The salient feature of the FieldNA device is the vertical stacking of the various fluidic modules, a collection of components designed to perform DNA isolation, as shown in Figures 1A,B. The topmost module consists of components a and b, which allows for sample loading and an incubation period. Through aligned holes and incline profiles, fluidic sample is regulated downwards in the separation plate (component c) and then towards the magnetic capture module (component d; Figures 1A,C,D), which is aligned and supported using a housing structure (component e). The purified DNA is collected in the bottom elution plate, component f, which again consists of a unique profile allowing for easy transfer of eluted DNA. The main goal for the device design is to facilitate gravity-driven fluid flow, with flow control implemented at various stages using modular components. Several optimization steps were implemented in printing of individual modules, most notably the magnetic capture module shown in Figures 1C,D. It consists of two components: a bottom surface with a 45^o inclined flow profile with an in-house synthesized polyvinylidene fluoride (PVDF) membrane, and a top enclosure with a built-in sluice gate-like flow control structure to control the speed and alignment of fluid flow over a Neodymium disc magnet (Diameter: 6 mm, height: 3 mm, Applied Magnets). Additional modules may be stacked above or below the current setup, depending on unique target application requirements.

The device was designed in SolidWorks 2022, and the reported prototype has a total footprint of approximately 52 mm × 52 mm with a height of 50 mm. In the outlet of component b in Figure 1, there is a PDMS circular insert to facilitate flow focusing and to reduce dead volume. The magnetic bead capture module is designed to be a single-use disposable unit which may be replaced following an extraction experiment. The design was processed for stereolithography (SLA) 3D printing using CHITUBOX (v1.9.5) with the following settings: exposure time: 10 s, lift distance: 7 mm, lift speed: 90 mm/min, bottom exposure time: 50 s, layer height: 0.050 mm, retract speed: 320 mm/min. Finally, printing was carried out using a Elegoo (Shenzen, China) Saturn 3 Ultra 3D printer, with Grey ABS-like resin UV Resin, as the printing material. The parts were post-processed in an isopropyl alcohol (IPA) bath followed by 10 min of further UV curing using the Elegoo Mercury X wash and curing stations. All components were coated in a Teflon coating (4,5-Difluoro-2,2-bis(trifluoromethyl)-1,3-dioxole, polymer with Tetrafluoroethylene).

Once the user loads the solution containing lysate media and bead-bound DNA into the device through the top module (a and b), the solution is retained for a desired incubation period. Following this, the content is driven downwards into the magnetic capture module by aligning the notch features in module b and c, achieved through rotation. This allows the solution to enter capture module d, where it flows over the inclined PVDF film covering the disc magnet. The magnet housed in the bottom structure captures the beads and with help from the control structure, the beads are immobilized uniformly on the PVDF film. The remaining solution passes through flowing downwards into a waste collection plate, used to collect all discard lysis and wash buffers. The magnetic beads, now immobilized, are in place for the washing, drying and elution steps. For the final step, when the elution media is added to detach purified DNA from beads, the waste collection plate is swapped with the elution plate mentioned above.

Olive leaf and olive oil samples were provided to BioCoS by Aprol Umbria (Umbria, Italy), including different olive varieties. Olive oil samples used in this study corresponded mainly to 2024 harvest year (EVOO-24), with a handful of experiments utilizing samples from 2023 harvest year (EVOO-23). The isolated DNA from olive leaf samples served as a positive control for the real-time PCR, coupled with High Resolution Melting (HRM) analysis. In this study, the Italian cultivar “Moraiolo” was utilized for both olive leaf and olive oil DNA isolation experiments. Due to the limited availability and the observed poor DNA quality in the older batch (EVOO-23), the experimental results and statistical analysis presented in the main manuscript text correspond to the EVOO-24 samples. Results obtained from EVOO-23 samples are reported separately in Supplementary attachment 2.

