, and economically viable path forward. This work firmly establishes integrated digital microfluidics as a foundational technology capable of unlocking the long-promised, unbiased diagnostic potential of mNGS for routine patient care. METHODS Library preparation Sequencing libraries were constructed using a commercial kit (Sansure Biotech, for Illumina platforms) according to the manufacturer’s protocol, which encompasses DNA fragmentation, end repair and A-tailing, adapter ligation, bead-based cleanup, library amplification, and final cleanup. For the conventional (off-AM-DMF) method, the protocol was executed manually at the standard reaction volumes defined in the kit instruction. For the on-AM-DMF method, the entire workflow was automated and miniaturized on the digital microfluidic chip. All reagent dispensing, droplet merging, mixing, incubation, and magnetic bead-based purification steps were programmatically controlled. Key reaction conditions (e.g., incubation time and temperature for tagmentation, ligation, and PCR) strictly followed the kit specification. The precise volume of each reagent used for both the on-AM-DMF and conventional protocols is detailed in Supplementary Table 1. DNA sequencing Purified libraries were quantified using a Qubit Fluorometer, and their fragment size distributions were analyzed on an Agilent 4200 TapeStation system with High Sensitivity D5000 ScreenTapes. Qualified libraries were sequenced on an Illumina Nextseq 500 platform. For bioinformatic analysis, reads were processed through a standardized pipeline to remove low-quality sequences and host-derived reads (alignment to the human reference genome hg38). The remaining reads were classified by alignment to pathogen databases. A positive pathogen identification required meeting predefined thresholds for read count, genome coverage, and reads per million (RPM), consistent with established clinical mNGS criteria 21 – 24 . Analytical Performance Assessment The limit of detection (LoD) was determined using a panel of five pathogens: Staphylococcus aureus, Klebsiella pneumoniae, Legionella pneumophila, Cryptococcus neoformans , and Adenovirus . Certified genomic copy numbers of S. aureus, K. pneumoniae, L. pneumophila , and C. neoformans were provided and measured by BeNa Culture Collection (Henan, China). Adenovirus standards were from BDS Biological Technology (Guangzhou, China). Serial dilutions of each pathogen (50–1000 copies/mL) were prepared in a background of 10⁵ cells/mL Jurkat cells and processed on the AM-DMF platform. The LoD was defined as the lowest concentration at which the pathogen was consistently detected. Linearity was assessed by plotting the log-transformed input concentration against the log-transformed RPM values from the same dilution series, and a linear regression was performed to calculate the coefficient of determination (R²). Gentamicin (0.1 mg/mL) and mucin (0.5 mg/mL) were added into the microbial community standard spike-in sample (ZymoBIOMICS™ Microbial Community Standard spiked into 10⁵ cells/mL Jurkat cells) to evaluate interference from substances commonly encountered in clinical samples. The impact of these interferents on pathogen detection efficiency was quantified through RPM value analysis. To evaluate specificity in co-infections, mixtures of Staphylococcus aureus and Staphylococcus epidermidis were prepared at defined genomic copy ratios (1:3, 1:1, 3:1). The observed RPM ratios in the resulting libraries were compared to the expected input ratios. To comprehensively evaluate the precision of the AM-DMF platform, we assessed intra-chip, inter-chip, and inter-method reproducibility using the microbial community standard spike-in sample (ZymoBIOMICS™ Microbial Community Standard spiked into 10⁵ cells/mL Jurkat cells). For intra-chip precision, the same sample was processed in five replicate reaction units on the same AM-DMF chip. For inter-chip precision, the same sample was processed on two independent AM-DMF chip batches (designated Chip Batch 1 and 2), with five replicates per batch. For inter-method precision, the same sample material was also processed in parallel using the conventional manual library preparation protocol (n = 5), serving as the reference. Precision was quantified by calculating the coefficient of variation (CV%) of the reads per million (RPM) values for each constituent pathogen across the replicate measurements within each group. The quantitative concordance between the AM-DMF platform and the manual method, as well as between the two chip batches, was further evaluated using Pearson correlation analysis. Fabrication of the AM-DMF chip The digital microfluidic (DMF) chip is a sealed, two-plate device fabricated through a combination of flat-panel display manufacturing and micro-assembly processes. The bottom plate was manufactured on a glass substrate using a standard flat-panel display process. It features an active-matrix array independently addressable thin-film transistor (TFT) pixel electrodes. A 300 nm-thick silicon nitride (SiNx) layer was deposited uniformly over the electrode array as the dielectric layer. The top plate consists of a flat, approximately 2-inch indium tin oxide (ITO)-coated glass sheet. Laser-drilled micro-holes (1 or 2 mm in diameter) serve as fluidic access ports for sample and reagent introduction/extraction. Prior to assembly, both the top ITO plate and the bottom TFT plate were ultrasonically cleaned and coated with a hydrophobic layer by spin-coating and baking Teflon-AF (1% w/w in Fluorinert FC-40). The two plates were aligned and bonded with a precisely controlled gap of 50 µm, maintained using UV-curable adhesive or plastic spacers. This cavity facilitates droplet formation and transport. The final chip packaging, including bonding flexible printed circuit (FPC) connectors to the chip via a standard hot-press process, was completed in a Class 10,000 cleanroom facility (ACXEL, Foshan, China). Study Cohort and Clinical Samples Residual fluid and tissue samples were collected from patients who underwent clinically indicated metagenomic next-generation sequencing (mNGS) testing at the Department of Laboratory Medicine, Sichuan Provincial People’s Hospital, School of Medicine, University of Electronic Science and Technology of China. Sample types included whole blood, bronchoalveolar lavage fluid (BALF), cerebrospinal fluid (CSF), ascites, synovial fluid, pus, urine, abscess drainage, biopsy tissue, and bone marrow. Inclusion criteria were: (1) availability of sufficient sample volume (≥ 600 µL) after routine testing, and (2) complete accompanying clinical information. Exclusion criteria were: (1) insufficient sample volume, or (2) prior failure of the clinical mNGS assay. All samples were collected between May and September 2025 as part of standard clinical care, and stored at − 80°C until nucleic acid extraction. Sample Processing and Nucleic Acids Extraction All samples were initially inactivated at 56°C for 30 minutes in a dry heating block, followed by type-specific pretreatments: blood samples were centrifuged at 1,600 × g for 10 min at 4°C to separate plasma; clear cerebrospinal fluid (CSF) was concentrated by centrifugation at 13,000 × g for 10 min, after which the supernatant was removed and the pellet was resuspended in approximately 100 µL of the residual fluid; viscous specimens (e.g., purulent CSF, bronchoalveolar lavage fluid, or pus) were mixed 1:1 (v/v) with 0.1 M dithiothreitol (DTT) and incubated at room temperature for 30 min to reduce viscosity; and tissue specimens were minced into fine fragments using sterile ophthalmic scissors. Following pretreatment, total nucleic acids were extracted from 350 µL of each processed lysate using a commercial Pathogen DNA/RNA Extraction Kit (Sansure Biotech, Changsha, China) according to the manufacturer’s instructions. To enhance the detection sensitivity for low-biomass samples, extracted nucleic acids were concentrated using a 1.8x bead ratio, typically yielding a 3–5 fold increase in concentration. A no-template control (NTC, nuclease-free water) was processed in parallel with each extraction batch to monitor potential contamination. Declarations Ethics approval statement: This study was approved by the Institutional Review Board of Sichuan Provincial People's Hospital (Approval 2025 No.557]). All procedures were performed in accordance with the relevant guidelines and regulations. Consent statement: Informed consent was waived by the approving ethics committee due to the retrospective nature of the study using residual clinical specimens, which were collected during routine clinical care and anonymized prior to analysis COMPETING INTERESTS One patent has been filed based on this work. AUTHOR CONTRIBUTIONS HM, LJ and ZY conceived and designed the study. YW and YC did statistical analyses and wrote the manuscript. YW, BZ, TW, YY, YC, JL, YiW, JJ, SH, JZ and LY contributed to data collection and the overall structure of the study. HM, LJ and ZY reviewed the manuscript and supervised the study. All authors read and approved the final manuscript and had final responsibility for the decision to submit for publication. ACKNOWLEDGMENTS This work was supported by National Key R&D Program of China (2023YFF0721500), Sichuan Science and Technology Program (2025ZDZX0103) and the Suzhou Basic Research Pilot Project (SSD2023013). References Gu W, Miller S, Chiu CY (2019) Clinical Metagenomic Next-Generation Sequencing for Pathogen Detection. Annu Rev Pathol 14:319–338. 