Microfluidic isothermal amplification method and system based on multi-channel spectral detection

Abstract

The present application relates to the technical field of microfluidics, and particularly relates to a microfluidic isothermal amplification method and system based on multi-channel spectral detection, comprising: adding a sample and a reaction premix to a microfluidic chip to form a reaction liquid; at least two fluorescent probes are added to the reaction premix; performing isothermal amplification reaction on the reaction liquid, collecting spectral time series data during the reaction process; performing spectral deconvolution on the spectral time series data to obtain independent amplification curves corresponding to each fluorescent probe and perform quantitative analysis. In view of the problem that the reaction amplification quantitative system in the prior art relies on high-precision quantitative PCR instruments and optical components, and the cost is high, the spectral deconvolution method is introduced to separate the independent amplification curves of each fluorescent probe at the calculation level, thereby eliminating the expensive PCR instrument, and reducing the requirement for the color filter and detection accuracy of the optical component, and the method can be suitable for a wide range of detection scenarios.

Description

Technical Field

[0001] This invention relates to the field of microfluidics, specifically to a microfluidic isothermal amplification method and system based on multi-channel spectral detection. Background Technology

[0002] In 2000, Japanese scholar Notomi published a novel isothermal nucleic acid amplification technique suitable for gene diagnosis in the journal Nucleic Acids Research (Nucleic Acids Research), namely loop-mediated isothermal amplification (LAMP). This method relies on four specific primers (two outer primers and two inner primers) that recognize six conserved sequence regions of the target DNA, and a DNA polymerase with strand displacement activity (such as Bst DNA polymerase). The reaction system generally includes four primers, Bst DNA polymerase buffer, Bst DNA polymerase, dNTPs, template DNA, betaine, MgSO4, etc. The main principle of LAMP technology is to continuously perform strand displacement synthesis using four specific primers and a DNA polymerase with strand displacement activity under isothermal conditions of approximately 65°C, thereby achieving self-circulating DNA amplification. Gene amplification and product detection can be completed in one step, featuring high amplification efficiency and strong specificity, increasing the copy number of the target DNA sequence to approximately 10^60 within 30–60 minutes. 9 ~10 10 times.

[0003] In the existing technology, there are already microfluidic isothermal amplification systems built based on the LAMP reaction principle.

[0004] For example, patent application CN201610879938.4 discloses a portable microfluidic chip LAMP visualization detector and its detection method, including a lid and a housing. The lid and housing are connected by a locking tab on the side. A microfluidic chip, a temperature control device, and a visualization detection system are disposed inside the housing. The temperature control device includes a temperature controller and a heating metal block. The microfluidic chip is placed on the heating metal block. A transparent baffle is arranged around the microfluidic chip and the heating metal block. A removable cover is placed on the transparent baffle. The visualization detection system is disposed on the cover and includes a miniature camera and a miniature ultraviolet fluorescent lamp. The detection method of the aforementioned visualization detector includes the fabrication of the microfluidic chip, the injection of primers and reaction solution, the isothermal reaction, and the reading of the detection results. It also relates to the application of the visualization detector. This enables real-time detection of the LAMP reaction, effectively saving users' time.

[0005] For example, patent application CN201110078450.9 discloses a microfluidic chip for multiplex LAMP detection and its fabrication method. This chip uses a polymer as the chip material and is fabricated using MEMS methods. The basic structure of the chip includes: spatially ordered amplification cells, which effectively distinguish multiplex LAMP signals through spatial signal differentiation; and capillary channels to prevent cross-mixing of LAMP primers, amplification products, and byproducts between different amplification cells. The amplification cells and capillary channels are connected by connecting pipes, allowing fluid to flow smoothly and uniformly from the capillary channels into the amplification cells. This provides an effective solution for the simultaneous detection of multiple pathogens in clinical settings using the LAMP method.

[0006] However, in actual implementation, the inventors found that in this type of technical solution, when there are multiple LAMP reaction signals, high-performance molecular diagnostic equipment, such as quantitative PCR (qPCR) instruments, is usually used. This relies on complex and expensive optical components, such as multiple independent excitation / emission filters, dichroic mirrors, and photomultiplier tubes to achieve multiple detection. This leads to high hardware platform costs, and the optical part usually needs to be redesigned after replacing probes with other fluorescent reactions. Summary of the Invention

[0007] To address the aforementioned problems in the existing technology, a microfluidic isothermal amplification method based on multi-channel spectral detection is provided; furthermore, a system for implementing the microfluidic isothermal amplification method is also provided; furthermore, a pathogen detection method based on the microfluidic isothermal amplification method is provided; furthermore, a genotype analysis method based on the microfluidic isothermal amplification method is provided; furthermore, a food safety detection method based on the microfluidic isothermal amplification method is provided.

[0008] The specific technical solution is as follows: A microfluidic isothermal amplification method based on multi-channel spectral detection, applicable to microfluidic isothermal amplification systems; the microfluidic isothermal amplification method includes: Step S1: adding a sample and a reaction premix to a microfluidic chip to form a reaction liquid; at least two fluorescent probes are added to the reaction premix; Step S2: performing an isothermal amplification reaction on the reaction liquid, and collecting spectral time series data during the reaction process; the spectral time series data covers the exponential growth period and plateau period of the amplification curve in the time domain; Step S3: performing spectral deconvolution on the spectral time series data to obtain independent amplification curves corresponding to each fluorescent probe and performing quantitative analysis.

[0009] On the other hand, the fluorescent probe is selected from 2-6 combinations of fluorescent dyes from the FAM, ROX, HEX, CY5, Texas Red, TAMRA, JOE, VIC, NED, and PET series.

[0010] On the other hand, the emission peak wavelength interval of each fluorescent probe is greater than 20 nm; the emission peak wavelength of the fluorescent probe is between 380-1000 nm.

