Method for screening antibodies for flow detection

Abstract

The invention provides an antibody screening method for flow cytometry, and belongs to the technical field of flow cytometry. Antibody purity analysis and concentration determination are carried out through high performance liquid chromatography; the binding characteristic of the antibody and the target cell is researched by adopting an equal-ratio dilution method; analyzing the combined signal by using an improved Hilbert-Huang decomposition algorithm; evaluating the combination stability through covariance matrix decomposition; collecting cell event data by using a flow cytometer; constructing a fluorescence intensity contribution matrix and performing signal optimization; calculating a signal stability contribution value to determine an effective detection area; a 0-1 knapsack optimization method is adopted to carry out multiple detection combination optimization; finally, repeatability verification is carried out through a bootstrap resampling method, the stable-performance antibody for flow detection is screened out, and the technical problems that in the prior art, the signal stability of antibody screening for flow detection is poor, and specificity is insufficient are solved.

Description

Technical Field

[0001] The invention belongs to the technical field of flow cytometry, and in particular relates to an antibody screening method for flow cytometry detection. Background Art

[0002] Flow cytometry is an indispensable analytical tool in modern biomedical research and clinical diagnosis. Its core principle is to recognize specific molecules on the cell surface or inside the cell through fluorescently labeled antibodies, thereby achieving cell classification and analysis. High-quality antibodies for flow cytometry are essential for obtaining accurate and reliable experimental results. Traditional antibody screening methods for flow cytometry mainly rely on enzyme-linked immunosorbent assay, Western blot analysis, immunohistochemical staining and other technical means for preliminary screening, and then verification by flow cytometry. This screening method has been widely used in scientific research and clinical fields, providing important support for applications such as disease diagnosis, immunophenotyping, and cell sorting.

[0003] However, traditional screening methods have several significant defects. First, the evaluation of the binding stability between antibodies and target antigens lacks quantitative indicators and mainly relies on empirical judgment, which leads to a strong subjectivity in the screening results. Secondly, it is difficult for existing methods to effectively distinguish specific binding signals from nonspecific background signals, especially in multiple staining experiments, where the crosstalk effect between different fluorescent channels will affect the accuracy of the test results. In addition, traditional methods fail to fully consider the impact of factors such as signal drift and instrument parameter fluctuations during flow cytometry on the test results, resulting in the screened antibodies having unstable performance in practical applications.

[0004] These technical defects directly affect the accuracy and repeatability of flow detection. Especially in multi-parameter flow detection, due to the lack of systematic signal analysis and optimization methods, the selection of antibody combinations lacks scientific basis, and the reliability of the test results is difficult to guarantee. Therefore, how to improve the screening efficiency of antibodies for flow detection and ensure the accuracy and repeatability of screening results has become a key technical problem to be solved in this field. In other words, there are technical problems in the prior art such as poor signal stability and insufficient specificity in antibody screening for flow detection. Summary of the invention

[0005] In view of this, the present invention provides a method for screening antibodies for flow cytometry, which can solve the technical problems of poor signal stability and insufficient specificity in the prior art of antibody screening for flow cytometry.

[0006] The present invention is implemented as follows: The present invention provides a method for screening antibodies for flow cytometry detection, including the following steps: performing purity analysis on the monoclonal antibody to be screened, determining the antibody purity by high-performance liquid chromatography, calculating the sample purity by second derivative peak shape analysis, measuring the protein concentration by the peak area method, and selecting a monoclonal antibody sample with a purity greater than 95% and a protein concentration in the range of 1 to 5 mg / ml; preparing a concentration gradient of the monoclonal antibody sample by the equal ratio dilution method, and incubating it with target antigen-positive cells and target antigen-negative cells at 2 to 8°C for 30 to 60 minutes respectively; calculating the binding affinity constant of the monoclonal antibody sample and the target antigen-positive cells by non-linear fitting; analyzing the binding signal of the monoclonal antibody sample and the target antigen-positive cells by an improved Hilbert-Huang decomposition algorithm, establishing a binding intensity matrix, and obtaining a stable binding component and a non-specific binding component; calculating a binding stability coefficient matrix by covariance matrix decomposition based on the stable binding component and the non-specific binding component; detecting the cell suspension labeled with the monoclonal antibody by flow cytometry; constructing a fluorescence intensity contribution matrix and performing calculations using a signal optimization equation set; constructing a fluorescence intensity contribution matrix based on the optimal signal distribution parameters and calculating the signal stability contribution value using an exponential decay function; performing multiple detection combination optimization on the monoclonal antibody samples within the effective detection region; verifying the repeatability of the optimal detection combination scheme by the bootstrap resampling method, and screening out the antibody for flow cytometry detection.

