Droplet Digital CRISPR for Nucleic Acid Detection
Yang Zhang, Roy S. K. Walker, Anwar Sunna, Tracie J. Barber, Ming Li
TL;DR
Droplet digital CRISPR combines CRISPR's precision with droplet technology to detect nucleic acids with high sensitivity and accuracy.
Contribution
This review introduces the integration of CRISPR with droplet digital technology for nucleic acid detection and quantification.
Findings
ddCRISPR enables single-molecule resolution and minimizes background interference through picoliter microdroplets.
Amplification-based and amplification-free detection strategies have been advanced for DNA and RNA biomarker detection.
Challenges include workflow automation, droplet stability, and assay portability, with future directions involving AI and point-of-care integration.
Abstract
Droplet digital (dd) clustered regularly interspaced short palindromic repeats (CRISPR) integrates the high sequence specificity of CRISPR‐based nucleic acid detection with the absolute quantification capability of digital droplet microfluidics, offering high sensitivity, precision, and scalability. By partitioning samples into thousands to millions of picoliter microdroplets, ddCRISPR enables single‐molecule resolution and minimizes background interference. This review summarizes the principles of droplet generation, manipulation, and detection in ddCRISPR platforms, as well as recent advances in amplification‐based and amplification‐free detection strategies. Representative applications are highlighted for viral, bacterial, and other DNA/RNA biomarker detection. Current challenges, including workflow automation, droplet stability, multiplexing, and assay portability, are discussed…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7
FIGURE 8
FIGURE 9|
|
|
|
|
|
| |
|---|---|---|---|---|---|---|
| E | N/A | DNA polymerase | DNA polymerase, recombinase | DNA polymerase, NEase | DNA polymerase, recombinase | Ligase, DNA polymerase |
| Primers (count) | N/A | 4‐6 | 2 | 1 | 2 | 2 |
| Temperature (°C) | 37.5 | 60 – 65 | 37–42 | 37 | 37 – 42 | 60 |
| Time (min) | 20‐30 | 60 | 30 – 90 | 30 – 90 | 70 | 90 |
| Amplicon | N/A | DNA | DNA | DNA | DNA | DNA |
| LOD | 1 copy/µL | 1 copy/mL | 100 copies/mL | 1 fM | 10 aM | 100 aM |
|
| |||||
|---|---|---|---|---|---|
| Cas | Target | Amplification | LOD | Refs. | |
|
| Cas12a | HPV 18 | No | 5 copies/µL | [ |
| Cas12a | HPV 18 | RAA | 0.015 fM | [ | |
| Cas13a | HPV 18 | RPA | 10 copies/µL | [ | |
| Cas12a | HPV 16 | No | 60 copies/µL | [ | |
| Cas12a | HPV 16 | RPA | 1 aM | [ | |
|
| Cas12a | African swine fever virus, Epstein‐Barr virus, Hepatitis B virus | No | 17.5 copies/µL | [ |
| Cas13a | JC virus | RPA | 100 copy/mL | [ | |
|
| Cas12a | Klebsiella pneumoniae ( | RAA | 10 aM | [ |
| Cas12a | Salmonella typhimurium ( | LAMP | 1 fM | [ | |
| Cas13a |
| RAA | 10 copies/µL | [ | |
|
| Cas12a | ND1, IL‐2 | No | 11.2 aM | [ |
| Cas12a | Telomerase DNA from breast cancer cells (MCF‐7) | HCA | 62.5 cells/µL | [ | |
- —Australian Research Council10.13039/501100000923
- —National Health and Medical Research Council10.13039/501100000925
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsInnovative Microfluidic and Catalytic Techniques Innovation · Electrowetting and Microfluidic Technologies · CRISPR and Genetic Engineering
Introduction
1
Accurate and sensitive detection of nucleic acids is fundamental to modern molecular diagnostics, enabling early disease diagnosis, monitoring of therapeutic response, and pathogen surveillance. Over the past few decades, polymerase chain reaction (PCR) and its derivatives, such as quantitative PCR (qPCR), have become the gold standard in nucleic acid quantification [1, 2]. These techniques rely on enzymatic amplification of target sequences, enabling detection down to low copy numbers. However, despite their sensitivity, conventional bulk detection methods face several limitations [3]. First, they require professional instruments such as thermal cycler, making them less suitable for point‐of‐care or field applications. Second, bulk reactions reflect the average signal from a mixed population of molecules, masking rare variants and reducing precision. Third, and most critically, PCR amplification introduces quantitative biases due to variable reaction efficiency, primer dimer formation, or nonspecific amplification. These issues pose challenges for absolute quantification, reproducibility, and detection in complex or inhibitor‐rich clinical samples. Moreover, current methods often require laborious workflows, increasing assay time, cost, and the risk of contamination [4].
To overcome these challenges, researchers have explored novel molecular detection strategies that offer high sensitivity, specificity, and portability without sacrificing accuracy or scalability. CRISPR‐based systems have gained attention due to their programmability, high specificity, and compatibility with isothermal conditions [5, 6, 7]. Originally derived from bacterial immune systems [8], CRISPR technologies utilize Cas enzymes to recognize and cleave target nucleic acid sequences [9, 10, 11]. Their unique collateral cleavage activity has been harnessed to develop rapid and sensitive detection platforms [12, 13, 14, 15] such as SHERLOCK [11, 16] (Specific High‐sensitivity Enzymatic Reporter unLOCKing) and DETECTR [5] (DNA Endonuclease‐Targeted CRISPR Trans Reporter) [17, 18]. However, many CRISPR‐based assays still require pre‐amplification to enhance sensitivity, which increases complexity and contamination risks [19, 20, 21]. Additionally, most current CRISPR‐based diagnostic platforms rely on fluorescence, electrochemical detection [22] or lateral flow strip tests [23] that offer only qualitative results, limiting their utility for applications requiring quantitative resolution.
Droplet digital microfluidics has emerged as a complementary and enabling technology due to the absolute quantification of targets and increased detection accuracy [24]. In these assays, bulk samples are partitioned into thousands or even millions of micro‐sized droplets, each containing single entities of interest. Unlike conventional assays that measure bulk concentration, the specific number of target molecules can be determined from the proportion of positive reactions based on the Poisson distribution [25]. This technology was first developed for droplet digital PCR (ddPCR) [26, 27, 28, 29], allowing for absolute target quantification without reliance on standard curves. Over time, advances in chip fabrication, droplet generation techniques, and automation have enhanced accuracy, scalability, and efficiency. By minimizing reagent consumption and reducing cross‐contamination, droplet digital microfluidics has become an essential tool for high‐throughput and precise molecular analysis. It can deliver precise, single‐molecule level and high‐throughput measurement, making it well‐suited for integration with CRISPR‐based diagnostics [30].
Recently, researchers have combined CRISPR's sequence specificity with the precision of droplet digital microfluidic platforms. This combination improves nucleic acid detection, even in complex biological samples. CRISPR/Cas12 and Cas13 systems have been adapted for fluorescence‐based detection inside picolitre‐scale droplets. To further enhance sensitivity, droplet digital CRISPR (ddCRISPR) has been integrated with isothermal amplification methods such as recombinase polymerase amplification (RPA) [31, 32, 33] and loop‐mediated isothermal amplification (LAMP) [34]. In parallel, recent efforts have focused on developing amplification‐free workflows. These approaches enable direct nucleic acid detection, reducing assay complexity and processing time. Additional advances have aimed at improving droplet stability, enabling detection multiplexing, and reducing costs to refine ddCRISPR technologies. As a result, ddCRISPR has been successfully applied to infectious disease diagnostics, cancer liquid biopsies, and environmental monitoring. Although review articles exist on CRISPR‐based biosensing [14, 35] and others have independently summarized advances in digital microfluidics [36] or reviewed CRISPR–microfluidics integration more broadly [37, 38], these works typically address the two fields separately or do not focus on digital quantification. With the rapid increase in studies applying digital droplet microfluidics to CRISPR‐based nucleic acid detection, a detailed and dedicated synthesis of these developments has become necessary.
In particular, ddCRISPR represents a distinct diagnostic paradigm that combines the absolute quantification capability of droplet digital microfluidics with the programmable, sequence‐specific recognition of CRISPR effectors. This integration enables high sensitivity, amplification‐free workflows, reduced background interference, and improved precision compared with conventional ddPCR or bulk CRISPR assays. Despite this progress, no existing review has systematically evaluated ddCRISPR as an emerging class of digital CRISPR technologies.
To address this gap, the present review provides a focused and integrated overview of ddCRISPR for nucleic acid detection. We systematically highlight recent progress in ddCRISPR assays for nucleic acid biomarker detection. We first present the fundamental principles of ddCRISPR systems, including droplet generation, manipulation and detection, target compartmentalization, and CRISPR reaction mechanisms. Next, we summarize current advances in both amplification‐based and amplification‐free ddCRISPR strategies. Representative applications are highlighted, covering the detection of DNA and RNA biomarkers. We further discuss current challenges and emerging opportunities including multiplexing, workflow automation, and point‐of‐care integration, aiming to establish ddCRISPR as a comprehensive quantitative framework for next‐generation molecular diagnostics (see Figure 1).