[0011] On the other hand, a calibration model is constructed before step S1; the calibration model is used to characterize the fluorescence signal corresponding to each fluorescent probe or combination of fluorescent probe concentrations; in step S3, spectral deconvolution is performed based on the calibration model using classical least squares, partial least squares, or principal component regression.

[0012] On the other hand, in step S3, spectral deconvolution is performed using any one of artificial neural networks, support vector machine regression, or random forest regression.

[0013] On the other hand, in step S3, the main components of the spectral time series data are first extracted as core features through principal component analysis, and then an artificial intelligence model is used to perform spectral deconvolution on the core features.

[0014] On the other hand, each time a fluorescence signal is measured, the output fluorescence signal is preprocessed; the fluorescence signal preprocessing includes at least one of dark signal correction, baseline correction, spectral normalization, and noise filtering.

[0015] On the other hand, step S2 includes: step S21: heating the reaction liquid to a predetermined temperature and maintaining it; step S22: when the predetermined temperature is reached, cyclically collecting and recording fluorescence signals on multiple spectral channels of the reaction liquid until a predetermined duration is reached; the predetermined duration is determined according to the plateau period of the isothermal amplification reaction; step S23: assembling the spectral time series data based on the recorded fluorescence signals.

[0016] On the other hand, after performing step S3, the method further includes: verifying the specificity of the amplification product by analyzing the melting curve, and distinguishing between specific amplification and non-specific amplification.

[0017] A microfluidic isothermal amplification system is provided for implementing the aforementioned microfluidic isothermal amplification method. The microfluidic isothermal amplification system includes: a microfluidic chip with two inlets for placing a sample and a reaction premix, respectively; a reaction region within the microfluidic chip connected to the inlets via channels; a heating platform above the microfluidic chip, the heating platform heating the reaction region to a predetermined temperature and maintaining it at a constant temperature; an optical module irradiating the reaction region and collecting fluorescence signals from the reaction region; and a processing device connected to and controlling the heating platform and the optical module, the processing device receiving and analyzing the fluorescence signals.

[0018] On the other hand, the microfluidic chip is made of at least one of polymethyl methacrylate, polycarbonate, cyclic olefin copolymer, glass or quartz.

[0019] On the other hand, the optical module includes: an excitation light source, the type of which is determined according to the excitation band of the fluorescent probe in the reaction premix; and a multi-channel spectral sensor, which acquires the fluorescence signals on multiple spectral channels of the reaction region.

[0020] On the other hand, the processing device includes a microcontroller, a storage unit, a communication interface, and an algorithm processing module; the processing device supports both real-time data processing and offline data analysis modes.

[0021] On the other hand, the excitation source is any one of a laser diode, a white LED, or a broadband light source; the multi-channel spectral sensor has 2-64 independent detection channels; the detection range of the multi-channel spectral sensor covers 350-1000 nm; on the other hand, the microfluidic isothermal amplification system is powered by a battery; the processing device has a low-power mode.

[0022] On the other hand, the microfluidic isothermal amplification system can achieve simultaneous quantitative analysis of at least two fluorescent probes by using a single, fixed optical detection path and only utilizing the spectral resolution capability of the multi-channel spectral sensor for mixed fluorescence signals and the deconvolution algorithm of the processing device.

[0023] A pathogen detection method is implemented based on the microfluidic isothermal amplification method described above; the pathogen includes at least one of viruses, bacteria, and fungi.

[0024] A genotype analysis method is implemented based on the microfluidic isothermal amplification method described above; the genotype analysis method includes single nucleotide polymorphism detection and gene deletion / insertion mutation detection.

[0025] A food safety testing method is implemented based on the above-mentioned microfluidic isothermal amplification method; the food safety testing method is used to detect pathogenic bacteria, allergens, and genetically modified components.

[0026] A reverse transcription analysis method is implemented based on the microfluidic isothermal amplification method described above; the reverse transcription analysis method is used for RNA detection.

[0027] The above technical solution has the following advantages or beneficial effects: In view of the problem that the existing reaction amplification quantitative system relies on high-precision quantitative PCR instruments and optical components, resulting in high costs, this embodiment introduces a spectral deconvolution method to separate the independent amplification curves of each fluorescent probe at the computational level, thereby eliminating the need for expensive PCR instruments and reducing the requirements for the color filtering and detection accuracy of optical components, making it applicable to a wider range of detection scenarios. Attached Figure Description

[0028] Embodiments of the invention will be described more fully with reference to the accompanying drawings. However, the drawings are for illustration and explanation only and do not constitute a limitation on the scope of the invention.

[0029] Figure 1 This is an overall schematic diagram of an embodiment of the present invention;

[0030] Figure 2 This is a schematic diagram of the system in an embodiment of the present invention;

[0031] Figure 3 This is a schematic diagram of the first calibration process in an embodiment of the present invention;

[0032] Figure 4 This is a schematic diagram of step S3 in an embodiment of the present invention;

[0033] Figure 5 This is a schematic diagram of the second calibration process in an embodiment of the present invention;

[0034] Figure 6 This is a schematic diagram of step S2 in an embodiment of the present invention;

[0035] Figure 7 This is a schematic diagram of the optical module in an embodiment of the present invention. Detailed Implementation

[0036] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0038] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0039] This invention includes: a microfluidic isothermal amplification method based on multichannel spectral detection, applicable to microfluidic isothermal amplification systems; such as... Figure 1 As shown, the microfluidic isothermal amplification method includes: Step S1: Adding a sample and a reaction premix to a microfluidic chip to form a reaction liquid; at least two fluorescent probes are added to the reaction premix; Step S2: Performing an isothermal amplification reaction on the reaction liquid, and collecting spectral time series data during the reaction process; the spectral time series data covers the exponential growth phase and plateau phase of the amplification curve in the time domain; Step S3: Performing spectral deconvolution on the spectral time series data to obtain independent amplification curves corresponding to each fluorescent probe and performing quantitative analysis.