[0007] Among them, the specific steps for preparing a concentration gradient of the monoclonal antibody sample by the equal ratio dilution method are as follows: First, dilute the monoclonal antibody sample with phosphate buffer to an initial concentration of 10 μg per 1 million cells, and then sequentially dilute it at a ratio of 1:2 to prepare 6 concentration gradient points, with the lowest concentration being 1 μg per 1 million cells.

[0008] Among them, the specific steps for analyzing the binding signal of the monoclonal antibody sample and the target antigen-positive cells by an improved Hilbert-Huang decomposition algorithm are as follows: First, perform a Hilbert transform on the original binding signal to obtain the instantaneous frequency and instantaneous amplitude of the signal, and then use the empirical mode decomposition method to decompose the signal into multiple intrinsic mode functions and a residual function. Identify the stable binding component and the non-specific binding component by screening the frequency characteristics of the function, and perform spectral analysis on the decomposed intrinsic mode functions to establish a 10- to 50-dimensional binding intensity matrix.

[0009] Among them, the specific steps for detecting the cell suspension labeled with monoclonal antibodies using a flow cytometer are as follows: Adjust the cell suspension to a concentration of 1 million cells per milliliter, filter out cell clumps through a 200-mesh cell sieve, set the forward scatter threshold at 5000 to 50000 and the side scatter threshold at 2000 to 20000 on the flow cytometer, and collect 50000 valid cell events at a collection rate of 200 to 500 cells per second.

[0010] Among them, the signal optimization equation set includes a signal gain equation, a signal compensation equation, a background correction equation, and a distribution fitting equation. The signal gain equation is used to correct the influence of the instrument gain coefficient on the fluorescence signal. The signal compensation equation is used to eliminate the spectral overlap between different fluorescence channels. The background correction equation is used to deduct the sample background and non-specific fluorescence. The distribution fitting equation is used to fit the distribution characteristics of the fluorescence signal.

[0011] Among them, the input data of the signal gain equation include the original fluorescence signal intensity collected by the flow cytometer, the voltage gain coefficient of the flow cytometer, and the channel sensitivity parameter of the flow cytometer. The output data of the signal gain equation is the corrected true fluorescence intensity value.

[0012] Among them, the input data of the signal compensation equation include the corrected true fluorescence intensity value, the inter-channel crosstalk coefficient matrix of the flow cytometer, and the single-staining control data of the monoclonal antibody sample. The output data of the signal compensation equation is the pure fluorescence signal after removing crosstalk.

[0013] Among them, the input data of the background correction equation include the pure fluorescence signal after removing crosstalk, the control signal value of the target antigen-negative cells, and the spontaneous fluorescence signal intensity of the target antigen-negative cells. The output data of the background correction equation is the specific fluorescence signal after deducting the background.

[0014] Among them, the input data of the distribution fitting equation include the specific fluorescence signal after deducting the background, the number distribution of valid cell events, and the forward scatter threshold and side scatter threshold. The output data of the distribution fitting equation is the optimal signal distribution parameter.

[0015] Among them, the specific steps for optimizing the multiplex detection combination of the monoclonal antibody sample within the effective detection region are as follows: Take the specificity of the monoclonal antibody sample as the value coefficient and the detection cost of the monoclonal antibody sample as the weight coefficient, and use the 0-1 knapsack optimization method to calculate the optimal detection combination scheme.

[0016] Compared with the prior art, the present invention provides a method for screening antibodies for flow cytometry detection, and the present invention proposes a method for screening antibodies for flow cytometry detection based on multi-dimensional signal analysis and optimization. This method realizes the precise quantification and evaluation of antibody binding characteristics by establishing a systematic evaluation system and combining mathematical tools such as improved Hilbert-Huang decomposition algorithm, covariance matrix decomposition, and 0-1 knapsack optimization.