Genomic DNA was extracted from cv. Moraiolo young olive leaf tissue using a modified CTAB method (Cipriani et al., 2002), which is found to be well-suited for plants with high polysaccharide and polyphenol content. 200 mg of fresh olive leaves were frozen in liquid nitrogen and ground to a fine powder. The ground tissue was mixed with 1 mL of pre-warmed (at 65 °C) CTAB extraction buffer (2% CTAB, 1.4 M NaCl, 100 mM Tris-HCl pH 8.0, 20 mM EDTA, 0.2% β-mercaptoethanol) and incubated at 65 °C for 30 min to facilitate cell lysis, followed by centrifugation for 15 min at 14000rpm at 4 °C. After the recovery of the clean lysate into a new tube, an equal volume of chloroform:isoamyl alcohol (24:1) was added, and the mixture was centrifuged for 15 min at 14000rpm at 4 °C. The aqueous phase containing DNA was transferred to a new tube, and DNA was precipitated with 100% ice-cold isopropanol at −20 °C for 30 min. After centrifugation for 20 min at 14000rpm at 4 °C, the DNA pellet was washed twice with 70% ethanol and air-dried. The DNA pellet was finally resuspended in 50uL of TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0) and stored at −20 °C until further analysis.

Six different DNA isolation protocols for Olive Oil were implemented in this work. All samples were extracted in quadruplicate (n = 4) using each protocol, and the yield and purity of the extracted DNA was measured using BioPhotometer D30 (Eppendorf). Figure 2 presents a detailed process flow diagram for each of the used protocols.

A real-time quantitative PCR (qPCR) coupled with HRM software was carried out to determine the performance of the different extraction protocols. For this analysis, a publicly available Simple Sequence Repeat (SSR) genetic marker sequence, namely, ssrOeUA-DCA3 (F-CCCAAGCGGAGGTGTATATTGTTAC, R-TGCTTTTGTCGTGTTTGAGATGTTG), was synthesized and purchased from IDT (Integrated DNA Technologies, United States) and utilized as primer set for the real-time PCR reaction. For the reactions, the protocol described by Gomes et al., 2018 was employed, with minor modifications. More specifically, qPCR amplification of ssrOeUA-DCA3 was performed in a final volume of 20 μL, containing 10 µL 2X Forget-Me-Not™ EvaGreen^® qPCR Master Mix (Biotium CA, United States), 0.5 μM of each primer, and 50 ng of isolated DNA. All qPCR reactions were performed on a Mic qPCR Cycler (Bio Molecular Systems) using the following protocol: a 2-min incubation at 95 °C, followed by 40 cycles of denaturation at 95 °C for 5 s, an annealing step at 50 °C for 10 s, and a 12 s extension at 72 °C. For the melting phase, conditioning was done at 95 °C for 30 s and 60 °C for 60 s, leading to a melt analysis over the temperature range of 65 °C–95 °C, with increments of 0.2 °C/s, with constant green fluorescence signal acquisition.

A 2% agarose gel electrophoresis was set up and the real-time PCR products were run at 50V for 60 min to confirm that the amplification observed in the real-time PCR matched the expected amplicon size, in comparison to the positive control (Supplementary Figure S1.1).

All DNA isolations were conducted in quadruplicates per protocol to capture inherent sample-to-sample variability and to ensure statistical robustness of the measurements. This approach increases the accuracy of variance estimation, reduces the influence of outliers, and provides improved power for subsequent statistical analyses, while maintaining experimental feasibility. To assess potential variability into the DNA isolations, two operators were involved, each having different laboratory experience (one is a senior scientist and the other a junior scientist). This way we can also address the user-friendliness of the FieldNA. Data obtained from the photometer (DNA concentration and A_260/280 ratio) were analyzed in R v 4.5.1 software using the car packages to assess homogeneity of variances with Levene’s test. Normality within groups was evaluated using the Shapiro–Wilk test implemented in the base stats package. Depending on the outcome of assumption checks, either a one-way ANOVA followed by Tukey’s HSD post hoc test, or a Kruskal–Wallis test followed by pairwise Wilcoxon comparisons with Holm correction, was performed. A significance threshold of α < 0.05 was applied for all statistical tests. All graphs were prepared with OriginPro, Version 2024b.