[0040] Specifically, in response to the problem that existing isothermal amplification reaction quantitative systems rely on high-precision quantitative PCR instruments and optical modules, resulting in high costs, this embodiment introduces a spectral deconvolution method to separate the independent amplification curves of each fluorescent probe at the computational level. This eliminates the need for expensive PCR instruments and reduces the requirements for color filtering and detection accuracy of the optical module, making it applicable to a wider range of detection scenarios.

[0041] Specifically, the microfluidic isothermal amplification method described above is mainly applicable to microfluidic isothermal amplification systems.

[0042] Here, "sample" refers to the target sample in the detection process, which may be a sample containing DNA or RNA. The most common sample to be detected is a DNA sample, which is amplified by adding a reaction premix containing the corresponding primers, DNA polymerase, and fluorescent probe, and then heating it to a predetermined temperature.

[0043] However, depending on the needs of the test, the sample can also be replaced with an RNA sample. In this case, additional RNA reverse transcriptase needs to be added, and the components in the reaction premix need to be adjusted accordingly. Then, it is heated to the predetermined temperature for amplification.

[0044] Taking DNA sample testing as an example, Figure 2 A simplified schematic diagram of a microfluidic isothermal amplification system is shown. The system includes at least one set of microfluidic chips A1 and a heating platform A2. The microfluidic chip A1 is fixed on the heating platform A2 to heat the DNA sample and reaction premix to the reaction temperature.

[0045] The isothermal amplification reaction is selected from any one of LAMP (Loop-mediated isothermal amplification), RPA (Recombinase Polymerase Amplification), NASBA (Nucleic Acid Sequence-Based Amplification), HAD (Helicase-dependent Amplification), and SDA (Strand Displacement Amplification).

[0046] The heating platform A2 also integrates an optical module A3, which is used to provide an excitation light source for the fluorescent probe in the isothermal amplification reaction process and to collect fluorescence signals for quantitative analysis by the processing device A4.

[0047] To address the issue of requiring color filters in existing technologies, this solution introduces a computational method for deconvolution of spectral channels. This method separates the spectral channels corresponding to multiple fluorescence signals at the computational level, forming independent amplification curves for each fluorescent probe. Quantitative analysis is then performed, reducing the precision requirements at the hardware level. The computational complexity replaces the expensive and bulky physical optical modules (such as multiple sets of filters or grating spectrometers) in traditional detection equipment, as well as the precision thermal cycler in PCR instruments. This significantly reduces system cost and size while maintaining analytical performance.

[0048] Taking the LAMP reaction system as an example, the biochemical basis of this system is loop-mediated isothermal amplification (LAMP) technology. The reaction process roughly involves simultaneously injecting a DNA sample and a LAMP reaction premix into a microfluidic chip and heating it to the amplification temperature, thereby rapidly amplifying the DNA sample.

[0049] The main components of the LAMP reaction premix are primers and biomarkers for isothermal amplification, such as fluorescent probes. A typical reaction premix consists of Bst polymerase, primers, dNTPs, and fluorescent probes. The staining method for the fluorescent probe is determined according to the assay and the type of DNA to be measured.

[0050] When the DNA sample and LAMP reaction premix are heated to the amplification temperature, typically 65 °C, the DNA sample amplifies rapidly. The positive antibody reacts with the fluorescent probe, generating a fluorescent signal during excitation. Because the LAMP reaction is fast, a significant reaction usually occurs within 40 minutes, thus the overall reaction time can be controlled to within 60 minutes.

[0051] Building upon this foundation, sampling points were inserted at multiple time points during the LAMP reaction process. The sampling operation at each point involved exciting the fluorescent probe with a light source of a specific wavelength, acquiring optical signal data from multiple spectral channels as the sampling data for that point, and assembling the sampling data according to time sequence to ultimately obtain the spectral time series data. The start and end times of the sampling points should cover the exponential growth and plateau phases of the amplification curve in the time domain.

[0052] A typical design involves starting sampling after heating to a predetermined temperature, inserting a set of sampling points every 30 seconds, and returning multi-channel spectral data until the reaction time is reached. Using a typical isothermal amplification reaction time of 40 minutes, 80 sampling points are generated within 40 minutes, sufficient to reconstruct the exponential growth phase and plateau phase of the amplification curve, thereby accurately calculating the "Time to Threshold" and performing related analyses.

[0053] The number of spectral channels should be determined according to the spectral range of the fluorescence reaction. A typical number of spectral channels is 12. The light intensity data of 12 channels are collected at each sampling point and combined to obtain the spectral vector at that sampling point.

[0054] Finally, the spectral vectors are added to the spectral time series data according to the acquisition order to form a matrix of data. Spectral deconvolution is then performed on this data matrix to separate the independent amplification curves, and quantitative analysis is then conducted.

[0055] When other isothermal amplification reactions are selected, the primers in the reaction premix and the reaction temperature of the heating platform can be adjusted according to the corresponding isothermal amplification reaction type.

[0056] The following are the reaction temperatures for common amplification reactions:

[0057] Recombinase polymerase amplification (RPA): Usually performed at 37-42 ℃.

[0058] Nucleic Acid Sequence-Based Amplification (NASBA): Its standard reaction temperature is 41 °C.

[0059] Helicase-dependent amplification (HDA): The reaction temperature is usually set at 65 °C.

[0060] Strand displacement amplification (SDA): The reaction temperature range is relatively wide, usually between 37-60°C.