10.1146/annurev-pathmechdis-012418-012751 Huang L et al (2021) Dynamic blood single-cell immune responses in patients with COVID-19. Signal Transduct Target Ther 6:110. 10.1038/s41392-021-00526-2 Gong B et al (2022) A genetic variant in IL-6 lowering its expression is protective for critical patients with COVID-19. Signal Transduct Target Ther 7:112. 10.1038/s41392-022-00923-1 Xie C et al (2020) Comparison of different samples for 2019 novel coronavirus detection by nucleic acid amplification tests. Int J Infect Dis 93:264–267. 10.1016/j.ijid.2020.02.050 Simner PJ, Miller S, Carroll KC (2018) Understanding the promises and hurdles of metagenomic next-generation sequencing as a diagnostic tool for infectious diseases. Clin Infect Dis 66:778–788 Tegally H, San JE, Giandhari J, de Oliveira T (2020) Unlocking the efficiency of genomics laboratories with robotic liquid-handling. BMC Genomics 21:729. 10.1186/s12864-020-07137-1 Rackus DG, Shamsi MH, Wheeler AR (2015) Electrochemistry, biosensors and microfluidics: a convergence of fields. Chem Soc Rev 44:5320–5340 Wang D et al (2022) Thin-film transistor arrays for biological sensing systems. Flex Print Electron 7:023004 Wang D, Jiang S, Ma H, Yu J, Nathan A (2025) Active-matrix digital microfluidics for high-throughput, precise droplet handling. Nat Reviews Electr Eng, 1–15 Hu S et al (2023) Large-area electronics-enabled high-resolution digital microfluidics for parallel single-cell manipulation. Anal Chem 95:6905–6914 Guo Z et al (2025) Deep Learning-Assisted Label‐Free Parallel Cell Sorting with Digital Microfluidics. Adv Sci 12:2408353 Jia Z et al (2025) Intelligent single-cell manipulation: LLMs-and object detection-enhanced active-matrix digital microfluidics. Microsystems Nanoengineering 11:133 Ji J et al (2025) AM-DMF‐ddRPA: An All‐in‐One Digital Microfluidic Platform for Rapid and Automatic Digital Nucleic Acid Analysis. Angew Chem Int Ed, e202501913 Yang Z et al (2024) AM-DMF-SCP: integrated single-cell proteomics analysis on an active matrix digital microfluidic chip. JACS Au 4:1811–1823 Li H et al (2024) Droplet-based microfluidics with mass spectrometry for microproteomics. Engineering 43:37–53 Liu J et al (2024) Rapid and multi-target genotyping of Helicobacter pylori with digital microfluidics. Biosens Bioelectron 256:116282 Hu S et al (2022) All-in-one digital microfluidics system for molecular diagnosis with loop-mediated isothermal amplification. Biosensors 12:324 Hess JF et al (2020) Library preparation for next generation sequencing: A review of automation strategies. Biotechnol Adv 41:107537. 10.1016/j.biotechadv.2020.107537 Kim H et al (2013) A microfluidic DNA library preparation platform for next-generation sequencing. PLoS ONE 8:e68988 Kim S et al (2017) High-throughput automated microfluidic sample preparation for accurate microbial genomics. Nat Commun 8:13919 Schlaberg R, Chiu CY, Miller S, Procop GW, Weinstock G (2017) Validation of Metagenomic Next-Generation Sequencing Tests for Universal Pathogen Detection. Arch Pathol Lab Med 141:776–786. 10.5858/arpa.2016-0539-RA Bittinger K et al (2014) Improved characterization of medically relevant fungi in the human respiratory tract using next-generation sequencing. Genome Biol 15:487. 10.1186/s13059-014-0487-y Doughty EL, Sergeant MJ, Adetifa I, Antonio M, Pallen MJ (2014) Culture-independent detection and characterisation of Mycobacterium tuberculosis and M. africanum in sputum samples using shotgun metagenomics on a benchtop sequencer. PeerJ 2:e585. 10.7717/peerj.585 Miao Q et al (2018) Microbiological diagnostic performance of metagenomic next-generation sequencing when applied to clinical practice. Clin Infect Dis 67:S231–S240 Additional Declarations There is NO Competing Interest. One patent has been filed based on this work. Supplementary Files supplementaryvideo1.mp4 Supplementary Video 1 rs.pdf Reporting Summary Supplementaryfigures.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8889289","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":594604696,"identity":"331d75cc-7316-4531-b960-e8a330628e0e","order_by":0,"name":"Hanbin Ma","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA5ElEQVRIiWNgGAWjYBACAwYGNoYEEIu9ASaWQKwWngMw1cRoAQOJBCK1mLOfPfbg4Q6GaP6Zbwwf/vxhx8DPnmPA8HMHbi2WPXnpBolnGHJn3M4xNuZJSGaQ7HljwNh7Bo/DDuSYSSS2MeQ23M4xk2ZIOMBgcCPHgJmxDY+W828gWubfPGP+8wdQiz1BLTegtmy4wWPGwAOyRYKgljfmBoltErkbz6QVS/OkJfNInHlWcLAXr8NyzB7+bLPJnXf88MaPP2zs5Pjbkzc++IlHCxRIwFk8IOIAQQ2jYBSMglEwCvACADCgT8r8opxYAAAAAElFTkSuQmCC","orcid":"","institution":"Guangdong ACXEL Micro \u0026 Nano Tech Co. Ltd. \u0026University of Electronic Science and Technology of China","correspondingAuthor":true,"prefix":"","firstName":"Hanbin","middleName":"","lastName":"Ma","suffix":""},{"id":594604697,"identity":"ee187c65-d664-4e14-82cd-89837e23c136","order_by":1,"name":"Yange Wang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Yange","middleName":"","lastName":"Wang","suffix":""},{"id":594604698,"identity":"674c1f55-433b-47fe-b8aa-f2a791528307","order_by":2,"name":"Bingbing Zhang","email":"","orcid":"","institution":"CAS Key Laboratory of Bio-medical Diagnostics, Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Bingbing","middleName":"","lastName":"Zhang","suffix":""},{"id":594604699,"identity":"74c8c8d6-c2c4-4561-9366-d4e9123a3a3c","order_by":3,"name":"Tong Wang","email":"","orcid":"","institution":"Guangdong ACXEL Micro\u0026Nano Tech Co. Ltd.","correspondingAuthor":false,"prefix":"","firstName":"Tong","middleName":"","lastName":"Wang","suffix":""},{"id":594604700,"identity":"96d150de-1625-4046-be52-cd2020ba8c95","order_by":4,"name":"Yihan Yao","email":"","orcid":"","institution":"University of Electronic Science and Technology of China","correspondingAuthor":false,"prefix":"","firstName":"Yihan","middleName":"","lastName":"Yao","suffix":""},{"id":594604701,"identity":"55cf606d-789e-49e7-b482-7785181c3452","order_by":5,"name":"Yong Chang","email":"","orcid":"","institution":"Sichuan Provincial People's Hospital, School of Medicine, University of Electronic Science and Technology of China","correspondingAuthor":false,"prefix":"","firstName":"Yong","middleName":"","lastName":"Chang","suffix":""},{"id":594604702,"identity":"da3047e8-3872-4589-a58b-9103924ef677","order_by":6,"name":"Juan Long","email":"","orcid":"","institution":"Sichuan Provincial People's Hospital, School of Medicine, University of Electronic Science and Technology of China","correspondingAuthor":false,"prefix":"","firstName":"Juan","middleName":"","lastName":"Long","suffix":""},{"id":594604703,"identity":"d5fee206-2aad-44fb-b1c0-de16713cb46c","order_by":7,"name":"Yili Wang","email":"","orcid":"","institution":"Sichuan Provincial People's Hospital, School of Medicine, University of Electronic Science and Technology of China","correspondingAuthor":false,"prefix":"","firstName":"Yili","middleName":"","lastName":"Wang","suffix":""},{"id":594604704,"identity":"77a4a420-7067-4425-92bf-b1aafd81879c","order_by":8,"name":"Jiajian Ji","email":"","orcid":"https://orcid.org/0000-0002-4058-2733","institution":"Chinese Academy of Science","correspondingAuthor":false,"prefix":"","firstName":"Jiajian","middleName":"","lastName":"Ji","suffix":""},{"id":594604705,"identity":"e17a5d30-2bb4-4b86-825d-a38a8953d0a0","order_by":9,"name":"Siyi Hu","email":"","orcid":"https://orcid.org/0000-0002-0686-5182","institution":"Suzhou Institute of Biomedical Engineering and Technology, Chinese Academy of Sciences","correspondingAuthor":false,"prefix":"","firstName":"Siyi","middleName":"","lastName":"Hu","suffix":""},{"id":594604706,"identity":"5eb4df65-0957-4bd9-b984-8c5f69689a73","order_by":10,"name":"Jie Zhang","email":"","orcid":"","institution":"Sichuan Academy of Medical Sciences and Sichuan Provincial People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"Jie","middleName":"","lastName":"Zhang","suffix":""},{"id":594604707,"identity":"9bfbc5a4-8c30-4880-b6cb-f0517fb4fe54","order_by":11,"name":"Li Yang","email":"","orcid":"","institution":"Sansure Biotech Incorporation","correspondingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Yang","suffix":""},{"id":594604708,"identity":"9f411150-0935-4efb-91f7-145d91ccc8fc","order_by":12,"name":"Zhenglin Yang","email":"","orcid":"https://orcid.org/0000-0002-8656-7862","institution":"Sichuan Provincial People's Hospital, University of Electronic Science and Technology of China","correspondingAuthor":false,"prefix":"","firstName":"Zhenglin","middleName":"","lastName":"Yang","suffix":""},{"id":594604709,"identity":"ab7a09b3-983f-42f0-bbd2-f62586858ded","order_by":13,"name":"Li Jiang","email":"","orcid":"","institution":"Sichuan Provincial People's Hospital, School of Medicine, University of Electronic Science and Technology of China","correspondingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Jiang","suffix":""}],"badges":[],"createdAt":"2026-02-16 03:20:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8889289/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8889289/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":103840542,"identity":"eacc6486-77e4-4785-87af-69cbbb0aadff","added_by":"auto","created_at":"2026-03-03 14:40:46","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":126944,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIntegrated workflow of mNGS library preparation on the AM‑DMF platform. a-b,\u003c/strong\u003eschematic of the fully automated, “sample‑in, library‑out” workflow on the AM‑DMF chip. The process integrates DNA tagmentation, adapter ligation, bead‑based purification, and PCR amplification into a single, sealed AM-DMF chip, with less than 15 minutes of hands‑on time;\u003cstrong\u003e c,\u003c/strong\u003e diagram of the AM‑DMF chip layout, showing the programmed locations and droplet transport paths (arrows) for each reaction step.