[0017] The method of the present invention solves the defects of the traditional technology from multiple aspects. In terms of signal analysis, by introducing an improved Hilbert-Huang decomposition algorithm, the effective separation of the stable binding component and the non-specific binding component is realized, and objective quantitative evaluation indexes are provided. In terms of signal optimization, by constructing a complete signal optimization equation set, multiple influencing factors such as instrument gain, channel crosstalk, and background noise are systematically considered to ensure the accuracy and reliability of the detection signal. In terms of the screening strategy, by introducing multiple detection combination optimization and repeatability verification, a scientific screening decision-making mechanism is established, which significantly improves the reliability of the screening results.

[0018] The present invention successfully solves the problems of signal stability and specificity in the screening of antibodies for flow cytometry detection, which are mainly reflected in the following aspects: First, the objective evaluation of antibody binding characteristics is realized through precise signal analysis; second, the stability of the detection results is ensured through systematic signal optimization; third, the optimal configuration of the antibody combination is guaranteed through a scientific screening strategy. These technological innovations make the screening process more standardized and provide reliable technical support for high-quality flow cytometry detection experiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flowchart of the method of the present invention.

[0020] Figure 2 is the high-performance liquid chromatography chart of the monoclonal antibody sample MAb-003 in Example 2 and the analysis result chart of its second derivative.

[0021] Figure 3 is the concentration gradient binding curve and the Langmuir fitting result chart of the two antibody samples MAb-003 and MAb-005 in Example 2. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0023] As Figure 1 shown, it is a flowchart of a method for screening antibodies for flow cytometry detection provided by the present invention, and this method includes the following steps:

[0024] S01. Analyze the purity of the monoclonal antibody to be screened. Determine the antibody purity by high-performance liquid chromatography, calculate the sample purity by the second derivative peak shape analysis method, measure the protein concentration by the peak area method, and select monoclonal antibody samples with a purity greater than 95% and a protein concentration in the range of 1 to 5 mg / ml.

[0025] S02. Prepare a concentration gradient of the monoclonal antibody sample by the equal ratio dilution method, and incubate it with target antigen-positive cells and target antigen-negative cells at a ratio of 1 to 10 μg per 1 million cells for 30 to 60 minutes at 2 to 8 °C.

[0026] S03. Calculate the binding affinity constant of the monoclonal antibody sample and the target antigen-positive cells by non-linear fitting.

[0027] S04. Analyze the binding signal of the monoclonal antibody sample and the target antigen-positive cells by using an improved Hilbert-Huang decomposition algorithm, establish a binding strength matrix of 10 to 50 dimensions, and obtain the stable binding component and the non-specific binding component.

[0028] S05. Based on the stable binding component and the non-specific binding component, calculate the binding stability coefficient matrix by using the covariance matrix decomposition method, obtain the principal component contribution rate through eigenvalue decomposition, and select monoclonal antibody samples with a principal component contribution rate greater than 0.9.

[0029] S06. Detect the cell suspension labeled with the monoclonal antibody by flow cytometry, set the forward scatter threshold and the side scatter threshold to remove cell debris, and collect 50,000 effective cell events.

[0030] S08. Construct a fluorescence intensity contribution matrix of 20 to 100 dimensions and calculate it by using a signal optimization equation set. The signal optimization equation set includes a signal gain equation, a signal compensation equation, a background correction equation, and a distribution fitting equation.

[0031] S08. Based on the fluorescence intensity contribution matrix constructed by the optimal signal distribution parameters, calculate the signal stability contribution value by using an exponential decay function, and select the signal region with a signal stability contribution value greater than 0.8 and a decay rate less than 10% as the effective detection region.

[0032] S09. Optimize the multiple detection combination of the monoclonal antibody sample in the effective detection region. Take the specificity of the monoclonal antibody sample as the value coefficient and the detection cost of the monoclonal antibody sample as the weight coefficient, and calculate the optimal detection combination scheme by using the 0-1 knapsack optimization method.