[0061] After changing the amplification reaction, the corresponding sampling interval and duration should be adjusted according to the type of amplification reaction so that the spectral time series data covers the exponential growth period and plateau period of the amplification curve in the time domain.

[0062] In one embodiment, the fluorescent probe is selected from 2 to 6 combinations of fluorescent dyes from the FAM, ROX, HEX, CY5, Texas Red, TAMRA, JOE, VIC, NED, and PET series.

[0063] In another embodiment, the fluorescent probe consists of FAM and ROX.

[0064] Specifically, when selecting fluorescent probes, multiple fluorescent probes are often added to achieve better measurement results. However, when two or more fluorescent probes have similar emission wavelengths, they may overlap in the spectrum, making effective separation difficult.

[0065] Therefore, when selecting fluorescent probes, the emission bands of each fluorescent probe should be kept as separate as possible.

[0066] Based on this, analysis of the fluorescent probes revealed that when FAM (emission peak approximately 517 nm) and ROX (emission peak approximately 604 nm) were used as fluorescent probes, fluorescence emission was primarily detected by the sensor's F4 (center wavelength 515 nm), FY (center wavelength 555 nm), and F5 (center wavelength 550 nm) channels. ROX fluorescence emission mainly fell within the response range of the FXL (center wavelength 600 nm) and F6 (center wavelength 640 nm) channels. FAM only caused slight spectral crosstalk in the ROX detection channel. Therefore, choosing these two dyes resulted in better analytical performance.

[0067] However, in actual testing, depending on the genotype to be detected, other combinations of fluorescent probes may be selected, such as combinations of 2-6 fluorescent dyes from the FAM, ROX, HEX, CY5, Texas Red, TAMRA, JOE, VIC, NED, and PET series.

[0068] Generally, to achieve better spectral deconvolution, the wavelength interval of the emission peaks of each fluorescent probe needs to be greater than 20 nm to reduce crosstalk.

[0069] At the same time, it is also necessary to control the emission peak wavelength within the detection range of the multispectral sensor, such as a typical range of 380-1000 nm.

[0070] Based on the above process design, optical deconvolution can be achieved using methods such as classical least squares, partial least squares, or principal component regression.

[0071] Taking the classical least squares method as an example, in this embodiment, a first calibration model is constructed for the microfluidic chip before executing step S1; in step S3, the spectral time series data is deconvolved based on the first calibration model; as follows... Figure 3 As shown, the first calibration process for generating the first calibration model includes: Step A01: For each fluorescent probe, the corresponding positive template DNA is used to react in a microfluidic chip. After the plateau phase, the fluorescence signal emitted by the fluorescent probe is measured as a first multi-channel spectral sample signal; the first multi-channel spectral sample signal has the same spectral channels as the spectral time series data; Step A02: The first calibration model is assembled based on all the first multi-channel spectral sample signals; in the first calibration model, each multi-channel spectral sample signal is used as a row of a matrix, and the spectral channels are used as columns of the matrix.

[0072] Specifically, due to the broad peak characteristics of the fluorescent dye emission spectrum and the limited spectral resolution of the sensor channel, the signal measured by this system is a mixed signal. At any given time point... The signal vector measured by the system (A vector containing 12 channel intensity values) is not the response of a single dye, but a linear superposition of the signals of all fluorescent components in the sample (FAM and ROX in this case).

[0073] This problem can be precisely described using a linear algebraic model. The data collected throughout the reaction process can be represented as a data matrix. Each row represents a measurement vector at a given time point. Each column represents the signal variation of a specific spectral channel at all time points. This is a mixed data matrix. It can be modeled as the product of two (or more) more fundamental matrices, plus an error term:

[0074] ;

[0075] In the formula, This represents the spectral time series data obtained from actual measurements. This represents the actual concentration matrix of each fluorescent probe. yes The pure component spectral matrix, also known as the calibration matrix, This represents the transpose of the spectral matrix of the pure component. Each row represents a pure component (such as FAM) in The response vectors on each spectral channel, i.e., their "spectral fingerprints," have a total of [number] rows. , This represents the error term to be determined. This matrix needs to be determined in advance through calibration experiments.

[0076] Before performing step S1, a first calibration model needs to be built in advance.

[0077] The process of constructing the first calibration model typically involves obtaining the spectral fingerprint of a single fluorescent probe. This spectral fingerprint may vary depending on the actual dye model, measurement system, etc., and therefore needs to be calibrated separately.

[0078] Specifically, for each fluorescent probe, an isothermal amplification reaction system containing only that type of labeled probe needs to be prepared beforehand. Sufficient positive template DNA is added, and the reaction is allowed to proceed to the plateau phase, at which point the FAM fluorescence signal reaches its maximum and stabilizes. Under these conditions, the spectral signals of its 12 channels are measured using this device, resulting in a 12-dimensional vector. This vector, after correlation processing, can be used as the corresponding spectral fingerprint.

[0079] Repeat this process to obtain the spectral fingerprint of each fluorescent probe, and then construct a calibration matrix. .

[0080] Calibration matrix of FAM and ROX probes in a 12-channel detection system For example, calibration matrix It is a 2×12 matrix:

[0081] .

[0082] In one embodiment, such as Figure 4 As shown, step S3 includes: step A31: calculating the first concentration matrix by least squares method on the first calibration model and spectral time series data; step A32: splitting the first concentration matrix into column vectors to serve as the amplification curves for each fluorescent probe; step A33: performing quantitative analysis on the amplification curves.

[0083] Specifically, to achieve a more convenient calculation effect, in this embodiment, the first concentration matrix is ​​first calculated using the least squares method on the first calibration model and the spectral time series data. Specifically:

[0084] ;

[0085] In the formula, The first concentration matrix that needs to be solved is... For spectral time series data, It is the first calibration model. It is the transpose matrix of the first calibration model.