\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8889289/v1/ac5cc7fcd97d235cde0466df.jpg"},{"id":103840479,"identity":"fc8cacef-4c34-4e4a-9c37-b0267285a948","added_by":"auto","created_at":"2026-03-03 14:40:37","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":457936,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePerformance validation of AM-DMF lib-prep across a broad range of DNA input. a, \u003c/strong\u003eschematic comparing the workflows of conventional manual and miniaturized, automated AM‑DMF mNGS library preparation.\u003cstrong\u003e b-d, \u003c/strong\u003eperformance across 5-25 ng inputs (6 PCR cycles): \u003cstrong\u003eb,\u003c/strong\u003elibrary concentration; \u003cstrong\u003ec\u003c/strong\u003e fragment size distributions; \u003cstrong\u003ed, \u003c/strong\u003ecorrelation of microbial profiles (log₁₀(RPM+1); Pearson's r indicated).\u003cstrong\u003e e-g,\u003c/strong\u003eperformance at low-input limits (1-5 ng; 8 PCR cycles): \u003cstrong\u003ee,\u003c/strong\u003e library yield; \u003cstrong\u003ef, \u003c/strong\u003efragment size distributions; \u003cstrong\u003eg,\u003c/strong\u003e correlation of microbial profiles. All libraries were prepared from the ZymoBIOMICS D6300 standard spiked into 10⁵ cells/mL Jurkat cells. Data are mean ± s.d. (n=3). \u003cstrong\u003eh-j\u003c/strong\u003e, performance at 25 ng input: \u003cstrong\u003eh,\u003c/strong\u003e library concentration, adapter-ligation rate, and PCR duplication rate for manual versus AM‑DMF libraries; \u003cstrong\u003ei, \u003c/strong\u003emetagenomic sequencing profiles (log₁₀(RPM+1); bubble size indicates RPM of all organisms;\u003cstrong\u003ej, \u003c/strong\u003ereadout of selected low-abundance fungi (\u003cem\u003eCryptococcus neoformans\u003c/em\u003e, 40 copies/mL; \u003cem\u003eSaccharomyces cerevisiae\u003c/em\u003e, 25 copies/mL) and GC-extreme bacteria (\u003cem\u003ePseudomonas aeruginosa\u003c/em\u003e, \u003cem\u003eListeria monocytogenes\u003c/em\u003e). \u003cstrong\u003ek-m,\u003c/strong\u003eperformance at 5 ng input: Panels as in h-j.\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8889289/v1/9d69463c5e16a48bf24a24a6.jpg"},{"id":103840573,"identity":"87f6631d-e35d-4368-883e-c1f3fd3ae39d","added_by":"auto","created_at":"2026-03-03 14:40:54","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":189939,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAssessment of cross‑contamination and sample specificity on the AM‑DMF platform.\u003c/strong\u003e \u003cstrong\u003ea,\u003c/strong\u003e experimental design for assessing inter-sample cross-contamination on the AM‑DMF Lib‑prep platform. \u003cstrong\u003eb\u003c/strong\u003e, on-AM-DMF layout of positive controls and no-template controls (NTCs), with a summary table of mNGS detection results. \u003cstrong\u003ec, \u003c/strong\u003eprincipal component analysis (PCA) of pathogen relative‑abundance profiles from four samples processed on AM‑DMF. Points are colored by sample identity. \u003cstrong\u003ed,\u003c/strong\u003e heatmaps of pairwise Jaccard indices for pathogen detection profiles among four clinical samples processed on a single microfluidic chip versus manually. \u003cstrong\u003ee,\u003c/strong\u003e paired comparison of Jaccard indices (upper left), library duplication rate (upper right), uniquely mapped read fraction (lower left), and Q30 score distribution (lower right) between AM‑DMF and manual libraries from the same samples (n=20).\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8889289/v1/0d98bf7233266fe9299b7333.jpg"},{"id":103840531,"identity":"7f4b40a3-f284-4e88-89ff-204efddb2627","added_by":"auto","created_at":"2026-03-03 14:40:43","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":225825,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAnalytical Performance of the AM‑DMF‑Lib‑prep method. a\u003c/strong\u003elimit of detection (LoD) determination for a panel of clinically relevant pathogens across 50–1000 copies/mL;\u003cstrong\u003e b\u003c/strong\u003e recovery of expected ratios in mixtures of \u003cem\u003eS. aureus\u003c/em\u003e and\u003cem\u003e S. epidermidis\u003c/em\u003e at defined genomic copy ratios; \u003cstrong\u003ec \u003c/strong\u003epathogen detection in the presence of common interferents (gentamicin, 0.1 mg/mL; mucin, 0.5 mg/mL) in a defined microbial community spike-in sample;\u003cstrong\u003e d \u003c/strong\u003ecomparison of the intra-method coefficient of variation (CV%) for pathogen RPMs between on‑AM‑DMF and manual library preparations across technical replicates; \u003cstrong\u003ee, \u003c/strong\u003econcordance of pathogen RPMs between methods (left) and across independent AM‑DMF chip batches (right).\u003c/p\u003e","description":"","filename":"Picture4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8889289/v1/bdd0499f6d083b0677ae82e2.jpg"},{"id":103840502,"identity":"f9eb3fec-21cc-47e6-98c9-b4f7813c7d37","added_by":"auto","created_at":"2026-03-03 14:40:41","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":330744,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDiagnostic accuracy and quantitative concordance across diverse clinical specimen types. a,\u003c/strong\u003e 94 clinical specimens used in this work;\u003cstrong\u003eb,\u003c/strong\u003e the distribution of infection categories;\u003cstrong\u003e c,d,\u003c/strong\u003e fraction and propotion of patient samples presenting single‑pathogen vs. mixed‑pathogen infections;\u003cstrong\u003e e,\u003c/strong\u003e contingency table comparing detection outcomes between the two library preparation methods;\u003cstrong\u003e f,\u003c/strong\u003e deming regression analysis of log₁₀‑transformed pathogen RPM;\u003cstrong\u003e g,\u003c/strong\u003e Bland–Altman plot evaluating agreement of log₁₀‑transformed pathogen RPM; \u003cstrong\u003eh,\u003c/strong\u003e paired comparison of log₁₀‑transformed pathogen RPM; \u003cstrong\u003ei,\u003c/strong\u003e heatmap for log₁₀‑transformed pathogen RPM values from ten clinical samples; \u003cstrong\u003ej\u003c/strong\u003e, deming regression analysis of pathogen RPMs\u003cstrong\u003e.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8889289/v1/5d68b8e1e467a87961a0d27e.jpg"},{"id":104782437,"identity":"3f3b79b8-5453-48bb-bf6a-f9199ebb2151","added_by":"auto","created_at":"2026-03-17 07:57:18","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2192364,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8889289/v1/88f83672-26a5-474c-8d51-34aac587e9ce.pdf"},{"id":103840538,"identity":"5f0fa3b7-73c5-4da3-8023-4ca1502abb6e","added_by":"auto","created_at":"2026-03-03 14:40:44","extension":"mp4","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":7062292,"visible":true,"origin":"","legend":"Supplementary Video 1","description":"","filename":"supplementaryvideo1.mp4","url":"https://assets-eu.researchsquare.com/files/rs-8889289/v1/65105c3417625a342011819a.mp4"},{"id":103840537,"identity":"231cf60c-640b-4e00-a6ee-364f9ce8ec34","added_by":"auto","created_at":"2026-03-03 14:40:44","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":83945,"visible":true,"origin":"","legend":"Reporting Summary","description":"","filename":"rs.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8889289/v1/892c12505d74ca368eb8d62a.pdf"},{"id":103840499,"identity":"6c401abc-75cd-4c20-bb35-2bf00286db20","added_by":"auto","created_at":"2026-03-03 14:40:41","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":3490648,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryfigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-8889289/v1/c364bb772254a5809a361811.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.