Overview of droplet digital CRISPR (ddCRISPR) for nucleic acid biomarker detection. The ddCRISPR platform integrates CRISPR‐based detectionoffering sequence‐specific target recognition—with droplet digital microfluidics for precise partitioning and absolute quantification. This approach enables ultrasensitive, high‐throughput, and streamlined molecular diagnostics of nucleic acid biomarkers from liquid biopsy as well as food and environmental samples.
Fundamentals of ddCRISPR Technology
2
The capabilities of ddCRISPR are dependent on its workflow and statistical analysis methods. The workflow includes droplet generation, manipulation and detection techniques as shown in Figure 2. Statistical analysis is crucial for accurate quantification, as the randomness of molecular distribution in droplets must be accounted to achieve precise measurements. These fundamental aspects ensure ddCRISPR provides reliable, high‐precision nucleic acid quantification and exhibits clear advantages over conventional detection methods.
Schematic illustration of workflow and key stages in ddCRISPR. (a) The ddCRISPR workflow begins with sample preparation followed by droplet partitioning, digital detection of CRISPR activation, and Poisson‐based analysis for absolute quantification; (b) Representative droplet generation strategies used in ddCRISPR, categorized as passive (flow‐focusing, T‐junction, co‐flow, step emulsification) and active (electrical, mechanical, thermal, and surface acoustic wave) approaches; (c) Droplet manipulation (mixing, merging, injection and splitting) used to control reagent distribution and reaction timing within ddCRISPR workflows; (d) Detection methods for distinguishing positive and negative droplets, including optical imaging, laser‐initiated fluorescence detection (FADS/FACS), and electrical sensing.
Core Principles of ddCRISPR
2.1
ddCRISPR relies on the principles that a bulk reaction can be divided into thousands or millions of small, isolated droplets [39, 40]. When nucleic acids are randomly distributed across these droplets, each compartment contains either zero, one or a few target molecules. This partitioning provides effective single‐molecule resolution, because a single DNA or RNA molecule confined in a picoliter‐scale droplet reaches a local concentration sufficient to activate Cas enzymes. The presence or absence of Cas activation in each droplet produces a binary outcome that forms the basis of the digital readout.
Specifically, the statistical foundation of ddCRISPR is derived from the partitioning process. Since target molecule loading into droplets follows a Poisson distribution, the number of target molecules per droplet is characterized by the parameter λ which represents the mean number of molecules in each droplet. In practical assays, λ can be obtained from the fraction of negative droplets (*P_0_ *) according to:
where *P_0_
- is the proportion of droplets that show no detectable CRISPR activation. The fraction of positive droplets (*P_1_ *) is therefore:
Meanwhile, the number of targets in the sample (m) is typically less than or on the order of the number of droplets (n). As a result, significant statistical variation exists in the number of molecules per partition. The average number of targets per partition (λ) depends on the sample concentration (C) and the droplet volume (*V_d_ *), and can be expressed as following:
Here, m is the number of targets in the sample.
Therefore, once λ is determined, the absolute concentration of the target in the original sample can be calculated. This is achieved by first multiplying λ by the total number of droplets (n) to obtain the total number of target molecules (m) and then dividing m by the volume of a single droplet (*V_d_ *) to obtain the sample concentration C. This allows ddCRISPR to quantify nucleic acids without calibration curves, threshold fitting, or dependence on reaction time.
Additionally, droplet size in ddCRISPR is typically measured after the droplets are collected and imaged in a static state. Under these conditions, the in‐plane radius obtained from brightfield microscopy is used to estimate droplet volume by assuming a spherical shape, which is the most common practice. However, when droplet size needs to be estimated on‑chip, droplets may adopt flattened Hele‐Shaw or plug‐like shapes while traveling through microchannels, and these non‐spherical profiles can introduce errors in the calculated volume. Such inaccuracies in droplet volume estimation can further affect the determination of the partition volume and the calculated λ value. To address these issues, a correction strategy can be applied: for droplets whose measured diameter exceeds the channel height, a correction factor based on an oblate spheroid model can be used to recalculate the volume more accurately. Alternatively, the fluorescence intensity profile across an individual droplet can serve as a morphological indicator: a parabolic profile confirms spherical morphology, whereas a flattened profile suggests a non‑spherical shape, prompting the application of the geometric correction described above. Furthermore, relevant studies [41, 42] have detailed that in droplet‑based microfluidics, the droplet size can be predicted using an analytic model that minimizes the interfacial energy based on the geometric configuration of the confined flow. This quasi‑static approach [43] relies on a simple mass balance and thermodynamic argument, yielding a droplet diameter that depends primarily on the channel height and the upstream flow ratio. Such theoretical predictions of droplet size can also be compared with experimentally measured dimensions to support validation and correction.
Moreover, the Poisson model also provides a straightforward framework for estimating measurement uncertainty. Confidence intervals can be calculated based on the binomial distribution of positive droplets or through the propagation of error from λ, enabling ddCRISPR to report quantitative results together with statistically meaningful precision. Moreover, the dynamic range of ddCRISPR is directly related to the number of partitions: large droplet counts support accurate quantification at both low and high concentrations by maintaining reliable estimates of *P_0_
By linking droplet partitioning with Poisson‐based digital counting, ddCRISPR avoids many of the factors that complicate quantitative interpretation in bulk CRISPR fluorescence assays. In bulk reactions, the fluorescence output is a continuous signal that depends on collateral cleavage kinetics, enzyme concentration, reporter availability, and the presence of inhibitory substances in complex samples. These factors can alter the rate and extent of signal accumulation and often require standard curves or carefully controlled reaction times to ensure consistent measurements [5, 45]. Previous studies have shown that Cas12/Cas13 cleavage kinetics [46] vary with buffer composition [47], temperature [48], and sample matrix [49], and that reporter depletion [50] or enzyme variability [16] can lead to nonlinear response curves or early signal plateauing in bulk fluorescence assays [49]. In contrast, ddCRISPR converts each reaction into a binary positive or negative outcome within isolated droplets, making quantification dependent only on the fraction of positive droplets rather than the magnitude or kinetics of fluorescence signals. This digital output improves robustness and accuracy and forms the basis for the workflow steps that follow, including droplet generation, manipulation, and detection.
Based on combining droplet compartmentalization with CRISPR's programmable sequence recognition, ddCRISPR achieves high sensitivity, strong specificity, and improved reproducibility across different sample types. These core principles underpin the performance advantages of ddCRISPR and link the subsequent workflow steps: droplet generation, manipulation, detection, and statistical analysis to the overall accuracy and robustness of the assay.
Workflow of ddCRISPR Assays
2.2
The performance of ddCRISPR depends not only on the core principles described above, but also on the specific workflow used to generate, control, and read out droplets. The overall workflow of ddCRISPR, including sample preparation, droplet partitioning, fluorescence detection, and Poisson‐based quantification, is summarized in Figure 2a. After a sample is digitally partitioned, each droplet must provide a stable and well‐defined reaction environment for reliable CRISPR activation. Therefore, the workflow of ddCRISPR typically includes three key stages: droplet generation, droplet manipulation, and droplet detection. Each stage directly influences important assay parameters such as droplet size uniformity, reagent distribution, reaction timing, background suppression, and signal readout quality. Together, these steps determine the sensitivity, accuracy, and reproducibility of ddCRISPR assays and translate the statistical framework of digital quantification into practical experimental performance.
Droplet Generation
2.2.1
Droplet generation defines how a bulk sample is partitioned into numerous microreactors and thereby establishes the central statistical parameters for ddCRISPR, including droplet size, the total number of partitions, and the average number of molecules per droplet (λ) [51, 52, 53, 54]. These factors form the basis of Poisson statistics and directly determine quantitative performance in applications. Generating stable, highly monodisperse droplets supports consistent λ values and reduces the uncertainty of Poisson estimation [25, 55], enabling sensitive and reproducible absolute quantification in complex biomedical samples [37]. Because different droplet generation strategies create different physical and biochemical microenvironments during compartmentalization, they ultimately influence ddCRISPR performance through their effects on droplet uniformity, shear exposure, and reaction initiation dynamics. These strategies can broadly be categorized into passive and active droplet generation [56], each offering distinct advantages and constraints for ddCRISPR assays. Common methods including passive and active approaches are illustrated in Figure 2b based on the driving mechanism.