[0086] After obtaining the first concentration matrix, matrix calculations are performed on each column of the first concentration matrix to obtain the amplification curve of each fluorescent probe. Finally, quantitative analysis is performed to obtain the corresponding results.

[0087] Taking partial least squares as an example, in this embodiment, a second calibration model is constructed for the microfluidic chip before step S1; in step S3, partial least squares is used to deconvolve the spectral time series data based on the second calibration model; as shown... Figure 5 As shown, the second calibration process for generating the second calibration model includes: Step B01: Constructing multiple concentration combinations for all fluorescent probes, and mixing the fluorescent probes based on the concentration combinations to obtain a test mixture; Step B02: Reacting the positive template DNA with the test mixture in a microfluidic chip, and measuring the second multi-channel spectral sample signal of the emitted mixed fluorescence signal after the plateau phase; The second multi-channel spectral sample signal has the same spectral channels as the spectral time series data; Step B03: Calculating the second calibration model based on all the second multi-channel spectral sample signals; The second calibration model is a regression model used to characterize the dye concentration from the second multi-channel spectral sample signal.

[0088] Specifically, to achieve better model robustness, this embodiment selects partial least squares (PLS) as an alternative during the deconvolution process. Unlike least squares, PLS is a supervised regression method that actively seeks out spectral change patterns most relevant to known concentrations when building the model, and can better handle collinearity and noise among variables.

[0089] To implement this processing step, in this embodiment, during the construction of the calibration model, a calibration set containing a series of standard samples is constructed based on the combination of multiple fluorescent probes. The combination of concentrations of each fluorescent probe in these samples is known and can cover any numerical range that may occur during actual measurement.

[0090] Based on this, fluorescent probes were mixed according to a pre-designed concentration combination to obtain a test mixture. Corresponding positive template DNA was then reacted with the test mixture in a microfluidic chip. After the plateau phase, the emitted mixed fluorescence signal was measured as a second multi-channel spectral sample signal, which was consistent with the concentration combination. Subsequently, a regression model from the 12-channel spectral data to the concentration of each fluorescent probe was constructed as a second calibration model.

[0091] The second calibration model is used to perform partial least squares calculations to achieve better deconvolution results.

[0092] In another embodiment, Principal Component Analysis (PCA) is used to achieve spectral deconvolution.

[0093] For the principal component analysis process, a calibration set containing a series of standard samples was first constructed based on the combination of multiple fluorescent probes. The combination of concentrations of each fluorescent probe in these samples is known and can cover any numerical range that may occur during actual measurement.

[0094] Based on this, fluorescent probes are mixed according to a pre-designed concentration combination to obtain a test mixture. Corresponding positive template DNA is then reacted with the test mixture in a microfluidic chip. After the plateau phase, the emitted mixed fluorescence signal is measured to obtain the third multi-channel spectral sample signal. A known concentration matrix is ​​constructed based on multiple third multi-channel spectral signals as prior parameters.

[0095] In actual testing, principal component analysis is performed on the preprocessed spectral time series data to decompose it into a few uncorrelated principal components (PCs). These principal components capture the main variances of the original spectral data.

[0096] Then, using the extracted principal components as independent variables and the previously known concentration matrix based on the standard samples as the dependent variable, a multiple linear regression model is established.

[0097] For unknown samples, their spectral data are first projected into the established PCA space to obtain their principal component scores, and then their component concentrations are predicted using a regression model. Compared to PLS, PCR only considers the variations in the spectral data itself when extracting principal components, without considering their correlation with concentration, but it can still provide stable and reliable deconvolution results in many application scenarios.

[0098] In addition, in some other embodiments, in step S3, spectral deconvolution is performed using any one of artificial neural networks, support vector machine regression, or random forest regression.

[0099] This type of deconvolution method mainly relies on pre-collecting samples, including adjusting different concentration combinations and collecting corresponding fluorescence spectral data and adding annotations based on a given probe type and corresponding amplification reaction, thereby forming a large-scale calibration dataset.

[0100] Then, using the spectral data as input features and the known concentration as the output label, a machine learning regression model is trained, for example:

[0101] Artificial neural networks: By constructing a network structure that includes an input layer, hidden layers, and an output layer, they learn the complex nonlinear relationship between spectra and concentrations.

[0102] Support Vector Machine Regression: Finds an optimal hyperplane that minimizes the error of all sample points to that hyperplane.

[0103] Random forest regression: This method involves constructing multiple decision trees and averaging their results.

[0104] In some embodiments, ensemble learning and hybrid algorithm strategies may also be used for deconvolution.

[0105] Taking the combination strategy of principal component analysis and artificial neural network as an example, this strategy requires the prior collection of samples, including, based on the given probe types and corresponding amplification reactions, adjusting different concentration combinations and collecting corresponding fluorescence spectral data and adding annotations, thereby forming a large-scale calibration dataset.

[0106] The artificial neural network is trained using a calibration dataset to obtain a well-trained artificial neural network.

[0107] In the actual processing, the collected spectral time series data are first preprocessed and denoised. Then, principal component analysis is used to reduce the dimensionality of the preprocessed spectral data and extract the first 5-10 principal components as core features.

[0108] The core features are then input into the trained artificial neural network, which ultimately outputs the concentration curves of each component after deconvolution.

[0109] In one embodiment, before each measurement of the fluorescence signal, the dark signal of the microfluidic isothermal amplification system is measured in advance; and after performing step S2 and before performing step S3, the measured value of the spectral time series data is subtracted from the dark signal in advance; and the drift baseline is subtracted from the spectral time series data using the asymmetric least squares method.

[0110] Specifically, considering the potential noise and baseline drift issues in the measurement system, in this embodiment, the above-described method is used to preprocess the spectral time series data output by the measurement system for each measurement, including the fluorescence signal output during pre-modeling and the spectral time series data output during actual measurement.