[0033] S10. Use the bootstrap resampling method to perform 3 - time repeatability verification on the optimal detection combination scheme, calculate the repeatability coefficient of variation through the analysis of variance method, and screen out monoclonal antibody samples with a repeatability coefficient of variation less than 10% and stable signal stability contribution values as antibodies for flow cytometry detection;

[0034] Among them, the signal optimization equation set includes:

[0035] The signal gain equation is used to correct the influence of the instrument gain coefficient on the fluorescence signal. The input data of the signal gain equation includes the original fluorescence signal intensity collected by the flow cytometer, the voltage gain coefficient of the flow cytometer, and the channel sensitivity parameter of the flow cytometer. The output data of the signal gain equation is the corrected true fluorescence intensity value;

[0036] The signal compensation equation is used to eliminate spectral overlap between different fluorescence channels. The input data of the signal compensation equation includes the corrected true fluorescence intensity value, the inter - channel crosstalk coefficient matrix of the flow cytometer, and the single - stain control data of the monoclonal antibody sample. The output data of the signal compensation equation is the pure fluorescence signal after removing crosstalk;

[0037] The background correction equation is used to deduct the sample background and non - specific fluorescence. The input data of the background correction equation includes the pure fluorescence signal after removing crosstalk, the control signal value of target - antigen - negative cells, and the spontaneous fluorescence signal intensity of target - antigen - negative cells. The output data of the background correction equation is the specific fluorescence signal after deducting the background;

[0038] The distribution fitting equation is used to fit the fluorescence signal distribution characteristics. The input data of the distribution fitting equation includes the specific fluorescence signal after deducting the background, the number distribution of effective cell events, the forward scatter threshold, and the side scatter threshold. The output data of the distribution fitting equation is the optimal signal distribution parameter.

[0039] The specific implementation manners of the above steps are described in detail below. The specific implementation manner of step S01 is to perform antibody purity analysis based on high performance liquid chromatography. The monoclonal antibody sample to be screened is separated using a reverse-phase chromatographic column. The absorbance signal of the protein is detected by an ultraviolet detector at a wavelength of 214 nm, the chromatogram data is collected, the sample purity is calculated using the second derivative peak shape analysis method, and the protein concentration is determined by the peak area method. The chromatogram is subjected to baseline correction and noise filtering, the Savitzky-Golay smoothing algorithm is used to smooth the original chromatographic data, the second derivative curve is calculated, the main peak and impurity peaks are identified and quantified, the sample purity is calculated based on the peak area ratio, and the monoclonal antibody sample with a purity greater than 95% and a protein concentration in the range of 1 to 5 mg / mL is selected. The purpose of this step is to ensure that the monoclonal antibody sample used for screening has a high purity and an appropriate concentration range, providing a reliable sample basis for subsequent binding force analysis. Among them, the reference value of the sample purity is 95% to 99.9%, and the reference value of the protein concentration is 1.5 to 4.5 mg / mL.

[0040] The specific implementation manner of step S02 is to prepare a concentration gradient using the equal ratio dilution method. First, the monoclonal antibody sample is diluted with phosphate buffer to an initial concentration of 10 μg per 1 million cells, and then diluted successively at a ratio of 1:2 to prepare 6 concentration gradient points, with the lowest concentration being 1 μg per 1 million cells. The prepared monoclonal antibody samples with different concentrations are incubated with target antigen-positive cells and target antigen-negative cells at 2 to 8°C for 30 to 60 minutes. A phosphate buffer containing 0.1% bovine serum albumin is added to the incubation system as a blocking agent to prevent non-specific binding. The purpose of this step is to investigate the binding characteristics of the monoclonal antibody to the cell surface target antigen through a concentration gradient experiment, providing experimental data for subsequent affinity analysis. Among them, the reference value of the antibody concentration gradient is 10, 5, 2.5, 1.25, and 1 μg per 1 million cells, the reference value of the incubation temperature is 4°C, and the reference value of the incubation time is 45 minutes.

[0041] The specific implementation manner of step S03 is to calculate the binding affinity constant of the monoclonal antibody sample to the target antigen-positive cells based on non-linear fitting. The Langmuir isothermal adsorption model is used to describe the binding equilibrium between the antibody and the antigen, and the Levenberg-Marquardt algorithm is used to perform non-linear fitting on the experimental data to obtain the binding affinity constant. First, the binding signal values at different concentrations are normalized to remove the influence of the background signal, and then the Langmuir isothermal adsorption equation is established, and the least squares method is used to optimize the fitting parameters to calculate the binding affinity constant. The purpose of this step is to quantitatively evaluate the binding strength between the monoclonal antibody and the target antigen, providing a basis for screening high-affinity antibodies. Among them, the reference value of the binding affinity constant is 10 -7 to 10 -10 mol / L.