Passive Droplet Generation
2.2.1.1
Passive droplet generation methods such as T‐junction [57], flow‐focusing [58], co‐flow [59], and step emulsification [43, 60], which rely on interfacial forces and generally operate under gentle and stable flow conditions [24]. Several widely used microfluidic geometries belong to this category. Flow‐focusing structures [61], where the aqueous stream is pinched by two converging oil streams, generate highly monodisperse droplets at high frequency with excellent size control. T‐junction devices, which rely on periodic necking at the intersection of two channels, provide predictable droplet breakup under controlled flow‐rate ratios. Co‐flow geometries form droplets coaxially and expose biomolecules to minimal shear stress, which is advantageous for encapsulating sensitive enzymes [62]. Step‐emulsification devices, which use a confined‐to‐unconfined expansion to trigger droplet breakup, provide exceptional robustness to flow fluctuations and are easily parallelized for high‐throughput generation [63]. Although step emulsification provides excellent monodispersity and robustness to flow variations, stable operation still depends on a well‐controlled pressure drop along the inlet channel and across the nozzle region. In practice, the confined‐to‐unconfined geometry often requires a two‐step fabrication process, where alignment of shallow nozzles and reliable curing of small features can be technically challenging.
These approaches naturally produce highly monodisperse droplets, reducing volume variation and promoting uniform reaction kinetics across droplets. Consistent droplet volumes facilitate tighter λ control [64] and improved separation between positive and negative droplets in endpoint fluorescence readout [65]. Moreover, the low shear associated with passive formation helps preserve Cas12/Cas13 enzyme activity and reduces nonspecific activation or premature reporter cleavage [66]. An important practical requirement for stable water‐in‐oil droplet formation in microfluidic channels is proper surface treatment. Hydrophobic modification of channel walls (e.g., silanization) is often necessary to prevent aqueous‐phase wetting, ensure clean droplet break‐off, and maintain consistent droplet size across experiments. For double‐emulsion (W/O/W) droplet systems, appropriate surface treatment is also essential, as different regions of the chip typically require hydrophobic or hydrophilic coatings to support sequential emulsification and maintain droplet integrity. These features are particularly advantageous when working with inhibitor‐rich clinical matrices such as plasma, saliva, or nasopharyngeal swabs, as droplet confinement restricts inhibitor diffusion and stabilizes the reaction microenvironment. As a result, passive generation methods often provide robust sensitivity and reproducibility for detecting ultralow‐copy viral RNA or circulating tumor DNA in biomedical settings.
Active Droplet Generation
2.2.1.2
While passive droplet generation relies solely on fluid dynamics, active droplet generation employs external energy inputs [67] such as electric fields, mechanical vibration, thermal modulation, or surface acoustic waves (SAW) [68, 69, 70] to modulate interfacial instabilities and control droplet breakup. Electrical actuation can tune droplet formation in real time and support on‐demand generation [71, 72]; SAW excitation enables contactless droplet production with precise size control [73, 74, 75, 76, 77]; Mechanical vibration accelerates necking dynamics [78], and thermal modulation alters interfacial tension to trigger droplet formation. These active strategies offer advantages that complement passive methods: they provide fine temporal control, adjustable droplet size, and highly regular “clocked” formation, enabling synchronized initiation of CRISPR reactions and compatibility with downstream operations such as merging or picoinjection. Such temporal precision reduces droplet‐to‐droplet variability in reaction start times and improves the uniformity of CRISPR activation kinetics. However, the local electric fields, acoustic energy, or thermal gradients generated during active actuation can influence enzyme stability or reporter integrity if not carefully optimized. When operated within suitable parameter windows, active droplet generation combines high throughput with acceptable biochemical compatibility, making it well suited for ddCRISPR workflows that demand strict timing, staged reagent delivery, or integrated automation in biomedical contexts.
Droplet Manipulation
2.2.2
After droplet formation, manipulation techniques provide additional control over reagent handling, synchronization of reaction initiation, and the execution of multi‐step workflows. Although many ddCRISPR assays operate effectively with droplet generation and detection alone, several advanced implementations benefit from integrating manipulation steps to achieve more precise biochemical control or to accommodate complex assay designs. In these cases, droplet manipulation directly influences CRISPR activation kinetics, background fluorescence levels, and ultimately the sensitivity and reproducibility of the digital readout.
A variety of hydrodynamic and microactuation strategies enable droplets to be mixed, merged, injected, or split with high spatial and temporal precision (Figure 2c). Droplet mixing can be achieved through methods such as induced vortices, chaotic advection, or geometrically assisted stretching, which accelerate homogenization of encapsulated reagents and reduce kinetic variability across droplets. Droplet merging can be normally performed through electrocoalescence, geometric confinement, or hydrodynamic synchronization. And other active approaches such as optofluidic [79] or acoustofluidic actuation may also be used to trigger controlled merging. These light‐ and sound‐based methods provide precise temporal regulation of droplet fusion and are increasingly used in microfluidic systems. These strategies allow Cas enzymes, target nucleic acids, and reporters to encounter each other at defined time points, providing precise temporal control. This is particularly valuable for cleavage‐based ddCRISPR systems, as it helps prevent premature activation and lowers nonspecific background signals that can obscure positive–negative discrimination.
Picoinjection, which delivers nanoliter‐scale reagent volumes into pre‐formed droplets without disrupting their integrity, enables multi‐step ddCRISPR workflows. For example, adding cofactors such as Mg^2^ ^+^ after droplet partitioning or initiating CRISPR reactions only after a stabilization period. By decoupling droplet formation from reaction initiation, picoinjection enhances synchronization across droplets and minimizes reaction heterogeneity that could otherwise compromise quantification accuracy. Droplet splitting may also be used to create parallel reaction pathways or modulate reagent concentrations, though it requires careful control to avoid volume variability.
As with droplet generation, the physical stresses imposed during manipulation such as electric fields during merging, pressure fluctuations during injection, or shear forces during mixing must be carefully tuned to preserve enzyme stability and prevent leakage of reporters or nucleic acids. Excessive perturbation can increase background fluorescence or trigger unwanted CRISPR activation, particularly in droplets containing complex biological matrices. When properly optimized, however, droplet manipulation provides powerful means to fine‐tune reaction timing, improve reproducibility, and support advanced ddCRISPR workflows designed for multiplexed, multi‐step, or high‐throughput biomedical applications.
Droplet Detection
2.2.3
Detection of fluorescence signals from individual droplets is the final step in ddCRISPR workflows and directly determines how positive and negative droplets are distinguished. Accurate and stable readout is essential because CRISPR activation produces a binary signal within each droplet, and the distribution of these signals forms the basis of digital quantification. Different detection strategies influence the signal‐to‐background ratio, throughput, and compatibility with clinical samples (Figure 2d).
Optical fluorescence detection is the most widely used approach [80, 81, 82]. In this method, droplets with emitted fluorescence are recorded by a camera or photodetector [83]. This approach provides simple operation and high compatibility with different droplet sizes. Uniform illumination and stable imaging conditions help reduce variability in signal intensity and improve the separation between positive and negative droplets. When combined with bright reporters and careful control of droplet size, optical detection can support sensitive quantification even when the target concentration is very low. Optical detection is also suited for high‐throughput applications, since large arrays of droplets can be imaged at once.
Laser‐induced fluorescence detection includes methods such as fluorescence‐activated droplet sorting (FADS) and fluorescence‐activated cell sorting (FACS) [84]. In these systems, laser excitation is used to trigger fluorescence emission from droplet contents, which is then detected and quantified. FADS enables rapid screening and sorting of water‐in‐oil droplets based on signal intensity, while FACS allows encapsulated cells or nucleic acid targets in double emulsion [85] or hydrogel droplets to be analyzed in commercially available instruments [86, 87]. These techniques are well‐suited for high‐throughput assays, enabling the identification of positive droplets in large populations [88]. However, laser‐based detection systems require precise calibration and may be less accessible for point‐of‐care settings.
Electrical detection provides an alternative readout mechanism that does not rely on optical components. In these systems, droplet passage changes electrical properties such as impedance or capacitance, which can be monitored to identify positive droplets [89, 90]. Electrical detection platforms are compact, low cost, and compatible with opaque samples [24]. Although they typically have lower sensitivity for very small changes in reporter concentration compared with optical methods, electrical detection is attractive for portable ddCRISPR devices and has potential for integration into fully automated diagnostic platforms [91].
Across these approaches, the stability of droplet fluorescence, the reduction of background signals, and consistent detection thresholds are essential for accurate digital quantification. Optical and laser‐based methods benefit from uniform droplet size and stable reaction conditions that limit signal variability, while electrical methods benefit from precise channel geometry and stable droplet formation. When paired with appropriate detection strategies, ddCRISPR can achieve sensitive and highly reproducible quantification in a wide range of biomedical applications.
Partitioning
2.3
Partitioning is the central principle that transforms CRISPR reactions from bulk signal accumulation into digital counting. While the fundamental concepts of partitioning and Poisson statistics have been outlined in earlier sections, several design considerations remain important when implementing ddCRISPR assays. These considerations ensure that the number and uniformity of droplets generated during the workflow support reliable digital quantification.