The FieldNA device is designed to integrate external cell lysis and incubation with magnetic bead-based isolation and purification into its workflow. After lysis and incubation, wash buffers containing specific reagents as described in the methods Section 2.3, can be introduced into the device’s input module for DNA purification on immobilized magnetic beads. The gravity-driven passive flow simplifies the device operation, as it requires no electrical power and ensures efficient transport of the sample to the capture module, optimizing the DNA isolation process. Other 3D printed platforms have incorporated equipment such as syringe pumps or used capillary forces, etc., to move fluid into and through their devices (Kibar et al., 2024; Kirkoyun Uysal et al., 2024; Holladay et al., 2025). It allows them to control flow rates more precisely and set specific flow rates, however complexity of the device increases.

Through flow validation tests, the average time taken for the fluid to flow into the magnetic capture module was determined to be within 1–2 min. Once in the capture module, the slant was designed to accommodate both the flow-through of the fluid, as well as to allow for the capture of the beads by the magnetic field. COMSOL Multiphysics simulation of the magnetic capture module design concept was conducted to verify the influence of the magnetic field in the capture module operation, as seen in Figure 3. Figure 3A shows the schematic rendition of the capture mechanism. The magnetic beads (diameter 1–3 μm) are located in the fluid layer as they pass through the capture module. The material of the surrounding area outside the cartridge was set to air with a boundary condition of magnetic insulation. Simulations were conducted using the magnetic fields, no current (mfnc) interface. The relevant parameters are relative permeability of permanent magnet 100, permanent magnet x axial remanence 0, permanent magnet y axial remanence 0, permanent magnet z axial remanence ±0.385 T. Figures 3B,C show the magnetic flux density distribution in this region, with Figure 3B’ denoting a cut plane in the fluid region. From literature, external magnetic fields ranging from 0.1 to 0.3 T are sufficient to influence and capture beads (Yadav et al., 2024; Dowling and Kostylev, 2025). From the surface plots, it is evident that the magnetic field permeates through the PVDF film, the thin polymer structural layer, and the fluid layer, with sufficient field strength to attract the beads in the fluid layer and capture them onto the PVDF membrane.

Following the initial proof-of-concept tests to demonstrate bead capture, a sluice gate-like control structure was designed to slow the flow of fluid and to facilitate a wider surface area for bead capture. As seen in the left-most film in Figure 3D, the proof-of-concept tests showed the feasibility of this process but resulted in a greater loss of sample and bead volumes. A flow control structure was introduced to combat this issue. Initially, this structure was designed as an external component inserted into the magnetic capture module (Figure 3D, centered figure), but it was later merged into the top component (Figure 3D, right-most figure). By introducing a large gate, the flow rate of the solution is reduced, and the fluid is concentrated over the magnet allowing for more capture time before passing through the module. This structure also serves another purpose in aiding with magnetic bead spread on the PVDF film. From the initial tests, the magnetic beads were captured on the film but lacked a uniform spread and tended to clump together. The structure was designed to increase the contact time of the beads with the film, enhancing surface interactions and resulting in greater immobilization and higher bead concentration. This was qualitatively validated for the various control structures based on the same amount of bead loading. The beads are spread horizontally widthwise across the film, aligned with the control structure. From left to right, the immobilized bead concentration increases due to optimizations in control structure design. It should be noted that the wash buffers have high concentrations of ethyl alcohol (ethanol). The behavior of this fluid is more unpredictable and harder to restrain (Srisomwat et al., 2024) so a wider spread in bead capture was desirable to allow for greater wash efficiency to accommodate for this variability in fluid flow. This was established through spectrophotometer characterization of the eluted DNA obtained from various prototypes designs with the flow control structures shown in Figure 3D.