[0111] Specifically, for the original measured signal, since the system itself has ambient light background and dark current, the dark signal of the microfluidic isothermal amplification system is measured in advance before each measurement, and the dark signal is directly subtracted when outputting the original measurement data to avoid the influence of background noise.

[0112] Then, during the isothermal amplification reaction, slow, non-linear baseline drift may occur due to reagent degradation or minor temperature fluctuations. Therefore, for each column of the spectral time-series data matrix (i.e., the time-series data for each channel), an advanced baseline correction algorithm, such as asymmetric least squares (AsLS), is applied. AsLS can intelligently identify and subtract drifted baselines while preserving the true amplification signal peaks, making it ideal for automated processing.

[0113] Simultaneously, data can be normalized according to spectral channels as needed to eliminate potential physical differences between different reaction wells. Commonly used methods include Standard Normal Variation (SNV) or Multiplicative Scatter Correction (MSC). These methods can effectively eliminate multiplicative interference effects and enhance the consistency between experimental data from different samples or batches.

[0114] Additionally, noise filtering is used to remove signal noise from the sampling system.

[0115] In one embodiment, such as Figure 6 As shown, step S2 includes: step S21: heating the reaction liquid to a predetermined temperature and maintaining it; step S22: when the predetermined temperature is reached, collecting and recording fluorescence signals on multiple spectral channels of the reaction liquid in a cyclic manner until a predetermined duration is reached; the predetermined duration is determined according to the plateau period of the isothermal amplification reaction; step S23: assembling spectral time series data based on the recorded fluorescence signals.

[0116] Specifically, to achieve better reaction results, a PTC (Positive Temperature Coefficient) heating device was selected as the heating platform in this embodiment. The PTC device is controlled in either open or closed loop, automatically stopping and maintaining the temperature when it reaches 65 °C. Then, according to a pre-configured timer, fluorescence signals from multiple spectral channels of the reaction liquid are collected and added to a data matrix for recording until the predetermined time is reached, at which point the data is output.

[0117] In one embodiment, after performing step S3, the method further includes: verifying the specificity of the amplification product by analyzing the melting curve, and distinguishing between specific amplification and non-specific amplification.

[0118] Specifically, the melting curve of the amplification products can be obtained by slowly lowering the temperature of the PTC heater while continuously monitoring the fluorescence signal. Specific and non-specific products typically have different melting temperatures (Tm values), which can be used to distinguish between true positive and false positive results.

[0119] Specifically, after the standard isothermal amplification reaction (such as LAMP reaction for 45 minutes) is completed, the system automatically enters the melting curve analysis stage.

[0120] Then, the heating platform needs to be controlled to raise the temperature at a slow and precise rate, for example, linearly increasing it from 65 °C to 95 °C at a rate of 0.5 °C / s. The upper temperature limit is determined based on the highest melting temperature of the specific product expected to appear in the current reaction.

[0121] At each temperature point during the heating process, the optical detection module simultaneously acquires full-channel spectral data as melting spectral data.

[0122] The data processing unit processes the acquired data in real time. It selects the spectral channel that is most sensitive to the signal of a double-stranded DNA intercalation dye (such as SYBR Green, which can be added to the premix) or a specific probe, and plots the raw melting curve of "fluorescence intensity vs. temperature".

[0123] By calculating the negative first derivative (-dF / dT) of the original melting curve, one or more peaks can be obtained. The temperature corresponding to the peak of each peak is the melting temperature (Tm value).

[0124] Specific products: typically appear as a single, sharp melting peak within the expected temperature range (e.g., 85–90 °C). Non-specific products (e.g., primer dimers): typically have lower Tm values ​​(e.g., <80 °C) and broader peak shapes or exhibit multiple mixed peaks.

[0125] The above methods are used to distinguish between specific and non-specific products.

[0126] A microfluidic isothermal amplification system is provided for implementing the aforementioned microfluidic isothermal amplification method; such as... Figure 2As shown, the microfluidic isothermal amplification system includes: a microfluidic chip A1, which has two inlets for placing DNA samples and reaction premixed solutions, respectively; a reaction region is provided in the microfluidic chip A1, and the reaction region is connected to the inlets via flow channels; a heating platform A2, which is positioned above the microfluidic chip A1, and the heating platform A2 heats the reaction region to a predetermined temperature and maintains it at a constant temperature; an optical module A3, which irradiates the reaction region and collects the fluorescence signal of the reaction region; and a processing device A4, which is connected to and controls the heating platform A2 and the optical module A3, and receives and analyzes the fluorescence signal.

[0127] Specifically, to achieve better measurement results, the aforementioned microfluidic isothermal amplification system was constructed in this embodiment. The microfluidic chip A1 is a reaction chip made of polymer material, in which channels and multiple chambers are etched for injecting DNA samples and reaction premixes, eliminating air bubbles in the liquid, uniformly mixing the liquid, and finally introducing it into the reaction area.

[0128] The reaction region is a chamber with a defined volume and optical path. When the detection system is assembled, the reaction region is precisely positioned at the center of the heating surface of the heating platform A2, so that the reaction region is uniformly heated to a precise predetermined temperature; and the reaction region is also located at the focal point of the optical module A3.

[0129] The heating platform A2 is a heating device that can uniformly heat the microfluidic chip A1 to a predetermined temperature and automatically maintain it, with minimal temperature fluctuations throughout the overall reaction process. Common designs include PTC heating modules, nichrome resistance wire heaters, Peltier heaters, etc., achieving good temperature control through open-loop or closed-loop control.

[0130] The optical module A3 mainly consists of a light source and a sensor, used to excite the fluorescent probe and collect light signals at specific points in the reaction process. This control process is implemented by the processing device A4.