\nOne patent has been filed based on this work.","formattedTitle":"AM‑DMF-Lib-prep: A clinical ready lib-prep-on-a-chip platform with active-matrix digital microfluidics","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eMetagenomic next-generation sequencing (mNGS) has emerged as a powerful, unbiased pathogen detection approach and is increasingly adopted in clinical diagnostics, outbreak surveillance and source tracing, microbiome discovery, as well as environmental microbiome surveillance\u003csup\u003e\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. However, its widespread adoption is hindered by substantial bottlenecks in library preparation, which contributes substantially to cost, hands-on time, and technical variability, particularly for low-biomass samples\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eCurrent mNGS library preparation relies on complex and labor-intensive manual protocols that are prone to operator‑dependent variability, cross‑contamination, and high plastic consumption/waste generation. While robotic liquid‑handling systems can improve throughput and standardization\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e, their high costs, large footprint, and ongoing consumable needs constrain scalable clinical deployment. Thus, a streamlined, cost-effective and clinic-ready solution that ensures standardization and minimizes contamination is urgently needed.\u003c/p\u003e \u003cp\u003eAs a compelling alternative, Digital Microfluidics (DMF) is an emerging liquid-handling technology that enables the precise manipulation of discrete droplets, ranging from picolitres to microliters, on an array of microelectrodes, typically using the principle of electrowetting on dielectric\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. The clinical translation potential of DMF is immense, stemming from its capacity for end-to-end automation and miniaturization, which minimizes sample loss and contamination. Active-Matrix Digital Microfluidics (AM-DMF), which leverages ubiquitous thin-film transistor (TFT) backplane technology\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e, represents a crucial paradigm shift in lab-on-a-chip systems by resolving the inherent scalability and pixel-density limitations of conventional passive-matrix devices\u003csup\u003e\u003cspan additionalcitationids=\"CR10 CR11\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. AM-DMF provides a robust, versatile, and programmable foundation for next-generation in vitro diagnostics and Point-of-Care Testing\u003csup\u003e\u003cspan additionalcitationids=\"CR14 CR15 CR16\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. Despite the promise of microfluidic and digital microfluidic systems, there is currently no fully integrated and clinically validated platform that enables mNGS library preparation from diverse clinical samples with robust analytical performance and cost-effectiveness \u003csup\u003e\u003cspan additionalcitationids=\"CR19\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eBuilding on this, we developed a fully integrated AM-DMF library preparation platform that executes the entire workflow on a single device. AM-DMF-Lib-prep achieves\u0026thinsp;~\u0026thinsp;90% reduction in reagent consumption, 25.6% lower intra‑assay variability, and 43.5\u0026ndash;76.7% improvement in the detection of low‑abundance pathogens in microbial community standard spike-in sample, while maintaining excellent concordance with conventional full‑volume method. Validation with 94 diverse clinical specimens further confirmed high agreement with reference mNGS reports. Taken together, this work establishes AM‑DMF library preparation as a precise, cost‑effective, and scalable solution that is ready for clinical application, directly addressing key barriers to the translation and scaling of mNGS‑based infectious disease diagnostics.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e"},{"header":"RESULTS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eFully Integrated Workflow of mNGS Library Preparation on an Active-Matrix Digital chip\u003c/h2\u003e \u003cp\u003eLibrary preparation is transformed into a sealed, automated \u0026ldquo;sample-in, library-out\u0026rdquo; process on an AM-DMF chip. This transformation is achieved through programmable manipulation of discrete droplets via an electrode array coupled with integrated thermal control. The system executes all steps including dispensing, merging, splitting, and transport, which fundamentally eliminates manual pipetting, inter-tube transfers, and the associated risk of human error and cross-contamination.\u003c/p\u003e \u003cp\u003eIn the initial step, a droplet containing input DNA is merged on AM-DMF with the fragmentation master mix. and rapidly mixed by electrode-driven droplet motion to ensure homogeneous reaction conditions prior to incubation. Using an integrated thermoelectric (TEC) heat exchanger, the platform executes precise thermal control with the following program: 30\u0026deg;C for 5 min, 72\u0026deg;C for 5 min, then hold at 4\u0026deg;C. Subsequently, adapter ligation is performed by merging the fragmented DNA with droplets containing ligation mix and adapters, followed by incubation at 20\u0026deg;C for 15 min. Throughout these steps, the AM‑DMF platform actively maintains non‑reaction phases at 4\u0026deg;C, preserving enzyme activity and nucleic acid integrity between thermal steps. This demonstrates that standard library preparation biochemistry can be faithfully translated to the droplet-based AM-DMF format without compromising thermal control or mixing efficiency.\u003c/p\u003e \u003cp\u003eFor each purification step, droplets containing solid phase reversible immobilization (SPRI) beads are merged with the reaction droplet for DNA binding. With beads immobilized by an external magnet, the supernatant is first removed and routed to waste region via droplet actuation. The bead-bound DNA is then washed by sequentially introducing and removing droplets of 80% ethanol. For elution, nuclease-free water is introduced and incubated at room temperature for 5 minutes to release DNA from the beads. Finally, with the beads again immobilized by the magnet, the eluate droplet containing the purified DNA is transferred away from the beads to a clean region for subsequent steps. This integrated droplet\u0026ndash;bead handling strategy provides precise control over wash and elution conditions, significantly reduces sample loss, and standardizes clean-up across all reactions.\u003c/p\u003e \u003cp\u003eFollowing purification, the eluted DNA is merged with PCR reagent and primers directly on the AM-DMF chip. The platform then executes thermal cycling protocol (95\u0026deg;C for 3 min; 8\u0026ndash;10 cycles of 95\u0026deg;C for 20 s, 60\u0026deg;C for 15 s, 72\u0026deg;C for 30 s; final extension at 72\u0026deg;C for 1 min), leveraging its precise temperature control. A final SPRI-based cleanup using the same droplet\u0026ndash;magnet workflow yields a ready-to-sequence mNGS library directly from the AM-DMF chip.\u003c/p\u003e \u003cp\u003eCollectively, this droplet-based AM-DMF workflow consolidates all reagent mixing, bead washes, and elution into a single, software-controlled process. By eliminating manual handling between steps, AM-DMF-Lib-prep minimizes sample loss and technical variability, providing a robust, automated, and highly reproducible route for mNGS library preparation. (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e)\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eOptimization and Performance Benchmarking of Miniaturized On‑AM‑DMF mNGS Library Preparation\u003c/h3\u003e\n\u003cp\u003eThe technical feasibility of the miniaturized AM‑DMF workflow was validated by successfully preparing libraries from 25 ng and 5 ng of human gDNA, while employing a ten-fold reduction in reaction volume. All critical steps, including bead handling, were reliably executed at this scale, confirming the platform's operational viability (Supplementary Fig.\u0026nbsp;1a‑b; Supplementary video 1). Following feasibility confirmation, key reaction parameters were optimized. Using a 5 ng input, we determined that 8 PCR cycles provided an optimal balance, yielding significantly higher library concentrations than 5‑7 cycles while avoiding the over‑amplification artifacts observed with \u0026ge;\u0026thinsp;10 cycles (Supplementary Fig.