[0042] The specific implementation of step S04 is to analyze the binding signal between the monoclonal antibody sample and the target antigen-positive cells using an improved Hilbert-Huang transform algorithm. First, perform a Hilbert transform on the original binding signal to obtain the instantaneous frequency and instantaneous amplitude of the signal. Then, use the empirical mode decomposition method to decompose the signal into multiple intrinsic mode functions and a residual function. By screening the frequency characteristics of the functions, identify the stable binding component and the non-specific binding component. Conduct a spectral analysis on the decomposed intrinsic mode functions to establish a binding strength matrix. The matrix dimension is selected from 10 to 50 dimensions according to the complexity of the signal. The purpose of this step is to separate the true antigen-antibody binding signal and the non-specific background signal from the complex binding signal, providing data support for evaluating the specificity of the antibody. Among them, the reference value for the dimension of the binding strength matrix is 20 to 30 dimensions.

[0043] The specific implementation of step S05 is to calculate the binding stability coefficient matrix based on the stable binding component and the non-specific binding component using the covariance matrix decomposition method. First, construct a covariance matrix for the stable binding component and the non-specific binding component, and obtain the eigenvectors and eigenvalues through singular value decomposition, and calculate the contribution rate of the principal components. Arrange the eigenvalues in descending order, accumulate the contribution rates of the eigenvalues, and when the cumulative contribution rate reaches more than 0.9, select the corresponding eigenvectors to construct the binding stability coefficient matrix. The purpose of this step is to evaluate the binding stability of the antibody through the principal component analysis method and screen out the monoclonal antibody samples with good binding stability. Among them, the reference value for the contribution rate of the principal components is 0.9 to 0.95, and the reference value for the eigenvalue threshold is 0.1.

[0044] The specific implementation of step S06 is to detect the cell suspension labeled with the monoclonal antibody using a flow cytometer. First, adjust the cell suspension to a concentration of 1 million per milliliter, filter out cell clumps through a 200-mesh cell sieve, and set the forward scatter threshold and side scatter threshold on the flow cytometer to remove cell debris. The forward scatter signal reflects the cell size, and the side scatter signal reflects the internal granularity of the cell. According to the scattering characteristics of normal cells, set the forward scatter threshold to 5000 to 50000, and the side scatter threshold to 2000 to 20000. Collect 50000 effective cell events at a collection rate of 200 to 500 cells per second, and monitor the stability of the sample flow rate and laser power in real time during the collection process. The purpose of this step is to obtain high-quality flow cytometry detection data, providing a reliable experimental basis for subsequent signal analysis. Among them, the reference value for the forward scatter threshold is 10000, the reference value for the side scatter threshold is 5000, and the reference value for the collection rate is 300 cells per second.

[0045] The specific implementation of step S07 is to construct a fluorescence intensity contribution matrix and perform calculations using signal optimization equations. First, based on the number of channels and signal characteristics of the flow cytometer, a fluorescence intensity contribution matrix with 20 to 100 dimensions is established. The signal gain equation uses a logarithmic conversion model to combine the original fluorescence signal intensity with the voltage gain coefficient and channel sensitivity parameters to calculate the corrected true fluorescence intensity value. The signal compensation equation is based on the spectral overlap matrix, calculates the crosstalk coefficient between channels using single-stained control data, and uses linear algebra methods to eliminate the crosstalk signals between different fluorescence channels. The background correction equation uses a non-parametric estimation method to calculate and subtract the background fluorescence of the sample through the control signals and autofluorescence signals of antigen-negative cells. The distribution fitting equation uses a mixture Gaussian model to perform probability density estimation on the specific fluorescence signals after background subtraction to obtain the optimal signal distribution parameters. The purpose of this step is to comprehensively optimize the flow detection signals and improve the accuracy and reliability of detection. Among them, the reference value for the dimension of the fluorescence intensity contribution matrix is 50 dimensions, the reference value for the signal compensation coefficient is 0.01 to 0.1, and the reference value for the background fluorescence threshold is 100 to 1000.