The number of partitions largely determines the dynamic range and statistical confidence of ddCRISPR measurements. A higher number of droplets reduces the uncertainty of Poisson estimates and improves the precision of concentration calculations, particularly when the target concentration is low. In practical ddCRISPR assays, the number of droplets generated directly influences the statistical confidence of single‐molecule detection. Digital systems typically produce 10 000 – 20 000 droplets, which is sufficient for routine quantification. For more reliable single‐molecule analysis, generating 20 000–50 000 droplets is generally recommended, as the relative error in λ decreases as the number of partitions increases. Producing more than 100 000 droplets can further reduce statistical uncertainty but is usually unnecessary for standard assays and mainly beneficial when detecting extremely low‐abundance targets [92].
Uniform droplet size is equally important, because variations in droplet volume led to inconsistent values of λ and increase the variance of positive and negative distributions. Maintaining stable partition volumes therefore contributes directly to clear thresholding and accurate endpoint classification.
Partitioning also influences the sensitivity of detection. When droplets are sufficiently small and numerous, the likelihood of isolating single target molecules increases, and each cleavage event produces a strong local fluorescence response. This confinement effect improves the signal‐to‐background ratio and enables the detection of molecules that would otherwise be difficult to quantify in bulk solutions. Furthermore, partitioning reduces the impact of inhibitors present in clinical samples by isolating them within individual droplets rather than allowing them to diffuse throughout the reaction mixture. This effect is important for ddCRISPR assays that target nucleic acids in plasma, saliva, or other complex matrices.
Although partitioning is essential for digital quantification, its benefits depend on stable droplet formation, compatible surfactant systems, and detection strategies that can distinguish positive from negative droplets with minimal variability. When these factors are well aligned, the partitioning process provides the statistical foundation for the high sensitivity, specificity, and reproducibility that differentiate ddCRISPR from bulk CRISPR assays [93].
Signal Detection Strategies
3
ddCRISPR assays generate a binary readout in which each droplet is classified as positive or negative. Therefore, detection in ddCRISPR focuses on accurately identifying fluorescence activation at the droplet level rather than measuring bulk signal accumulation. The key challenges include distinguishing weak positive droplets from background, ensuring consistent illumination, and maintaining clear separation between positive and negative populations. In this section, we summarize amplification‐free and amplification‐assisted ddCRISPR detection strategies with emphasis on their roles in digital signal classification. A concise comparison of amplification‐free and amplification‐assisted ddCRISPR workflows is provided in Table 1 to highlight the analytical and operational characteristics most relevant to digital CRISPR detection.
Amplification‐Free ddCRISPR Assays
3.1
A growing number of researchers focus on the development of amplification‐free ddCRISPR assays, which leverage the intrinsic sensitivity of CRISPR/Cas systems combined with droplet digital techniques to enable direct detection of nucleic acids. While amplification‐assisted strategies, such as LAMP and RPA, have significantly improved the sensitivity and specificity of ddCRISPR assays, they also introduce additional complexity. Moreover, these approaches require more reagents and increase the risk of issues like cross‐contamination.
In practical ddCRISPR systems, droplets typically range from 30 to 60 µm in diameter, corresponding to volumes of approximately 10–120 pL. Even within these experimentally realistic volumes, confining a single nucleic acid molecule yields an effective local concentration in the tens of femtomolar range, which is still several orders of magnitude higher than in bulk assays. More importantly, ddCRISPR does not rely on detecting this concentration directly. Instead, target recognition triggers Cas12 or Cas13 activation, and the enzyme produces extensive collateral cleavage of reporter molecules (Figure 3). A single activated Cas12 or Cas13 enzyme can cleave more than 10^4^ reporters, generating a strong localized fluorescence burst within the droplet. This ultralocalized signal amplification, which has been demonstrated and explained in an amplification‐free Cas13a assay in picoliter droplets [47], creates a clear binary distinction between positive and negative droplets, explaining why ddCRISPR can achieve reliable single‐molecule detection without nucleic acid amplification.
Mechanisms of CRISPR/Cas12 and Cas13‐based nucleic acid amplification‐free detection. Cas12a cleaves dsDNA/ssDNA targets and activates non‐specific ssDNA cleavage, whereas Cas13a cleaves ssRNA targets and activates non‐specific ssRNA cleavage.
Normally, each droplet contains the necessary reagents, including the CRISPR‐Cas system, target nucleic acids, and detection components (e.g., fluorescent reporters). These reactions proceed independently within each droplet, creating isolated environments for highly specific and localized CRISPR activity. The CRISPR reaction is illustrated in Figure 3. If a target nucleic acid is present in the droplet, the crRNA guides the Cas enzyme to bind it. This binding activates the Cas enzyme, initiating the reaction. If no target is present, the Cas enzyme remains inactive. This ensures only correct targets trigger the reaction, which reduces false signals and increases detection reliability. Once activated, the Cas enzyme cleaves the target nucleic acid. Certain Cas enzymes, like Cas12a and Cas13a, also cleave nearby reporter molecules, which typically are DNA/RNA labelled with quenched fluorophore. Cleavage of the reporter releases a fluorescence signal: droplets containing target nucleic acid are lit up, while droplets without the target remain dark. The intensity of fluorescence in each droplet is linked to the amount of nucleic acid present. And positive droplets can be counted to determine the target concentration in the original sample.
Amplification‐free ddCRISPR assays offer several advantages. They make the steps simpler, need fewer reagents, work faster, and reduce the chance of contamination. These assays are particularly useful when quick and sensitive detection is needed, like pathogen screening, water quality testing or disease diagnostics. Also, since each droplet is isolated, multiplexed detection in a single sample is possible by using different CRISPR enzymes. Despite these advantages, there are still some challenges. The fluorescence output in amplification‐free ddCRISPR is often modest because only a small number of Cas enzymes are present in each droplet. High‐performance Cas enzymes, improved reporter systems, and more sensitive signal readout methods are required. Other factors like droplet stability, enzyme kinetics, and background noise can also influence assay performance. Future developments in droplet microfluidic chips, enhanced Cas enzymes, and advanced detection strategies may further improve amplification‐free ddCRISPR assays in real‐world testing [94, 95]. To provide a concise overview of the detection formats used in ddCRISPR systems, we introduce Table 1 as a comparative summary of amplification‐free and amplification‐assisted strategies. The table compiles the key biochemical and operational parameters including enzyme requirements, primer design, reaction conditions, amplification time, amplicon characteristics, and representative LOD.
Isothermal Amplification Methods for ddCRISPR
3.2
To further enhance detection signals, many studies have incorporated various isothermal amplification techniques into ddCRISPR workflows [5, 11]. Isothermal amplification refers to methods that exponentially amplify nucleic acids at a constant temperature, eliminating the need for thermal cycling. In these assays, amplification occurs within individual, isolated droplets, each serving as a microreactor containing the necessary reagents and, ideally, a single target nucleic acid molecule. Amplification‐assisted ddCRISPR generally provides stronger fluorescence signals and clearer positive/negative separation, especially for ultra‐low abundance targets such as viral RNA or rare DNA variants [96, 97].
Loop‐Mediated Isothermal Amplification (LAMP)
3.2.1
LAMP amplifies nucleic acids at a constant temperature of 60–65°C using four to six primers that recognize six to eight distinct regions of the target sequence [34, 98]. A strand‐displacing DNA polymerase initiates synthesis without thermal cycling, forming looped amplicons that undergo exponential self‐primed amplification (Figure 4a). First, the inner primers (FIP and BIP) bind to the target and initiate strand‐displacement synthesis, forming a short double‐stranded region. This newly generated strand folds back to create a dumbbell‐shaped structure, which serves as the starting point for amplification. The dumbbell then converts into a stem–loop structure that enables continuous self‐priming. Loop primers bind to the loop regions and introduce additional initiation sites, accelerating the reaction. Through repeated cycles of extension and strand displacement, LAMP rapidly produces large, multi‐loop “cauliflower‐like” DNA products
Schematic illustration of three isothermal amplification methods in ddCRISPR. (a) LAMP uses inner and loop primers to initiate dumbbell‐shaped DNA structures, followed by strand displacement and cyclic amplification. (b) RPA relies on recombinase–primer complexes to form D‐loops, enabling strand invasion, DNA synthesis, and duplex formation at a constant temperature. (c) RCA involves padlock probe hybridization, ligation to form a circular template, and continuous DNA synthesis by strand‐displacing polymerase. (d) RAA employs recombinase and single‐stranded DNA binding proteins to assist primer binding and initiate rapid isothermal amplification. (e) HCA proceeds through sequential primer binding, extension, and repeated strand displacement to generate highly branched DNA structures.
This process produces complex, cauliflower‐like DNA structures within 15–60 min, facilitating rapid and sensitive nucleic acid detection [99]. When integrated with ddCRISPR platforms, LAMP reactions are partitioned into thousands of discrete droplets, where the amplified products initiate Cas12/Cas13 activation. This often results in high sensitivity and fast detection.