\u0026nbsp;2a‑c). The condition of 8 PCR cycles was therefore selected for subsequent low‑input library preparations. For higher DNA inputs in later comparisons, 6-cycles was adopted to achieve sufficient yield while minimizing PCR‑induced duplicates and bias, in line with standard library preparation principles. Then, a rigorous, head-to-head benchmark against the manual protocol was conducted using a defined mock microbial community standard spike-in sample (ZymoBIOMICS\u0026trade; Microbial Community Standard D6300 spiked into 10⁵ cells/mL Jurkat cells) across a 1\u0026ndash;25 ng DNA input range, to evaluate the performance of the automated, miniaturized on‑AM‑DMF workflow (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003eAM-DMF-Lib-prep demonstrated robust performance across mid-to-high DNA inputs (5, 12, and 25 ng; 6 PCR cycles). It consistently produced libraries with concentrations\u0026thinsp;\u0026gt;\u0026thinsp;5 ng/\u0026micro;L (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb), near‑identical fragment size distributions (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec), and microbial profiles highly correlated with those from manual 25 ng libraries (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed). This robustness extended to the low-input limits of 1 ng and 3 ng, where libraries maintained sufficient yield (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ee), expected fragment sizes (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ef), and strongly correlated pathogen profiles with manual 5 ng libraries (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eg).\u003c/p\u003e \u003cp\u003eTo precisely quantify performance advantages, we conducted head-to-head comparisons at key input levels. At 25 ng, on‑AM‑DMF libraries achieved comparable yield and ligation efficiency to manual libraries but exhibited markedly lower PCR duplication rates, indicating superior library complexity (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eh). Metagenomic sequencing confirmed the absence of systematic bias, with all pathogens quantitatively detected (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ei), and revealed a specific 43.7\u0026ndash;76.7% enhancement in detecting low‑abundance fungi (\u003cem\u003eCryptococcus neoformans\u003c/em\u003e, \u003cem\u003eSaccharomyces cerevisiae\u003c/em\u003e), while recovery of GC‑extreme bacteria remained comparable (\u003cem\u003ePseudomonas aeruginosa\u003c/em\u003e, \u003cem\u003eListeria monocytogenes\u003c/em\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ej). The same trend was observed at 5 ng input, where the on‑AM‑DMF workflow again yielded higher library complexity (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ek) and provided 65.3% and 51.9% greater detection of low‑abundance fungi, with unbiased community profiling (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003el, m).\u003c/p\u003e \u003cp\u003eCollectively, these results demonstrate that the miniaturized on-AM‑DMF lib-prep platform supports robust and unbiased mNGS library preparation across a broad input range (1\u0026ndash;25 ng), achieving equivalent or superior performance compared to a conventional manual protocol in terms of library complexity, quantitative accuracy, and sensitivity for low‑abundance targets.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eCross‑contamination Assessment of the AM‑DMF Library‑preparation Platform\u003c/h3\u003e\n\u003cp\u003eMinimizing inter‑sample cross‑contamination is paramount for the reliability of high-throughput mNGS library preparation. To rigorously evaluate this risk on the AM‑DMF lib-prep platform, a two‑phase assessment was conducted: first, a test for carry-over using defined bacterial standards alongside no‑template controls (NTCs); second, an assessment using clinical samples to determine whether parallel on‑chip processing altered sample‑specific pathogen profiles (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003eIn the direct carry-over test, \u003cem\u003eKlebsiella pneumoniae\u003c/em\u003e and \u003cem\u003eStaphylococcus aureus\u003c/em\u003e (each at a clinically relevant high load of 1,000 copies/mL in a 10⁵ cells/mL Jurkat cells) were processed adjacent to NTCs on the same AM‑DMF chip (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb). Target pathogens were robustly detected exclusively in their designated wells, with no detectable reads in adjacent NTCs, confirming the absence of measurable carry‑over between neighboring reaction sites. Principal component analysis (PCA) of the pathogen profiles showed clear separation into distinct clusters according to biological origin, with NTCs forming an isolated cluster, demonstrating that parallel on‑chip processing does not blur biological signatures (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec).\u003c/p\u003e \u003cp\u003eBuilding on this, a quantitative, profile-based assessment was then performed using clinical samples. The pairwise Jaccard index\u0026mdash;a stringent metric of compositional similarity\u0026mdash;was significantly lower for samples processed on‑AM‑DMF than for manually prepared ones (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed). This provides direct statistical evidence that the compartmentalized, droplet‑based workflow more effectively maintains sample identity and minimizes cross-sample contamination compared to conventional open-tube processing.\u003c/p\u003e \u003cp\u003eFurther analysis revealed additional library quality advantages of the AM-DMF-Lib-prep platform. Libraries prepared on AM‑DMF exhibited significantly lower PCR duplication rates and a higher fraction of uniquely mapped reads (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee), indicating superior library complexity and more efficient conversion of input DNA into unique, informative sequencing fragments. Although both methods achieved equally high Q30 scores, all key quality metrics showed substantially tighter distributions for AM‑DMF libraries, underscoring the enhanced process control, reproducibility, and operational consistency of the automated AM-DMF workflow (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee).\u003c/p\u003e \u003cp\u003eIn summary, the sealed, droplet-based design of the AM‑DMF lib-prep platform not only maintains sample‑specific pathogen signatures without introducing detectable cross‑contamination, but also consistently generates higher‑complexity libraries with reduced technical variability. These results thoroughly validate the platform's reliability for clinical mNGS applications requiring parallel sample processing.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eAnalytical Performance of the AM-DMF Library Preparation Platform\u003c/h3\u003e\n\u003cp\u003eRigorous analytical validation is critical to establish the diagnostic reliability of any library preparation method within an mNGS workflow. Here, we evaluated the fully integrated AM‑DMF lib-prep platform across key metrics for library construction: analytical sensitivity (limit of detection), specificity, interference tolerance, and reproducibility.\u003c/p\u003e \u003cp\u003eThe limit of detection (LoD) was determined using a panel of five clinically relevant pathogens (\u003cem\u003eStaphylococcus aureus, Klebsiella pneumoniae, Legionella pneumophila, Cryptococcus neoformans\u003c/em\u003e, and \u003cem\u003eAdenovirus\u003c/em\u003e) across a concentration range of 50\u0026ndash;1000 copies/mL spiked into 10⁵ cells/mL Jurkat cells. The experimentally determined LoDs were 50, 100, 125, 500, and 1000 copies/mL, respectively. The on-AM-DMF libraries achieved the 100 copies/mL benchmark for all bacterial and fungal targets, and four of the five pathogens were detectable at or below 100 copies/mL. A strong linear correlation was observed between the input microbial titer and sequencing reads (RPM) across the concentration range tested, confirming that the AM-DMF Lib-prep platform preserves the quantitative relationship between input abundance and sequencing readout over clinically relevant concentrations (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003eSpecificity was assessed using defined mixtures (1:3, 1:1 and 3:1) of closely related species (\u003cem\u003eStaphylococcus aureus\u003c/em\u003e and \u003cem\u003eStaphylococcus epidermidis\u003c/em\u003e). Libraries generated on-AM-DMF accurately recapitulated the expected abundance ratios in all mixtures, demonstrating on-AM-DMF libraries maintain high specificity without significant competitive bias (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb). Furthermore, the workflow proved robust against common assay interferents. Although the addition of gentamicin (0.1 mg/mL) or mucin (0.5 mg/mL) caused modest fluctuations in absolute RPM values, all target pathogens remained reliably detectable, and the original microbial community composition faithfully reproduced, confirming resilience to inhibitors (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec).\u003c/p\u003e \u003cp\u003eA core advantage of the automated AM‑DMF workflow is its superior technical reproducibility. The intra‑method coefficient of variation (CV) for pathogen RPMs across technical replicates was 25.6% lower for on‑AM‑DMF preparations compared to manual processing, indicating substantially reduced technical variability (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed). In addition, excellent inter-chip consistency was demonstrated, as CVs across different AM‑DMF chip batches showed no significant difference. This was further supported by strong correlations of pathogen RPMs both between AM‑DMF and manual libraries and across independent AM‑DMF chips (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee).\u003c/p\u003e \u003cp\u003eIn summary, the AM‑DMF library preparation platform constitutes a clinic-ready solution, demonstrating excellent analytical sensitivity, linear quantitative response, and robustness against common interferents. Most importantly, it delivers substantially improved precision and reproducibility\u0026mdash;both within and across chips\u0026mdash;compared to manual methods. These performance characteristics strongly support its suitability for standardized clinical mNGS testing.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eValidation of the AM-DMF Library Preparation Platform Using Diverse Clinical Specimens\u003c/h3\u003e\n\u003cp\u003eTo evaluate the performance of the AM‑DMF library preparation platform in a clinical context, a retrospective analysis was conducted using 94 patient specimens with previously established mNGS-based diagnoses. The cohort comprised 84 clinically positive cases and 10 negative cases. The specimens encompassed a range of types, including whole blood (n\u0026thinsp;=\u0026thinsp;30), bronchoalveolar lavage fluid (BALF; n\u0026thinsp;=\u0026thinsp;30), sterile body fluids (n\u0026thinsp;=\u0026thinsp;19), urine (n\u0026thinsp;=\u0026thinsp;4), pus (n\u0026thinsp;=\u0026thinsp;4), and tissue (n\u0026thinsp;=\u0026thinsp;7) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea-b). Among the positive cases, 44.1% were single-pathogen infections and 55.9% were mixed infections (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec-d).\u003c/p\u003e \u003cp\u003eSequencing analysis of on-AM‑DMF libraries correctly identified 82 of the 84 positive cases and all 10 negative cases, achieving 97.7% overall concordance with the reference clinical reports. This corresponds to a positive percent agreement (PPA) of 97.4% and a negative percent agreement (NPA) of 100% (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ee). Quantitative comparison of pathogen readout (RPM) showed strong agreement between AM‑DMF‑prepared and manually prepared libraries: Deming regression demonstrated a highly linear correlation (slope\u0026thinsp;=\u0026thinsp;1.474, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0001; Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ef), and Bland\u0026ndash;Altman analysis confirmed no significant systematic bias (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eg). For low‑abundance pathogens (RPM\u0026thinsp;\u0026le;\u0026thinsp;1), detection rates obtained from AM‑DMF libraries were comparable to those from conventional libraries (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.6941; Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eh). The two discordant cases (false negatives) involved samples with extremely low pathogen loads, high host nucleic acid background, and a history of freeze‑thaw cycles, reflecting the inherent stochastic sampling limits of mNGS in extreme low-biomass scenarios.\u003c/p\u003e \u003cp\u003ePerformance under limited DNA input was further evaluated using a subset of clinical samples. For ten positive and two negative samples, all target pathogens were accurately detected from libraries prepared at both 3 ng and 5 ng input, with no significant degradation in library quality (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ei, Supplementary Fig.\u0026nbsp;3a, b). Deming regression analysis of these libraries yielded slopes close to 1 (1.133 at 3 ng, 1.036 at 5 ng) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ej), and Bland‑Altman plots showed\u0026thinsp;\u0026gt;\u0026thinsp;95% of data points within the limits of agreement, confirming strong quantitative concordance even with scarce input material (Supplementary Fig.\u0026nbsp;3c).\u003c/p\u003e \u003cp\u003eCollectively, these findings establish that the AM‑DMF library preparation platform delivers high pre‑analytical accuracy, producing libraries whose downstream sequencing results show strong quantitative concordance with those from standard clinical workflows across a wide range of challenging specimen types. Taken together, AM-DMF-Lib-prep\u0026rsquo;s reliability, robustness, and validated clinical concordance demonstrate its readiness for clinical deployment.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eThe translation of mNGS from a powerful research tool into a routine, reliable clinical assay has been persistently constrained by the pre-analytical library preparation phase. Conventional methods force a trade-off between performance, cost, labor, and standardization. Here, we present an AM‑DMF platform that offers a transformative approach to these challenges by consolidating the entire \u0026ldquo;sample-in, library-out\u0026rdquo; workflow. This work demonstrates how AM‑DMF addresses the tripartite challenge of automation, performance, and cost in a unified manner, potentially changing how clinical mNGS is implemented.\u003c/p\u003e \u003cp\u003eA primary advance of this platform is its capacity to enable end-to-end standardization. By performing all steps within a sealed, programmable droplet system, it dramatically reduces the inter-operator and inter-batch variability inherent in manual protocols. The negligible cross-contamination and high inter-chip reproducibility (R\u0026sup2; = 0.9998) stem directly from this integrated design, creating a level of operational robustness that is critical for deploying consistent mNGS testing across diverse laboratories\u0026mdash;a key step towards establishing universally comparable diagnostic criteria.\u003c/p\u003e \u003cp\u003eThe analytical performance translates directly into diagnostic utility. The negligible cross-contamination and excellent inter-chip reproducibility (R\u0026sup2; = 0.9998) ensure high test specificity and consistent run-to-run performance, mitigating risks of false positives and technical variability. Crucially, the 43.7%\u0026ndash;76.7% increase in sensitivity for low-abundance pathogens addresses a core diagnostic gap. This enhancement, validated down to 3 ng input, is particularly relevant for improving diagnostic yield in paucibacillary samples like cerebrospinal fluid, tissue biopsies, or specimens from antibiotic-treated patients.\u003c/p\u003e \u003cp\u003eFrom a practical implementation standpoint, the platform offers compelling operational advantages. The ~\u0026thinsp;90% reduction in reagent consumption significantly lowers the direct cost per test. Combined with minimal hands‑on time and reduced rates of assay failure or repeat testing due to contamination, these factors enhance the financial and operational feasibility of integrating mNGS into clinical laboratories. The robust performance across a wide input range, reliably down to 3 ng, alleviates a major constraint of conventional protocols for precious, low‑biomass samples. While library generation from 1 ng inputs was feasible in this study, the analytical robustness and clinical sensitivity at this extreme low limit require further validation. Nevertheless, the ability to process samples consistently at lower input amounts could expand access to mNGS testing in settings where specimen volume is inherently limited, such as pediatrics or neurology.\u003c/p\u003e \u003cp\u003eThis study has limitations. Clinical validation was retrospective and single-centered; prospective, multi-center trials assessing impact on clinical decision-making and patient outcomes are the essential next step. While benchmarked against a robust manual protocol, comparisons with other automation systems will further define its competitive landscape. The current chip throughput, while suitable for batch testing, requires scaling up for high-demand laboratories.\u003c/p\u003e \u003cp\u003eTaken together, the AM‑DMF library preparation platform represents more than a technical advance; it is an enabling infrastructure for clinical metagenomics. By concurrently solving the intertwined problems of manual complexity, performance variability, and high cost, it provides a scalable, precise, and economically viable path forward. This work firmly establishes integrated digital microfluidics as a foundational technology capable of unlocking the long-promised, unbiased diagnostic potential of mNGS for routine patient care.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eLibrary preparation\u003c/h2\u003e \u003cp\u003eSequencing libraries were constructed using a commercial kit (Sansure Biotech, for Illumina platforms) according to the manufacturer\u0026rsquo;s protocol, which encompasses DNA fragmentation, end repair and A-tailing, adapter ligation, bead-based cleanup, library amplification, and final cleanup. For the conventional (off-AM-DMF) method, the protocol was executed manually at the standard reaction volumes defined in the kit instruction. For the on-AM-DMF method, the entire workflow was automated and miniaturized on the digital microfluidic chip. All reagent dispensing, droplet merging, mixing, incubation, and magnetic bead-based purification steps were programmatically controlled. Key reaction conditions (e.g., incubation time and temperature for tagmentation, ligation, and PCR) strictly followed the kit specification. The precise volume of each reagent used for both the on-AM-DMF and conventional protocols is detailed in Supplementary Table\u0026nbsp;1.