[0131] The processing device A4 is a computer device capable of running specific programs to execute the aforementioned processes. It can be implemented using a microprocessor, microcontroller, or, in the form of a data acquisition card, directly connected to an external computer. Common designs are based on microcontrollers such as STM32 or Raspberry Pi.

[0132] This microfluidic isothermal amplification system can achieve simultaneous quantitative analysis of at least two fluorescent probes by using a single, fixed optical detection path and relying solely on the spectral resolution of mixed fluorescence signals by a multi-channel spectral sensor and the deconvolution algorithm of the processing device.

[0133] In one embodiment, the microfluidic chip A1 is made of polymethyl methacrylate.

[0134] Specifically, to achieve good thermal conductivity and optical performance, polymethyl methacrylate (PMMA) was selected as the material for the microfluidic chip in this embodiment. The selection factors include:

[0135] Optical transparency: In particular, the material used to construct the detection area must have high transmittance in the excitation and emission wavelength range of the system (approximately 450-650 nm), and its own background fluorescence (autofluorescence) should be as low as possible to ensure a high signal-to-noise ratio.

[0136] Biocompatibility: The inner surface of the chip must be chemically inert and must not adsorb or inhibit key biomolecules (such as DNA polymerase, primers and template DNA) in the isothermal amplification reaction.

[0137] Thermal conductivity: The bottom of the chip needs to have good thermal contact with the PTC heater to ensure that heat can be efficiently and evenly transferred to the reaction liquid.

[0138] In other embodiments, at least one of polycarbonate, cyclic olefin copolymer, glass, or quartz may also be used.

[0139] In one embodiment, such as Figure 7 As shown, the optical module A3 includes: a laser light source A31, the type of which is determined according to the excitation band of the fluorescent probe in the reaction premix; and a multi-channel spectral sensor A32, which collects fluorescence signals from multiple spectral channels in the reaction region.

[0140] Specifically, to achieve better measurement results, in this embodiment, the corresponding laser source was first determined according to the excitation wavelength of the fluorescent probe in the reaction premix. In portable devices, a constant wavelength LED laser source is typically used for excitation. Taking FAM (maximum excitation wavelength approximately 495 nm) and ROX (maximum excitation wavelength approximately 578 nm) fluorescent probes as examples, a typical high-power blue LED (center wavelength approximately 470 nm) can achieve good excitation efficiency. Two different wavelength LED laser sources can also be used for time-division excitation as needed.

[0141] In addition, white LEDs or broadband light sources can be used for excitation as needed.

[0142] The A32 multichannel spectral sensor is a sensor that integrates multiple spectral channels. It is a lower-cost instrument compared to a grating spectrometer, but can achieve the same detection effect based on the above calculation process.

[0143] A typical design is the AMS AS7343 multichannel spectral sensor, which integrates 12 spectral channels covering the visible to near-infrared bands, as well as a fully transparent channel and a scintillation detection channel. These 12 spectral channels provide enough data points to effectively separate the mixed spectra of 2-3 fluorescent dyes.

[0144] In other embodiments, different detection channels may be used depending on the device selection, but typically there are 2-64 independent detection channels, covering a detection range of 350-1000 nm.

[0145] Because the above method is used for spectral deconvolution, the multi-channel spectral sensor can directly acquire the spectrum without adding additional color filters.

[0146] In addition, the processing device includes a microcontroller, a storage unit, a communication interface, and an algorithm processing module; the processing device supports both real-time data processing and offline data analysis modes, providing greater detection flexibility.

[0147] The processing device has a low-power mode, which, when combined with battery power, makes the detection system suitable for rapid on-site testing.

[0148] In addition, a pathogen detection method is provided, based on the microfluidic isothermal amplification method described above; the pathogen includes at least one of viruses, bacteria, and fungi.

[0149] A genotyping method is based on the microfluidic isothermal amplification method described above; the genotyping method includes single nucleotide polymorphism (SNP) detection and gene deletion / insertion mutation detection.

[0150] A food safety testing method is implemented based on the microfluidic isothermal amplification method described above; the food safety testing method is used to detect pathogenic bacteria, allergens, and genetically modified components.

[0151] A reverse transcription analysis method is implemented based on the microfluidic isothermal amplification method described above; the reverse transcription analysis method is used for RNA detection.

[0152] The above are merely preferred embodiments of the present invention and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should recognize that any equivalent substitutions and obvious changes made based on the description and illustrations of the present invention should be included within the protection scope of the present invention.

Claims

1. A microfluidic isothermal amplification method based on multi-channel spectral detection, characterized in that, Suitable for microfluidic isothermal amplification systems; The microfluidic isothermal amplification method described is a microfluidic isothermal amplification method for non-diagnostic scenarios. The microfluidic isothermal amplification method includes: Step S1: Add sample and reaction premix to the microfluidic chip to form reaction liquid; At least two fluorescent probes are added to the reaction premix; The emission peak wavelength interval of each fluorescent probe is greater than 20 nm; Step S2: Perform an isothermal amplification reaction on the reaction liquid and collect spectral time series data during the reaction process; The spectral time series data covers the exponential growth phase and plateau phase of the amplification curve in the time domain; The spectral time series data includes mixed fluorescence signals corresponding to the fluorescent probe; Step S3: Perform spectral deconvolution on the spectral time series data to obtain independent amplification curves corresponding to each fluorescent probe and perform quantitative analysis; Prior to step S1, a corresponding spectral deconvolution algorithm is selected, and a calibration model corresponding to the spectral deconvolution algorithm is constructed. The calibration model is used to characterize the fluorescence signal corresponding to each fluorescent probe or combination of fluorescent probe concentrations; In step S3, spectral deconvolution is performed based on the selected spectral deconvolution algorithm and the calibration model.