However, LAMP‐ddCRISPR systems face specific challenges. The intrinsic complexity of primer design in LAMP increases the risk of non‐specific amplification, and despite droplet partitioning, occasional background activation may occur. Moreover, incubation at elevated temperatures can lead to droplet evaporation, fusion, and instability, potentially compromising quantification accuracy. Strategies such as fluorinated oil encapsulation, surface treatments to minimize evaporation, and microfluidic chip optimization have been developed to mitigate these issues [100]. Overall, LAMP‐ddCRISPR combines rapid amplification and high specificity but requires careful system design to maintain robustness in low‐concentration detection scenarios.
Recombinase Polymerase Amplification (RPA)
3.2.2
Recombinase Polymerase Amplification (RPA) is another widely used isothermal method that is well‐suited for integration with ddCRISPR assays [32, 101]. The reaction employs three enzymes: a recombinase that facilitates primer‐template pairing, a single‐stranded DNA‐binding protein (SSB) that stabilizes the displaced strand, and a strand‐displacing DNA polymerase that extends the primers without denaturation (Figure 4b) [31, 102]. This allows target amplification to be completed within 5–20 min under simple, field‐deployable conditions.
RPA operates at lower temperatures (37–42°C) and unlike LAMP, does not require complex primer designs. The recombinase enzyme facilitates primer binding to the target DNA, followed by strand displacement and polymerization mediated by the strand‐displacing DNA polymerase. Its low operating temperature improves droplet stability and limits evaporation, making it an attractive choice for ddCRISPR platforms, such as the SHERLOCK platform [16].
Integration of RPA with ddCRISPR platforms leverages the rapid kinetics and low‐temperature operation of RPA for sensitive, absolute quantification of nucleic acids [103, 104]. Droplet compartmentalization minimizes template competition and non‐specific amplification, while CRISPR effectors provide sequence‐specific detection through collateral cleavage of reporter probes. Importantly, RPA does not require hot‐start amplification, unlike PCR or LAMP; the reaction can be activated at room temperature by adding MgOAc. This property allows the reaction initiation to depend on the presence of Mg ions [105, 106, 107, 108], which can also serve as a trigger in two‐step ddCRISPR‐RPA schemes where target amplification [109, 110] and Cas‐mediated collateral cleavage are temporally separated to reduce false‐positive quantification errors [111]. In addition, two‐step RPA–ddCRISPR workflows often require controlled mixing or droplet merging to initiate the Cas reaction at a defined time point. These additional droplet manipulation steps allow separation of the amplification stage from CRISPR activation, reducing premature cleavage and improving quantification accuracy, but they also introduce extra technical requirements for stable droplet handling.
Rolling Circle Amplification (RCA)
3.2.3
Rolling Circle Amplification (RCA) is a linear amplification technique that uses a short DNA or RNA primer to initiate replication on a circular template [112]. A strand‐displacing DNA polymerase extends the primer continuously, generating long single‐stranded DNA (ssDNA) concatemers that consist of multiple repeats of the template sequence [113, 114]. RCA operates at a constant temperature, typically 30–37°C and can be performed under mild conditions without thermal cycling or complex primer design (Figure 4c). When integrated with ddCRISPR systems, RCA enables isothermal signal amplification with high sequence fidelity [115]. Target nucleic acids—especially miRNAs or short DNA fragments—can be first ligated into circular templates via padlock probes [116], which then serve as substrates for RCA inside droplets. The compartmentalized environment ensures that each RCA event occurs independently, allowing absolute quantification and reducing background interference.
In ddCRISPR, a key advantage of RCA is that each concatemer contains many repeated target motifs. These multiple recognition sites greatly increase the likelihood of Cas12/Cas13 activation within a droplet, producing stronger fluorescence signals and clearer separation between positive and negative droplets. Droplet partitioning also helps confine nonspecific RCA events, reducing background compared with bulk RCA. In particular, the droplet structure prevents amplified concatemers or partially ligated templates from diffusing throughout the reaction mixture, avoiding the diffusion‐driven background signals commonly observed in bulk RCA assays.
The long concatemeric RCA products can activate CRISPR/Cas12a or Cas13a through direct hybridization or after digestion into shorter trigger fragments. However, RCA proceeds more slowly than LAMP or RPA, increasing exposure to droplet evaporation or fusion during incubation, and the long ssDNA products can fold or tangle, limiting diffusion and generating heterogeneous fluorescence intensities among positive droplets, which complicates digital thresholding. Despite these challenges, RCA‐ddCRISPR remains promising for applications requiring high fidelity and strong signal amplification, and continued improvements in padlock probe chemistry, droplet stability and Cas enzyme activation are expected to further support its integration into ddCRISPR workflows.
Other Methods
3.2.4
In addition to commonly used isothermal amplification methods such as LAMP, RPA, and RCA, which are widely adopted in ddCRISPR assays, recent studies have explored the integration of alternative isothermal strategies into droplet digital formats. These methods vary in kinetics, enzyme requirements, and operating conditions, offering different opportunities and challenges for digital CRISPR detection.
For example, RAA (Recombinase‐Aided Amplification) shares the same core principle as RPA, relying on recombinase‐primer complexes and strand‐displacing polymerases for isothermal amplification. However, its practical implementation differs in several key aspects. RAA is predominantly developed and commercialized by Chinese manufacturers, offering a more accessible and cost‐effective alternative to RPA [117], whose core reagents are proprietary to TwistDx. When combined with ddCRISPR, RAA benefits from droplet partitioning, which reduces nonspecific amplification and contains spurious reactions within individual droplets rather than allowing them to propagate across the entire sample. This isolation helps mitigate false positives that might occur in bulk reactions due to primer dimerization or unintended primer–template interactions [118, 119].
Hyperbranched RCA amplification (HCA) is another isothermal nucleic acid amplification strategy that generates highly branched, tree‐like DNA structures through repeated cycles of primer annealing, extension, and strand displacement. In this mechanism, an initial primer binds to a target region and initiates DNA synthesis [120]. The newly synthesized strand then serves as a template for additional primers, enabling further rounds of extension and strand displacement as shown in Figure 4e. As the reaction progresses, multiple new priming sites are exposed on each synthesized strand, resulting in exponential signal amplification through continuous generation of branching DNA products [121, 122, 123]. In ddCRISPR systems, these branched products provide multiple reusable Cas recognition motifs, often producing strong fluorescence in positive droplets and enhancing digital separation. At the same time, their rapid growth and structural complexity can cause variability in signal intensity across droplets or restrict diffusion, complicating thresholding and potentially reducing quantification precision.
Although these alternative isothermal amplification methods can enhance ddCRISPR sensitivity, their successful integration depends on maintaining stable droplet environments and minimizing nonspecific amplification, both of which strongly influence digital readout quality. As with other amplification‐assisted strategies, improvements in droplet stability, reagent delivery, and Cas activation control will be necessary to fully exploit these chemistries in digital nucleic acid detection.
Nucleic Acid Biomarker Detection Using ddCRISPR
4
The integration of CRISPR technology with droplet digital assays has expanded the capabilities of nucleic acid detection, enabling high sensitivity, specificity, and precision in diverse applications, such as disease diagnostics, environmental monitoring, and food safety [124]. In this section, we review the published ddCRISPR studies applied to nucleic acid biomarker detection. For clarity, we categorize existing work according to the type of target molecule—DNA and RNA biomarkers—and summarize the key assay parameters and applications. A consolidated overview of these studies is provided in Table 2 at the end of this section.
DNA Biomarker Detection
4.1
Human Papillomavirus (HPV) DNA
4.1.1
HPV infection is a leading cause of cervical and other cancers, making early and accurate DNA detection crucial for disease prevention and management [125].
HPV infection is a leading cause of cervical and other cancers, making early and accurate DNA detection crucial for disease prevention and management. Therefore, early and accurate detection of HPV DNA is critical for timely prevention, screening, and clinical management [125]. Recent advancements in ddCRISPR assays for HPV detection have demonstrated ultra‐sensitive viral DNA quantification, even at attomolar levels, with minimal false positives [126, 128, 129, 130, 131]. Isothermal amplification techniques, such as RPA, have been used to improve detection efficiency while maintaining specificity.