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eDNA sequencing\u003c/h2\u003e \u003cp\u003ePurified libraries were quantified using a Qubit Fluorometer, and their fragment size distributions were analyzed on an Agilent 4200 TapeStation system with High Sensitivity D5000 ScreenTapes. Qualified libraries were sequenced on an Illumina Nextseq 500 platform. For bioinformatic analysis, reads were processed through a standardized pipeline to remove low-quality sequences and host-derived reads (alignment to the human reference genome hg38). The remaining reads were classified by alignment to pathogen databases. A positive pathogen identification required meeting predefined thresholds for read count, genome coverage, and reads per million (RPM), consistent with established clinical mNGS criteria\u003csup\u003e\u003cspan additionalcitationids=\"CR22 CR23\" citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eAnalytical Performance Assessment\u003c/h2\u003e \u003cp\u003eThe limit of detection (LoD) was determined using a panel of five pathogens: \u003cem\u003eStaphylococcus aureus, Klebsiella pneumoniae, Legionella pneumophila, Cryptococcus neoformans\u003c/em\u003e, and \u003cem\u003eAdenovirus\u003c/em\u003e. Certified genomic copy numbers of \u003cem\u003eS. aureus, K. pneumoniae, L. pneumophila\u003c/em\u003e, and \u003cem\u003eC. neoformans\u003c/em\u003e were provided and measured by BeNa Culture Collection (Henan, China). \u003cem\u003eAdenovirus\u003c/em\u003e standards were from BDS Biological Technology (Guangzhou, China). Serial dilutions of each pathogen (50\u0026ndash;1000 copies/mL) were prepared in a background of 10⁵ cells/mL Jurkat cells and processed on the AM-DMF platform. The LoD was defined as the lowest concentration at which the pathogen was consistently detected. Linearity was assessed by plotting the log-transformed input concentration against the log-transformed RPM values from the same dilution series, and a linear regression was performed to calculate the coefficient of determination (R\u0026sup2;). Gentamicin (0.1 mg/mL) and mucin (0.5 mg/mL) were added into the microbial community standard spike-in sample (ZymoBIOMICS\u0026trade; Microbial Community Standard spiked into 10⁵ cells/mL Jurkat cells) to evaluate interference from substances commonly encountered in clinical samples. The impact of these interferents on pathogen detection efficiency was quantified through RPM value analysis. To evaluate specificity in co-infections, mixtures of \u003cem\u003eStaphylococcus aureus\u003c/em\u003e and \u003cem\u003eStaphylococcus epidermidis\u003c/em\u003e were prepared at defined genomic copy ratios (1:3, 1:1, 3:1). The observed RPM ratios in the resulting libraries were compared to the expected input ratios. To comprehensively evaluate the precision of the AM-DMF platform, we assessed intra-chip, inter-chip, and inter-method reproducibility using the microbial community standard spike-in sample (ZymoBIOMICS\u0026trade; Microbial Community Standard spiked into 10⁵ cells/mL Jurkat cells). For intra-chip precision, the same sample was processed in five replicate reaction units on the same AM-DMF chip. For inter-chip precision, the same sample was processed on two independent AM-DMF chip batches (designated Chip Batch 1 and 2), with five replicates per batch. For inter-method precision, the same sample material was also processed in parallel using the conventional manual library preparation protocol (n\u0026thinsp;=\u0026thinsp;5), serving as the reference. Precision was quantified by calculating the coefficient of variation (CV%) of the reads per million (RPM) values for each constituent pathogen across the replicate measurements within each group. The quantitative concordance between the AM-DMF platform and the manual method, as well as between the two chip batches, was further evaluated using Pearson correlation analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eFabrication of the AM-DMF chip\u003c/h2\u003e \u003cp\u003eThe digital microfluidic (DMF) chip is a sealed, two-plate device fabricated through a combination of flat-panel display manufacturing and micro-assembly processes. The bottom plate was manufactured on a glass substrate using a standard flat-panel display process. It features an active-matrix array independently addressable thin-film transistor (TFT) pixel electrodes. A 300 nm-thick silicon nitride (SiNx) layer was deposited uniformly over the electrode array as the dielectric layer. The top plate consists of a flat, approximately 2-inch indium tin oxide (ITO)-coated glass sheet. Laser-drilled micro-holes (1 or 2 mm in diameter) serve as fluidic access ports for sample and reagent introduction/extraction. Prior to assembly, both the top ITO plate and the bottom TFT plate were ultrasonically cleaned and coated with a hydrophobic layer by spin-coating and baking Teflon-AF (1% w/w in Fluorinert FC-40). The two plates were aligned and bonded with a precisely controlled gap of 50 \u0026micro;m, maintained using UV-curable adhesive or plastic spacers. This cavity facilitates droplet formation and transport. The final chip packaging, including bonding flexible printed circuit (FPC) connectors to the chip via a standard hot-press process, was completed in a Class 10,000 cleanroom facility (ACXEL, Foshan, China).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eStudy Cohort and Clinical Samples\u003c/h2\u003e \u003cp\u003eResidual fluid and tissue samples were collected from patients who underwent clinically indicated metagenomic next-generation sequencing (mNGS) testing at the Department of Laboratory Medicine, Sichuan Provincial People\u0026rsquo;s Hospital, School of Medicine, University of Electronic Science and Technology of China. Sample types included whole blood, bronchoalveolar lavage fluid (BALF), cerebrospinal fluid (CSF), ascites, synovial fluid, pus, urine, abscess drainage, biopsy tissue, and bone marrow. Inclusion criteria were: (1) availability of sufficient sample volume (\u0026ge;\u0026thinsp;600 \u0026micro;L) after routine testing, and (2) complete accompanying clinical information. Exclusion criteria were: (1) insufficient sample volume, or (2) prior failure of the clinical mNGS assay. All samples were collected between May and September 2025 as part of standard clinical care, and stored at \u0026minus;\u0026thinsp;80\u0026deg;C until nucleic acid extraction.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eSample Processing and Nucleic Acids Extraction\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eAll samples were initially inactivated at 56\u0026deg;C for 30 minutes in a dry heating block, followed by type-specific pretreatments: blood samples were centrifuged at 1,600 \u0026times; g for 10 min at 4\u0026deg;C to separate plasma; clear cerebrospinal fluid (CSF) was concentrated by centrifugation at 13,000 \u0026times; g for 10 min, after which the supernatant was removed and the pellet was resuspended in approximately 100 \u0026micro;L of the residual fluid; viscous specimens (e.g., purulent CSF, bronchoalveolar lavage fluid, or pus) were mixed 1:1 (v/v) with 0.1 M dithiothreitol (DTT) and incubated at room temperature for 30 min to reduce viscosity; and tissue specimens were minced into fine fragments using sterile ophthalmic scissors. Following pretreatment, total nucleic acids were extracted from 350 \u0026micro;L of each processed lysate using a commercial Pathogen DNA/RNA Extraction Kit (Sansure Biotech, Changsha, China) according to the manufacturer\u0026rsquo;s instructions. To enhance the detection sensitivity for low-biomass samples, extracted nucleic acids were concentrated using a 1.8x bead ratio, typically yielding a 3\u0026ndash;5 fold increase in concentration. A no-template control (NTC, nuclease-free water) was processed in parallel with each extraction batch to monitor potential contamination.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003eEthics approval statement: This study was approved by the Institutional Review Board of Sichuan Provincial People\u0026apos;s Hospital (Approval 2025 No.557]). All procedures were performed in accordance with the relevant guidelines and regulations. Consent statement: Informed consent was waived by the approving ethics committee due to the retrospective nature of the study using residual clinical specimens, which were collected during routine clinical care and anonymized prior to analysis\u003c/p\u003e\u003cp\u003e \u003ch2\u003eCOMPETING INTERESTS\u003c/h2\u003e \u003cp\u003eOne patent has been filed based on this work.