[0046] The specific implementation of step S08 is to calculate the signal stability contribution value using an exponential decay function based on the fluorescence intensity contribution matrix constructed with the optimal signal distribution parameters. First, the fluorescence intensity contribution matrix is sorted according to the signal intensity, an exponential decay model is established to describe the decay characteristics of the signal over time, and the decay coefficient is determined by the maximum likelihood estimation method. Calculate the stability contribution value of each signal region, and select the region with a stability contribution value greater than 0.8 and a decay rate less than 10% as the effective detection region. The purpose of this step is to determine the signal region most suitable for flow detection and ensure the stability and repeatability of the detection results. Among them, the reference range for the signal stability contribution value is 0.8 to 1.0, the reference value for the signal decay rate is 5% to 10%, and the reference value for the signal intensity range of the effective detection region is 1000 to 100000.

[0047] The specific implementation of step S09 is to optimize the multiple detection combinations of monoclonal antibody samples within the effective detection region. First, use the specificity of the monoclonal antibody sample as the value coefficient and the detection cost as the weight coefficient to establish a 0-1 knapsack optimization model. Use the dynamic programming algorithm to solve the optimal detection combination scheme, and select the combination scheme with the maximum value under the cost constraint by iteratively calculating the total value and total cost of different combination schemes. The purpose of this step is to achieve the optimal combination of multiple detections under limited detection costs and improve detection efficiency and economy. Among them, the reference range for the value coefficient is 0 to 1, the reference range for the weight coefficient is 1 to 10, and the reference value for the number of combination schemes is 3 to 8.

[0048] The specific implementation of step S10 is to use the bootstrap resampling method to verify the repeatability of the optimal detection combination scheme. First, samples are randomly drawn from the original data to construct a resampling data set, and the experiment is repeated 3 times. The repeatability coefficient of variation is calculated by the analysis of variance method. The contribution values of signal stability for each repeated experiment are statistically analyzed to evaluate the dispersion and stability of the data. Monoclonal antibody samples with a repeatability coefficient of variation less than 10% and stable signal stability contribution values are selected as antibodies for flow cytometry detection. The purpose of this step is to verify the reliability of the screening results through statistical methods and ensure that the selected antibodies have good repeatability and stability. Among them, the reference value of the repeatability coefficient of variation is 5% to 10%, and the reference value of the standard deviation of the signal stability contribution value is less than 0.05.

[0049] The following details the calculation formulas or equations involved in the specific implementation of the present invention.

[0050] The combined affinity constant calculation formula is specifically expressed as follows:

[0051]

[0052] In the formula, K a is the combined affinity constant; k on is the binding rate constant; k off is the dissociation rate constant; [AB] is the concentration of the antigen-antibody complex; [A] is the concentration of free antibody; [B] is the concentration of free antigen. The combined affinity constant calculation formula adopts the form of a concentration ratio, reflecting the quantitative relationship between the antigen-antibody complex and the free components in the equilibrium state; among them, the ratio form of k on and k off embodies the dynamic equilibrium characteristics of the binding process, and the ratio of the binding rate to the dissociation rate determines the final affinity; this equation takes into account the concentration changes of each component in the reaction system and can be directly measured through experiments.

[0053] The Langmuir isothermal adsorption equation is specifically expressed as follows:

[0054]

[0055] In the formula, Y is the measured binding signal value; B max is the maximum number of binding sites; X is the antibody concentration; K d is the dissociation constant; ε is the measurement error. The Langmuir isothermal adsorption equation is based on the principle of reversible binding kinetics and uses a hyperbolic function to describe the antigen-antibody binding process; among them, the molecular term B max X represents the degree to which the binding sites are occupied, and the denominator term K d+X represents the competitive effect, and the introduction of the error term ε takes into account the random error in the measurement process; this equation is applicable to the single binding site model, and the binding parameters can be obtained through non-linear fitting.