For example, the Picoinjection Aided Digital reaction unlocking (PADLOCK)‐CRISPR system developed by Cui et al. [126] achieved absolute quantification of HPV16 DNA (Figure 5a). In this system, a droplet generator coupled with a microfluidic picoinjector introduced magnesium acetate (MgOAc) into preformed RPA droplets, precisely controlling the reaction onset and effectively preventing premature amplification. The combination of RPA with CRISPR/Cas13a yielded single‐molecule detection sensitivity, uniform fluorescence reporting, and a high signal‐to‐noise ratio (>6). The entire assay was completed within 30 min at 37 °C, and clinical validation with 22 patient swab samples showed 100% concordance with qPCR, underscoring the robustness and translational potential of the PADLOCK‐CRISPR platform for rapid HPV DNA quantification.
ddCRISPR platforms for HPV DNA detection. (a) Schematic illustration of the Picoinjection Aided Digital reaction unLOCKing (PADLOCK) CRISPR platform, which integrates picoinjection‐triggered droplet digital RPA with Cas13a detection for absolute quantification of HPV16 DNA. Reproduced with permission [126]. Copyright 2022, Elsevier; (b) Workflow of the d3CRISPR assay for amplification‐free detection and absolute quantification of HPV18 DNA, using double‐emulsion droplets and flow cytometry readout. Reproduced under terms of the CC‐BY 4.0 license [127]. Copyright 2025, the Author(s), published by Elsevier.
Similarly, several recent studies have integrated RPA with ddCRISPR for HPV detection. Xu et al. [129] reported a microfluidic‐free digital RPA–Cas12a assay using vortex‐generated polydisperse droplets, achieving detection of HPV18 DNA at15 aM and SARS‐CoV‐2 RNA at 10 fM with 100% sensitivity and specificity. Zhao et al. [130] designed a dual‐droplet microfluidic platform (M‐D3) for simultaneous detection of HPV16 and HPV18, integrating multiplexed RPA and CRISPR‐Cas12a in parallel droplets with green/red fluorescence coding. The platform demonstrated single‐copy sensitivity (∼10^−^ ^1^ ^8^ m) and accurately identified HPV subtypes in 20 clinical samples within 30 min. Together, these RPA‐assisted ddCRISPR systems highlight the potential of combining isothermal amplification with droplet compartmentalization and CRISPR collateral cleavage for rapid, quantitative, and clinically relevant viral diagnostics [132].
In addition, Li et al. [133] developed an amplification‐free platform for point‐of‐care detection of HPV18 in cervical epithelial cells, achieving a limit of detection (LOD) down to 10 copies/µL. This work represents a versatile, contamination‐free detection platform with potential to enhance integrated public health surveillance efforts. In our previous work [127], we developed an amplification‐free droplet digital CRISPR/Cas12a (d^3^CRISPR) assay that integrates double‐emulsion (DE) droplet compartmentalization with flow cytometry for absolute nucleic acid quantification (as shown in Figure 5b). By eliminating the need for pre‐amplification, this system simplifies the assay workflow while maintaining attomolar‐level sensitivity, with a LOD of 100 aM for HPV18 DNA.
Other Viral DNA
4.1.2
JC virus (JCVs) is a circular double‐stranded DNA (dsDNA) virus belonging to the Polyomaviruses (PyVs) family [134]. It is highly prevalent, with 80–100% of adults carrying latent infections. However, in immunocompromised individuals, such as renal transplant patients, JCV can reactivate and cause polyomavirus‐associated nephropathy (PVAN), a serious complication that may lead to kidney transplant failure. Early and quantitative detection of JCV in clinical samples is crucial for monitoring infection progression, guiding treatment decisions, and predicting patient prognosis. Liu et al. [132] developed a microfluidic digital CRISPR/Cas13a‐RPA platform (MdCaR) for JCVs detection (Figure 6a), focusing on JCV DNA detection in urine samples from renal transplant patients. The MdCaR system demonstrated superior sensitivity and accuracy compared to qPCR, achieving a LOD of 1 copy/mL (100 folder higher sensitivity than qPCR) within 40 min. Notably, one discrepant sample was negative in qPCR but positive in MdCaR and subsequent next‐generation sequencing (NGS) confirmed the presence of JCV, highlighting MdCaR's higher sensitivity.
ddCRISPR platforms for other viral DNA biomarker detection. (a) Schematic illustration of JCV DNA detection platform combing CRISPR/Cas13a with RPA. Reproduced with permission [132]. Copyright 2024, Elsevier. (b) Workflow of contamination‐free digital CRISPR platform for Point‐of‐Care detection of viral HPV18 DNA and influenza A RNA. Reproduced with permission [133]. Copyright 2024, American Chemical Society; (c) Schematic illustration of African swine fever virus (ASFV) detection in suspected swine serum samples by dual‐crRNA targeting droplet Cas12a assay. Reproduced with permission [135]. Copyright 2021, American Chemical Society.
Li et al. reported a contamination‐free nucleic acid detection platform that combines digital microfluidic nucleic acid extraction with an amplification‐free CRISPR assay for POC detection of viral DNA/RNA in clinical samples [133]. The platform achieved LODs of 10 copies/µL for DNA and 5 copies/µL for RNA, with a compact, sample‐to‐result workflow of ∼75 min (Figure 6b). In small clinical sets, this platform achieved around 100% agreement with qPCR for influenza A detection in nasopharyngeal swabs (n = 15). These results highlight the potential of amplification‐free ddCRISPR platforms for rapid, quantitative, and contamination‐free viral diagnostics at the point of care.
A rapid amplification‐free droplet Cas12a assay, developed by Yue et al. [135], demonstrated excellent performance in quantifying African swine fever virus (ASFV), Epstein−Barr virus (EBV), and Hepatitis B virus (HBV) from clinical serum samples without the need of nucleic acid amplification (Figure 6b). Before detection, the researchers comprehensively investigated the effect of reaction conditions, such as reporter sequence, temperature dependence, the components and concentration of the reaction buffer on Cas12a reaction efficiency. The optimal conditions increased the sensitivity over 50‐fold compared to previously reported Cas12a assays [5, 136]. Furthermore, the assay exhibited high specificity by detecting genomic DNA extracted from four pathogenic bacteria—Pseudorabies virus (PRV), Porcine circovirus (PCV), Haemophilus parasuis (HPS), and Streptococcus suis (SS)—all associated with swine respiratory diseases. To validate its applicability, 18 unamplified ASFV‐suspected swine serum samples were tested. Positive samples, confirmed by qPCR, were correctly diagnosed, with target viral DNA concentrations ranging from 80 to 5200 copies/µL.
Li et al. reported a contamination‐free nucleic acid detection platform that combines digital microfluidic nucleic acid extraction with an amplification‐free CRISPR assay for POC detection of viral DNA/RNA in clinical samples. The platform achieved LODs of 10 copies/µL for DNA and 5 copies/µL for RNA, with a compact, sample‐to‐result workflow of ∼75 min (Figure 6c). In small clinical sets, this platform achieved 100% agreement with qPCR for influenza A detection in nasopharyngeal swabs (n = 15). These results highlight the potential of amplification‐free ddCRISPR platforms for rapid, quantitative, and contamination‐free viral diagnostics at the point of care.
Pathogenic Bacterial DNA
4.1.3
Detecting pathogenic bacterial DNA is vital for preventing and controlling infectious diseases. ddCRISPR enables rapid and accurate identification, even at low concentrations and in complex samples, offering higher sensitivity than traditional culture‐based methods. Early and precise detection helps prevent outbreaks, ensures food and water safety, and supports the monitoring of bacterial species, strains, and antibiotic resistance.
Salmonella enterica serovar Typhimurium is a leading cause of bacterial foodborne illness worldwide, frequently associated with contaminated meat, dairy, and produce. Rapid and quantitative detection of Salmonella is essential for outbreak prevention and food safety monitoring [137]. Among its genetic biomarkers, the invA gene encodes a conserved inner membrane protein involved in type III secretion and host invasion. It is widely recognized as a gold standard target for Salmonella detection due to its high specificity and diagnostic relevance [138].
Wu et al. [139] developed DropCRISPR, a droplet digital LAMP–CRISPR/Cas12a platform for the absolute quantification of Salmonella invA gene. The invA gene encodes a conserved protein essential for host invasion and is recognized as the gold standard biomarker for rapid and quantitative identification of Salmonella enterica in food safety monitoring [138]. The system leverages a two‐step microfluidic workflow: LAMP pre‐amplification is performed in picoliter droplets, followed by picoinjection of Cas12a reagents to initiate collateral cleavage‐based fluorescence detection (Figure 7a). This spatial and temporal separation prevents reagent incompatibility and nonspecific background signals, ensuring high reaction efficiency. DropCRISPR achieved a limit of detection of 3 fM for target DNA and as low as 10^2^ CFU/mL for live bacteria, with precise quantification across a wide dynamic range. Notably, the system performed robustly in complex matrices such as LB medium and raw milk, without requiring nucleic acid extraction.
ddCRISPR platforms for pathogenic bacterial DNA detection. (a) Workflow of the hybrid LAMP‐CRISPR/Cas12a DropCRISPR platform. Reproduced with permission [139]. Copyright 2022, Elsevier; (b) Schematic of encoded droplet generation and heat map displaying the determined DNA concentration in spiked lettuce by digital one‐pot CRISPR/Cas13a and qPCR assays. Reproduced with permission [140]. Copyright 2023, Elsevier.