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eAUTHOR CONTRIBUTIONS\u003c/h2\u003e \u003cp\u003eHM, LJ and ZY conceived and designed the study. YW and YC did statistical analyses and wrote the manuscript. YW, BZ, TW, YY, YC, JL, YiW, JJ, SH, JZ and LY contributed to data collection and the overall structure of the study. HM, LJ and ZY reviewed the manuscript and supervised the study. All authors read and approved the final manuscript and had final responsibility for the decision to submit for publication.\u003c/p\u003e\u003ch2\u003eACKNOWLEDGMENTS\u003c/h2\u003e \u003cp\u003eThis work was supported by National Key R\u0026amp;D Program of China (2023YFF0721500), Sichuan Science and Technology Program (2025ZDZX0103) and the Suzhou Basic Research Pilot Project (SSD2023013).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eGu W, Miller S, Chiu CY (2019) Clinical Metagenomic Next-Generation Sequencing for Pathogen Detection. Annu Rev Pathol 14:319\u0026ndash;338. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1146/annurev-pathmechdis-012418-012751\u003c/span\u003e\u003cspan address=\"10.1146/annurev-pathmechdis-012418-012751\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang L et al (2021) Dynamic blood single-cell immune responses in patients with COVID-19. Signal Transduct Target Ther 6:110. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41392-021-00526-2\u003c/span\u003e\u003cspan address=\"10.1038/s41392-021-00526-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGong B et al (2022) A genetic variant in IL-6 lowering its expression is protective for critical patients with COVID-19. Signal Transduct Target Ther 7:112. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41392-022-00923-1\u003c/span\u003e\u003cspan address=\"10.1038/s41392-022-00923-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXie C et al (2020) Comparison of different samples for 2019 novel coronavirus detection by nucleic acid amplification tests. Int J Infect Dis 93:264\u0026ndash;267. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.ijid.2020.02.050\u003c/span\u003e\u003cspan address=\"10.1016/j.ijid.2020.02.050\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSimner PJ, Miller S, Carroll KC (2018) Understanding the promises and hurdles of metagenomic next-generation sequencing as a diagnostic tool for infectious diseases. Clin Infect Dis 66:778\u0026ndash;788\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTegally H, San JE, Giandhari J, de Oliveira T (2020) Unlocking the efficiency of genomics laboratories with robotic liquid-handling. BMC Genomics 21:729. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12864-020-07137-1\u003c/span\u003e\u003cspan address=\"10.1186/s12864-020-07137-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRackus DG, Shamsi MH, Wheeler AR (2015) Electrochemistry, biosensors and microfluidics: a convergence of fields. Chem Soc Rev 44:5320\u0026ndash;5340\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang D et al (2022) Thin-film transistor arrays for biological sensing systems. Flex Print Electron 7:023004\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang D, Jiang S, Ma H, Yu J, Nathan A (2025) Active-matrix digital microfluidics for high-throughput, precise droplet handling. Nat Reviews Electr Eng, 1\u0026ndash;15\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu S et al (2023) Large-area electronics-enabled high-resolution digital microfluidics for parallel single-cell manipulation. Anal Chem 95:6905\u0026ndash;6914\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo Z et al (2025) Deep Learning-Assisted Label‐Free Parallel Cell Sorting with Digital Microfluidics. Adv Sci 12:2408353\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJia Z et al (2025) Intelligent single-cell manipulation: LLMs-and object detection-enhanced active-matrix digital microfluidics. Microsystems Nanoengineering 11:133\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJi J et al (2025) AM-DMF‐ddRPA: An All‐in‐One Digital Microfluidic Platform for Rapid and Automatic Digital Nucleic Acid Analysis. Angew Chem Int Ed, e202501913\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang Z et al (2024) AM-DMF-SCP: integrated single-cell proteomics analysis on an active matrix digital microfluidic chip. JACS Au 4:1811\u0026ndash;1823\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi H et al (2024) Droplet-based microfluidics with mass spectrometry for microproteomics. Engineering 43:37\u0026ndash;53\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu J et al (2024) Rapid and multi-target genotyping of Helicobacter pylori with digital microfluidics. Biosens Bioelectron 256:116282\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu S et al (2022) All-in-one digital microfluidics system for molecular diagnosis with loop-mediated isothermal amplification. Biosensors 12:324\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHess JF et al (2020) Library preparation for next generation sequencing: A review of automation strategies. Biotechnol Adv 41:107537. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.biotechadv.2020.107537\u003c/span\u003e\u003cspan address=\"10.1016/j.biotechadv.2020.107537\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim H et al (2013) A microfluidic DNA library preparation platform for next-generation sequencing. PLoS ONE 8:e68988\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim S et al (2017) High-throughput automated microfluidic sample preparation for accurate microbial genomics. Nat Commun 8:13919\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchlaberg R, Chiu CY, Miller S, Procop GW, Weinstock G (2017) Validation of Metagenomic Next-Generation Sequencing Tests for Universal Pathogen Detection. Arch Pathol Lab Med 141:776\u0026ndash;786. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5858/arpa.2016-0539-RA\u003c/span\u003e\u003cspan address=\"10.5858/arpa.2016-0539-RA\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBittinger K et al (2014) Improved characterization of medically relevant fungi in the human respiratory tract using next-generation sequencing. Genome Biol 15:487. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13059-014-0487-y\u003c/span\u003e\u003cspan address=\"10.1186/s13059-014-0487-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDoughty EL, Sergeant MJ, Adetifa I, Antonio M, Pallen MJ (2014) Culture-independent detection and characterisation of Mycobacterium tuberculosis and M. africanum in sputum samples using shotgun metagenomics on a benchtop sequencer. PeerJ 2:e585. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7717/peerj.585\u003c/span\u003e\u003cspan address=\"10.7717/peerj.585\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiao Q et al (2018) Microbiological diagnostic performance of metagenomic next-generation sequencing when applied to clinical practice. Clin Infect Dis 67:S231\u0026ndash;S240\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8889289/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8889289/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Clinical adoption of metagenomic next‑generation sequencing (mNGS) is limited by costly, labor-intensive library preparation requiring substantial DNA input. Here, we present an active‑matrix digital microfluidics library preparation platform (AM‑DMF-Lib-prep) that overcomes these barriers by automating the “sample‑in, library‑out” workflow in an enclosed, digitalized lab-on-a-chip system. The platform accommodates a wide DNA input range (1–25 ng), producing high-quality libraries with exceptional reproducibility (inter-chip R² = 0.9998) and 25.6% lower variability than conventional manual methods. It achieves ~90% reagent reduction while enhancing detection sensitivity for low-abundance pathogens by 43.7–76.7% in microbial community standard spike-in sample. Validation across 94 diverse clinical specimens confirmed high concordance with reference diagnostics. Collectively, AM‑DMF-Lib-prep establishes a precise, cost‑effective, and scalable solution with demonstrated clinical readiness, directly addressing the pre‑analytical bottleneck in routine mNGS.","manuscriptTitle":"AM‑DMF-Lib-prep: A clinical ready lib-prep-on-a-chip platform with active-matrix digital microfluidics","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-03 14:38:12","doi":"10.21203/rs.3.rs-8889289/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"6e22f4e1-ac3c-4cb5-94e0-6a1bd5c0b325","owner":[],"postedDate":"March 3rd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":63279267,"name":"Health sciences/Molecular medicine"},{"id":63279268,"name":"Physical sciences/Engineering/Biomedical engineering"}],"tags":[],"updatedAt":"2026-03-14T22:40:22+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-03 14:38:12","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8889289","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8889289","identity":"rs-8889289","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}