Another motivation for the sluice gate control structure is improving the quality of the releases DNA. Purification is crucial for assuring the quality and yield of DNA. The control structure allows for a wider spread of the wash buffers as well as a lower flow rate for the elution buffer. The lower flow rate allows for longer elution buffer incubation time, mimicking the incubation step in commercial DNA extraction column kits where the elution buffer is deposited on the column and incubated at room temperature before centrifuging into an elution collection tube. By slowing down this flow through speed, the purification process has been optimized for this system.

To evaluate the different protocols, we applied criteria including average DNA yield and purity (A_260/280 ratio), with results presented in Table 1 as mean values from four replicates. In addition, a per-sample quantitative assessment of solvent use (liquid waste), plastic tubes, and pipette tips was performed, and these data are also summarized in Table 1.

Across the six protocols, DNA yield (in ng/μL) values differed markedly (one-way ANOVA: F (5,18) = 30.97, p = 3.06 × 10⁻⁸), driven by substantially higher measurements for P1. Tukey’s HSD confirmed that P1 exceeded each of P2–P6 (all p_adj<0.001), while no pairwise differences among P2–P6 were significant after multiplicity correction. In a sensitivity analysis restricted to P2–P6, the omnibus effect was not significant (F (4,15) = 2.82, p = 0.063) and no adjusted pairwise contrasts reached significance, indicating that, apart from P1, the remaining methods performed comparably in this dataset (Supplementary Table S1.2). On the other hand, regarding the ratio A_260/280, P1 maintained acceptable purity ranting between 1.5 and 2.1, while on the other extreme P4 showed poor purity (0.79–1.00). P6 purity ranged between 1.44 and 1.97, which was comparable with P1, P2 and P5. Overall, from Figure 4B, it is clear that the silica-based spin columns (P3 and P4) provided the poorest purity compared to the rest of the methods. The statistical significance is shown in Figure 4B.

The performance of the different DNA isolation protocols was subsequently evaluated using real-time PCR to assess parameters such as cycle values (Cq), PCR efficiency, efficiency R², and melting temperature (T_m) (Supplementary Table S1.1). It is anticipated that the DNA isolated from the olive leaf samples will exhibit a lower Cq value compared to the DNA isolated from olive oil. This is mainly attributed to the DNA degradation that occurs naturally in the olive oil, since the genomic material is–to its greater degree–stripped and flows freely in the fatty matrix. The main reason for this “free DNA” is the disruption of the olive fruit cells during the crushing (and malaxation step of olive paste), before the olive oil extraction step.

Real-time PCR coupled with High-Resolution Melting (HRM) demonstrated marked performance differences among DNA isolation methods (Figures 5A,B). The positive control (olive leaf gDNA) amplified earliest, as expected, with Cq ≈ 17.75, amplification efficiencies of 0.70–0.74, and excellent linearity (R² > 0.999), confirming assay robustness. Among the different methods of olive oil isolation, P6 yielded the earliest amplification (Cq ≈ 35) with efficiency ≈0.76 and R² > 0.999. P1 showed reproducible amplification at slightly later cycles (Cq 36.3–38.3), efficiencies of 0.73–0.79, and similarly high linearity (R² > 0.999). P5 amplified substantially later (Cq ≈ 41.9), indicative of lower template input despite acceptable efficiency (∼0.79) and high linearity. P3 exhibited variability (Cq 36.0 and 40.8; efficiency 0.72–0.74), consistent with uneven recovery. P4 did not amplify under the conditions tested. No amplification was detected in the no-template control (NTC).

Melting analysis (HRM derivatives) corroborated these results (Figure 5B). The positive control produced a single amplicon with T_m = 77.67 °C. P1 (77.61 °C) and P6 (77.76 °C) matched this T_m and were the only methods to pass the instrument-defined high-stringency quality filter, supporting specific target amplification. This was further supported by the results of the gel electrophoresis carried out in the PCR amplicons (Supplementary Figure S1.1). P2, P3, and P5 displayed T_m values concordant with the expected product but failed the quality filter—consistent with low template abundance and/or suboptimal melt-curve metrics. Overall, P6 provided the most reliable template for downstream real-time PCR, with P1 performing comparably; the remaining kits produced higher Cq values and less robust melt profiles.