2. A microfluidic isothermal amplification method based on multi-channel spectral detection, characterized in that, Suitable for microfluidic isothermal amplification systems; The microfluidic isothermal amplification method described is a microfluidic isothermal amplification method for non-diagnostic scenarios. The microfluidic isothermal amplification method includes: Step S1: Add sample and reaction premix to the microfluidic chip to form reaction liquid; At least two fluorescent probes are added to the reaction premix; The emission peak wavelength interval of each fluorescent probe is greater than 20 nm; Step S2: Perform an isothermal amplification reaction on the reaction liquid and collect spectral time series data during the reaction process; The spectral time series data covers the exponential growth phase and plateau phase of the amplification curve in the time domain; The spectral time series data includes mixed fluorescence signals corresponding to the fluorescent probe; Step S3: Perform spectral deconvolution on the spectral time series data to obtain independent amplification curves corresponding to each fluorescent probe and perform quantitative analysis; Before step S1, a corresponding machine learning regression model is selected, and a calibration dataset corresponding to the machine learning regression model is constructed to train the model; The calibration dataset is used to characterize the fluorescence signal corresponding to each fluorescent probe or combination of fluorescent probe concentrations; In step S3, spectral deconvolution is performed based on the pre-trained machine learning regression model.

3. The microfluidic isothermal amplification method according to claim 1 or 2, characterized in that, The fluorescent probes are selected from 2-6 combinations of fluorescent dyes from the FAM, ROX, HEX, CY5, Texas Red, TAMRA, JOE, VIC, NED, and PET series.

4. The microfluidic isothermal amplification method according to claim 1 or 2, characterized in that, The emission peak wavelength of the fluorescent probe is between 380 and 1000 nm.

5. The microfluidic isothermal amplification method according to claim 1, characterized in that, In step S3, spectral deconvolution is performed based on the calibration model using classical least squares, partial least squares, or principal component regression.

6. The microfluidic isothermal amplification method according to claim 2, characterized in that, In step S3, spectral deconvolution is performed using any one of artificial neural networks, support vector machine regression, or random forest regression.

7. The microfluidic isothermal amplification method according to claim 2, characterized in that, In step S3, the main components of the spectral time series data are first extracted as core features through principal component analysis, and then an artificial intelligence model is used to perform spectral deconvolution on the core features.

8. The microfluidic isothermal amplification method according to claim 1 or 2, characterized in that, For each measurement of fluorescence signal, the output fluorescence signal is preprocessed; The fluorescence signal preprocessing includes at least one of dark signal correction, baseline correction, spectral normalization, and noise filtering.

9. The microfluidic isothermal amplification method according to claim 1 or 2, characterized in that, Step S2 includes: Step S21: Heat the reaction liquid to a predetermined temperature and maintain it; Step S22: When the predetermined temperature is reached, the fluorescence signals on multiple spectral channels of the reaction liquid are collected and recorded in a cyclic manner until the predetermined time is reached; The predetermined duration is determined based on the plateau phase of the isothermal amplification reaction; Step S23: Assemble the spectral time series data based on the recorded fluorescence signals.

10. The microfluidic isothermal amplification method according to claim 1 or 2, characterized in that, After performing step S3, the method further includes: The specificity of the amplification products was verified by melting curve analysis, which distinguished between specific and non-specific amplification.

11. A microfluidic isothermal amplification system, characterized in that, Used to implement the microfluidic isothermal amplification method as described in any one of claims 1-10; The microfluidic isothermal amplification system includes: A microfluidic chip, wherein the microfluidic chip is provided with two liquid inlets, for inserting a sample and a reaction premix, respectively; The microfluidic chip is provided with a reaction region, and the reaction region is connected to the liquid inlet through flow channels; A heating platform, wherein the microfluidic chip is disposed above the heating platform, and the heating platform heats the reaction region to a predetermined temperature and maintains a constant temperature; An optical module that irradiates the reaction region and collects the fluorescence signal of the reaction region; A processing device, which is connected to and controls the heating platform and the optical module, receives the fluorescence signal and analyzes it. The microfluidic isothermal amplification system can achieve simultaneous quantitative analysis of at least two fluorescent probes by using a single, fixed optical detection path and relying solely on the spectral resolution capability of the multi-channel spectral sensor for mixed fluorescence signals and the deconvolution algorithm of the processing device.

12. The microfluidic isothermal amplification system according to claim 11, characterized in that, The microfluidic chip is made of at least one of polymethyl methacrylate, polycarbonate, cyclic olefin copolymer, glass, or quartz.

13. The microfluidic isothermal amplification system according to claim 11, characterized in that, The optical module includes: An excitation light source, the type of which is determined according to the excitation band of the fluorescent probe in the reaction premix; A multi-channel spectral sensor, wherein the multi-channel spectral sensor acquires fluorescence signals on multiple spectral channels in the reaction region.

14. The microfluidic isothermal amplification system according to claim 13, characterized in that, The processing device includes a microcontroller, a storage unit, a communication interface, and an algorithm processing module; The processing device supports both real-time data processing and offline data analysis modes.

15. The microfluidic isothermal amplification system according to claim 13, characterized in that, The excitation source is any one of a laser diode, a white LED, or a broadband light source; The multi-channel spectral sensor has 2-64 independent detection channels; The detection range of the multi-channel spectral sensor covers 350-1000 nm.

16. The microfluidic isothermal amplification system according to claim 13, characterized in that, The microfluidic isothermal amplification system is powered by a battery. The processing device has a low-power mode.

17. A food safety testing method, characterized in that, This is achieved based on the microfluidic isothermal amplification method according to any one of claims 1-10; The food safety testing method is used to detect pathogens, allergens, and genetically modified ingredients.

Images