Similarly, Lin's group introduced a one‐pot CRISPR/Cas13a‐based droplet microfluidic system capable of multiplex detection of Escherichia coli, Salmonella enterica, and Listeria monocytogenes in a single reaction [140]. Target‐specific CRISPR/Cas13a reactions are encapsulated into individual droplets, each labelled with a distinct fluorescent reporter, enabling simultaneous identification of multiple pathogens without cross‐reactivity (Figure 7b). The assay achieved detection limits as low as 10^2^ CFU/mL for all three pathogens in spiked milk samples, with a total assay time under 90 min. This multiplexing capability greatly enhances the practicality of ddCRISPR platforms for comprehensive food safety surveillance.
It was recently reported that target DNA from the highly pathogenic Shiga‐toxin‐producing Escherichia coli (STEC) O157:H7 was detected and quantified using the d^3^CRISPR platform [127]. The integration of ddCRISPR/Cas12a‐based assays with flow cytometry enabled absolute quantification of E. coli O157:H7 DNA across a wide dynamic range, with single‐molecule sensitivity and limits of detection from 60 to 6.0 × 10^5^ copies/µL in various sample matrices.
Overall, these studies highlight the versatility of ddCRISPR platforms in rapid, sensitive, and quantitative detection of diverse pathogenic bacteria. Their capability to function in complex sample matrices, enable multiplexed detection and integrate with high‐throughput readout technologies such as flow cytometry positions ddCRISPR as a powerful tool for food safety surveillance and infectious disease control.
RNA Biomarker Detection
4.2
SARS‐CoV‐2 RNA
4.2.1
Since the emergence of severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) in late 2019, the development of rapid, sensitive, and reliable diagnostic tools have been a primary focus of global health efforts. SARS‐CoV‐2, the causative agent of coronavirus disease 2019 (COVID‐19), has led to unprecedented public health challenges, highlighting the need for scalable molecular diagnostics capable of detecting viral RNA in diverse sample types, including nasopharyngeal swabs, saliva, and wastewater. Early detection and accurate quantification of viral RNA are critical for controlling disease transmission, monitoring infection dynamics, and assessing treatment efficacy [141].
In this regard, many recent studies have demonstrated the effectiveness of droplet digital CRISPR assays in the detection of SARS‐CoV‐2 [47, 129, 142, 143, 144, 145, 146, 147, 148, 149]. For example, Luo et al. [150] developed a droplet digital reverse transcription loop‐mediated isothermal amplification (RT‐LAMP) enhanced Cas12b‐based RNA detection platform (ddRECD) (Figure 8a). This method integrates RT‐LAMP with a Cas12b‐mediated detection system in a droplet digital format. Cas12b, known for its superior thermal stability, provides a homogeneous reaction environment compatible with isothermal conditions, minimizing premature amplification and improving specificity. The platform achieved single‐molecule sensitivity and absolute quantification over a wide dynamic range (1–10^4^ copies/µL), with a limit of detection as low as 14 copies/mL. Notably, ddRECD resolved concentration differences as small as 1.5‐fold in RNA extracted from infected cells, surpassing conventional bulk assays.
ddCRISPR platforms for SARS‐CoV‐2 RNA detection. (a) Schematic of a RT‐LAMP–Cas12b assay in droplets, enabling absolute quantification of SARS‐CoV‐2 RNA via thermophilic Cas12b cleavage. Reproduced with permission [150]. Copyright 2021, Elsevier; (b) Vortex‐generated ddCRISPR assay with one‐pot RPA, enabling digital readout without microfluidic chips. Reproduced with permission [129]. Copyright 2025, Elsevier; (c) Ultralocalized Cas13a system for amplification‐free single‐molecule detection in picoliter droplet arrays, achieving high turnover and signal confinement. Reproduced with permission [47]. Copyright 2020, American Chemical Society.
Another similar work was reported by Wang's group [129]. A RAA/RT‐RAA assisted CRISPR/Cas12a system was reported (Figure 8b), operating at a lower temperature (37–42°C). The platform successfully detected SARS‐CoV‐2 infection using nasopharyngeal swabs from 10 clinical samples. Five positive cases previously confirmed by qRT‐PCR were successfully identified, achieving a LOD down to10 fM (equal to around 6,022 copies/µL).
Besides, droplet digital assays for CRISPR/Cas13a‐based detection of SARS‐CoV‐2 have been implemented without any target amplification [144, 145, 146, 147, 151]. This amplification‐free format avoids false‐positive droplets commonly observed in isothermal amplification systems and eliminates the need for complex primer design, reducing both assay cost and turnaround time. For example, Tian et al. [47] introduced an ultralocalized Cas13a platform capable of single‐molecule RNA detection without preamplification (Figure 8c). In this approach, Cas13a and reporter RNAs are densely immobilized in femtoliter‐sized hydrogel reaction chambers, increasing local target and enzyme concentrations. This spatial confinement enhances the signal‐to‐noise ratio and eliminates the need for amplification. The system successfully detected SARS‐CoV‐2 RNA from clinical nasopharyngeal swab RNA extracts down to the single‐molecule level within 15 min, demonstrating sensitivity comparable to amplification‐based assays while avoiding the associated biases and complexity.
Other Viral RNA
4.2.2
There are also several works [133, 152, 153] about viral RNA detection using ddCRISPR assays. For example, Zuo et al. [153] focuses on Severe Fever with Thrombocytopenia Syndrome Virus (SFTSV), a tick‐borne virus with outbreaks in China, Japan, and Korea. They developed a digital RPA‐ assisted CRISPR‐Cas13a assay, which directly reverse transcribes SFTSV RNA into cDNA as a template, integrating RPA, T7 RNA polymerase transcription and CRISPR/Cas13a detection system into one‐step workflow (Figure 9a). To optimize the reaction system, they systematically investigated the effects of various components, such as reaction buffer, pH, enzymes, and other additives. The optimized one‐pot Cas13a‐based assay achieved a LOD of 5 copies/µL. Using this system to quantify viral RNA in simulated SFTSV samples produced results consistent with those obtained from qPCR‐based cDNA measurements.
Amplification‐free and one‐pot ddCRISPR strategies for other RNA biomarker detection. (a) One‐pot digital RPA–Cas13a platform for SFTSV RNA quantification, integrating transcription and cleavage within droplets. Reproduced with permission [153]. Copyright 2023, Elsevier; (b) Polydisperse droplet‐based Cas13a assay (PddCas13a) for single‐molecule detection of circRNAs and miRNAs. Reproduced under the terms of the CC‐BY license [156]. Copyright 2024, the Author(s), published by Elsevier; (c) Schematic illustration of the ultralocalized Cas13a assay for amplification‐free miRNA‐17 detection. Right: Comparative analysis of miR‐17 expression across different cell lines using qRT‐PCR, ddPCR, and the ultralocalized Cas13a assay. Reproduced with permission [47]. Copyright 2021, American Chemical Society.
MicroRNA
4.2.3
MicroRNAs (miRNAs) play crucial roles in gene regulation and are increasingly recognized as biomarkers for disease diagnosis and prognosis [154]. These non‐coding RNAs (ncRNAs) are particularly important in cancer detection, as their altered expression levels are associated with tumorigenesis and disease progression [155]. Traditional detection methods, such as quantitative reverse transcription PCR (qRT‐PCR) and RNA sequencing (RNA‐seq), often require amplification, complex sample preparation, and specialized equipment, limiting their suitability for point‐of‐care (POC) applications. Recent advances in CRISPR/Cas13a‐based droplet digital assays provide amplification‐free, highly specific, and quantitative detection of microRNAs, improving both accuracy and clinical applicability.
The polydisperse droplet digital Cas13a (PddCas13a) assay [156] enables absolute detection of miR‐21 and circHIPK3, both relevant biomarkers for colorectal cancer (CRC) diagnostics (Figure 9b). This technique combines chemically modified crRNAs with Cas13a‐mediated collateral cleavage in polydisperse droplets generated by vortex emulsification. The assay was validated using serum‐derived exosomes, cell lysates, and clinical CRC patient samples, achieving limits of detection of 10 aM for miR‐21 and 1 pM for circHIPK3. Compared to qRT‐PCR (LOD ∼250 fM), the PddCas13a assay demonstrated 100‐fold greater sensitivity, enabling precise single‐molecule detection. Additionally, a portable automated device was developed, integrating droplet generation, incubation, and fluorescence imaging, facilitating real‐time cancer diagnostics via liquid biopsy.
Besides, the ultralocalized Cas13a assay developed by Zhou'group (Figure 9c) enables amplification‐free, single‐molecule detection of microRNAs. They directly quantify miRNA‐17 expression levels in four different cell lines, including human breast normal cells (Hs578Bst), adenocarcinoma cells (MCF‐7 and MDA‐MB‐231), and human glioma cells (U87). The platform was further applied to distinguish serum miR‐21 levels between breast cancer patients and healthy individuals. While minor discrepancies were noted compared with gold‐standard qRT‐PCR, the assay successfully achieved robust amplification‐free digital quantification at the single‐molecule level (Table 2).
Challenges and Future Perspectives
5
Although ddCRISPR platforms offer excellent sensitivity and quantification capability, most current systems still require multiple manual steps, including sample preparation, droplet generation, amplification, and signal readout. These steps increase processing time, raise the risk of contamination, and limit the use of technology in field testing or resource‐limited settings. Achieving fully automated systems will require improved integration of microfluidic chips, fluid control, and detection modules, while maintaining compact, affordable, and easy to use designs.
To help address these equipment‐related challenges, several instrument‐free or low‐instrument approaches have been reported for generating digital droplet assays. These methods aim to reduce the dependence on pump‐driven microfluidics and complex pressure controllers while still providing large numbers of isolated reaction volumes. For example, gravity‐driven [159, 160] or vacuum‐driven devices [161] allow samples to self‐partition into microdroplets or confined reaction sites through passive flow, enabling digital compartmentalization without external pumps. Capillary‐based emulsification [162] can form water‐in‐oil droplets through spontaneous breakup driven by interfacial forces, while centrifugal microfluidics [163] can generate uniform droplets using rotational forces alone. Vortex‐based emulsification [164, 165] also provides a simple benchtop option for rapidly producing picoliter‐scale droplets. In the context of ddCRISPR, these strategies represent promising directions for simplifying droplet generation and improving the accessibility of digital assays in point‐of‐care or resource‐limited settings.
Besides, the performance of droplet‐based systems depends strongly on droplet stability and uniformity, which are typically maintained using surfactants that may interfere with downstream reactions, leading to reduced efficiency or increased background signal. Producing high‐quality droplets also requires precise control of flow rates and channel geometry, which can vary across devices or batches. Complex samples such as milk or blood may contain inhibitors that affect droplet stability and detection accuracy. Although CRISPR‐based detection offers excellent specificity, assays that rely on isothermal amplification still require carefully designed primers, and the CRISPR/Cas system itself is constrained by PAM sequence requirements. In addition, crRNA selection and reaction optimization must be tailored for each target, which limits the flexibility of multiplex detection. The activity of Cas proteins can also be affected by buffer composition, temperature, and sample matrix, restricting portability across platforms. For very low target concentrations, amplification steps are still needed, which adds to assay complexity and time.
While ddCRISPR platforms show immense potential for precision diagnostics of nucleic acid biomarkers, their future success depends on advances in key aspects, such as AI‐driven analysis, point‐of‐care testing, and multiplexed detection. Continued interdisciplinary efforts will be essential to move ddCRISPR from lab prototypes toward scalable, reliable, and widely accessible diagnostic solutions.
Artificial Intelligence (AI)‐Driven Diagnostics
5.1
AI is expected to become a key aspect of next‐generation digital assays. The integration of AI algorithms into droplet digital assays can improve data analysis, increasing the accuracy and efficiency of nucleic acid detection [166, 167]. Recent studies have shown that convolutional neural networks and other deep learning algorithms can significantly improve droplet segmentation and classification accuracy, even in noisy or complex imaging environments [168, 169, 170]. More importantly, AI not only enhances endpoint image analysis, but can also guide droplet generation in real time [171]. AI‐assisted control systems can adjust flow rates or voltages to stabilize droplet size and shape based on feedback from imaging or flow sensors [172]. Such feedback‐guided optimization minimizes assay‐to‐assay variation, which is especially critical in clinical applications. In addition, machine learning algorithms can simplify the entire data analysis pipeline by learning decision boundaries from large datasets, effectively distinguishing true positives from background or leaky signals, thereby improving assay precision and robustness [173].
Point‐of‐Care Testing (POCT)
5.2
ddCRISPR platforms also hold great potential for point‐of‐care applications. The integration of microfluidic droplet systems with temperature control, reagent handling, and detection modules can enable fully self‐contained ddCRISPR platforms for on‐site testing. Recent progress in centrifugal microfluidics [174, 175] and active droplet generation systems [176] shows promise for simplifying workflows in compact, portable formats [177]. Combined with low‐temperature amplification methods such as RPA, these systems can operate with minimal instrumentation. Moreover, smartphone‐based fluorescence detection has been developed for ddPCR and CRISPR assays, providing cost‐effective and accessible signal readout tools. These advances enable rapid nucleic acid quantification in outpatient clinics, rural hospitals, or field conditions, supporting timely disease screening, pathogen surveillance, and personalized therapy monitoring. From a broader perspective, the push toward POCT applications is reshaping microfluidic system design. Trends include the use of low‐cost and multifunctional materials (e.g., thermoplastics [178], superhydrophobic materials [179, 180], and liquid metal [181, 182]), modular chip architectures, and battery‐operated components, all aiming to balance performance with field usability.
Multiplexed Detection
5.3
Future droplet digital assays are expected to incorporate enhanced multiplexed detection capabilities, allowing simultaneous detection of multiple targets within a single assay [183, 184]. Multiplexing not only increases diagnostic throughput but also enables applications such as mutation profiling, pathogen differentiation, and biomarker panel screening. In practice, spectral overlap and cross‐reactivity often limit multiplexing in droplet digital systems. Recent strategies to overcome these challenges include amplitude‐based multiplexing [185], where varying probe concentrations create distinguishable signal clusters within the same fluorescence channel [140, 158, 186, 187], and barcode‐based approaches [173, 188, 189, 190], where different droplet populations carry unique crRNA–reporter pairs or sequence‐tagged amplification products [187, 189]. To support reliable multiplexing, upstream steps such as target enrichment, hybridization capture, or digital pre‐amplification may also be integrated into the droplet workflow, improving both sensitivity and specificity for low‐abundance analytes.
Conclusions
6
This review summarizes recent advances in droplet digital CRISPR (ddCRISPR) technology, including droplet generation, manipulation and detection methods, as well as signal amplification strategies, and representative applications in biomarker analysis. ddCRISPR offers distinct advantages over conventional molecular diagnostics, particularly in enabling absolute quantification, high specificity and sensitivity, and robust performance in various sample matrices. These capabilities position ddCRISPR as a potential tool for applications ranging from clinical diagnostics and food safety monitoring to environmental surveillance. While the technology shows great promise, challenges remain in automation, droplet stability, multiplexing, and integration with portable platforms. Future efforts should focus on addressing these challenges and exploring advances to enable robust, field‐deployable diagnostics.
Conflicts of Interest
The authors declare no conflict of interest.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1T. Nolan , R. E. Hands , and S. A. Bustin , “Quantification of m RNA Using Real‐Time RT‐PCR,” Nature Protocols 1 (2006): 1559–1582, 10.1038/nprot.2006.236.17406449 · doi ↗ · pubmed ↗
- 2C. B. Vogels , A. F. Brito , A. L. Wyllie , et al., “Analytical Sensitivity and Efficiency Comparisons of SARS‐Co V‐2 RT–q PCR Primer–Probe Sets,” Nature Microbiology 5 (2020): 1299–1305, 10.1038/s 41564-020-0761-6.PMC 924136432651556 · doi ↗ · pubmed ↗
- 3P. Song , L. R. Wu , Y. H. Yan , et al., “Limitations and Opportunities of Technologies for the Analysis of Cell‐Free DNA in Cancer Diagnostics,” Nature Biomedical Engineering 6 (2022): 232–245, 10.1038/s 41551-021-00837-3.PMC 933653935102279 · doi ↗ · pubmed ↗
- 4G. Xu , H. Yang , J. Qiu , et al., “Sequence Terminus Dependent PCR for Site‐Specific Mutation and Modification Detection,” Nature Communications 14 (2023): 1169, 10.1038/s 41467-023-36884-4.PMC 997802336859350 · doi ↗ · pubmed ↗
- 5J. S. Chen , E. Ma , L. B. Harrington , et al., “CRISPR‐Cas 12a Target Binding Unleashes Indiscriminate Single‐Stranded D Nase Activity,” Science 360 (2018): 436–439, 10.1126/science.aar 6245.29449511 PMC 6628903 · doi ↗ · pubmed ↗
- 6G. J. Knott and J. A. Doudna , “CRISPR‐Cas Guides the Future of Genetic Engineering,” Science 361 (2018): 866–869, 10.1126/science.aat 5011.30166482 PMC 6455913 · doi ↗ · pubmed ↗
- 7M. Adli , “The CRISPR Tool Kit for Genome Editing and Beyond,” Nature Communications 9 (2018): 1911, 10.1038/s 41467-018-04252-2.PMC 595393129765029 · doi ↗ · pubmed ↗
- 8M. Jinek , K. Chylinski , I. Fonfara , M. Hauer , J. A. Doudna , and E. Charpentier , “A Programmable Dual‐RNA–Guided DNA Endonuclease in Adaptive Bacterial Immunity,” Science 337 (2012): 816–821, 10.1126/science.1225829.22745249 PMC 6286148 · doi ↗ · pubmed ↗