[0056] The calculation formula of the Hilbert transform is specifically expressed as follows:

[0057]

[0058] In the formula, x(t) is the original signal; H[x(t)] is the signal after the Hilbert transform; t is the time variable; τ is the integration variable. The Hilbert transform adopts the Cauchy principal value integral form to obtain the analytical representation of the signal; the instantaneous phase and instantaneous frequency information of the signal can be obtained after the transformation, and the integral form ensures the reversibility of the transformation; this equation realizes the time-frequency analysis of non-stationary signals through the conversion between the time domain and the frequency domain.

[0059] The mathematical expression of the intrinsic mode function decomposition is specifically expressed as follows:

[0060]

[0061] In the formula, x(t) is the original signal; c i (t) is the i-th intrinsic mode function; r n (t) is the residual function; n is the decomposition level. The intrinsic mode function decomposition adopts a recursive screening method to decompose complex signals into oscillation components of different scales; among them, each intrinsic mode function represents the signal characteristics in a specific frequency range, and the residual function reflects the overall trend of the signal; this equation can adaptively separate meaningful signal components based on the local characteristic scales of the signal.

[0062] The binding strength matrix is specifically expressed as follows:

[0063]

[0064] In the formula, M is the binding strength matrix; m ij represents the binding strength value of the i-th sample on the j-th feature dimension; p is the number of samples; n is the number of feature dimensions. The binding strength matrix adopts a two-dimensional array form to organize and store multi-dimensional feature data; the matrix element m ij represents the measured value of the sample on different feature dimensions, and the number of rows and columns of the matrix reflects the scale and dimension of the data; this square matrix structure facilitates subsequent numerical calculations and feature extraction.

[0065] The covariance matrix decomposition formula is specifically expressed as follows:

[0066]

[0067] In the formula, C is the covariance matrix; X is the original data matrix; is the mean matrix; U is the left singular matrix; ∑ is the diagonal matrix of singular values; V is the right singular matrix. The covariance matrix decomposition adopts the form of singular value decomposition, realizing the dimensionality reduction and feature extraction of data; among them, the left and right singular matrices contain the direction information of the principal components, and the singular values reflect the importance of each principal component; this equation transforms the original data into a new feature space, which is beneficial to extracting the main variation patterns of the data.

[0068] The calculation formula for the contribution rate of the principal component is specifically expressed as follows:

[0069]

[0070] In the formula, η k is the contribution rate of the k-th principal component; λ k is the k-th eigenvalue; n is the total number of eigenvalues. The calculation of the contribution rate of the principal component adopts the form of eigenvalue ratio, quantifying the relative importance of each principal component; the numerator represents the size of a single eigenvalue, the denominator represents the sum of all eigenvalues, and the ratio reflects the degree to which this principal component explains the variation of the original data; this equation is used to determine the optimal number of features and achieve effective dimensionality reduction of the data.

[0071] The calculation formula of the signal gain equation is specifically expressed as follows:

[0072] F corrected = F raw ·e -αV ·S + β;

[0073] In the formula, F corrected is the corrected true fluorescence intensity value; F raw is the original fluorescence signal intensity; V is the voltage gain coefficient; α is the gain correction coefficient; S is the channel sensitivity parameter; β is the baseline drift correction term. The signal gain equation adopts the form of an exponential function, describing the non-linear effect of voltage gain on the fluorescence signal; among them, the exponential term e -αV represents the gain effect, the linear term S represents the channel sensitivity, and the constant term β represents the baseline drift; this equation considers various physical effects in the signal amplification process and realizes accurate signal correction.

[0074] The matrix expression of the signal compensation equation is specifically expressed as follows:

[0075]

[0076] In the formula, is the fluorescence signal after compensation for the i-th channel; F i is the original fluorescence signal of the i-th channel; s ijis the crosstalk coefficient of channel i to channel j; n is the number of fluorescence channels. The signal compensation equation adopts the form of a linear transformation matrix to process the crosstalk between multi-channel fluorescence signals; the diagonal elements of the matrix being 1 represent the signal itself, and the non-diagonal elements represent the crosstalk coefficients between channels; this equation is based on the principle of spectral overlap and realizes the separation of multi-channel signals.

[0077] The background correction equation is specifically expressed as follows:

[0078] F specific = F compensated -(F control + k·F auto ) + δ;

[0079] In the formula, F specific is the specific fluorescence signal after background subtraction; F compensated is the compensated fluorescence signal; F control is the negative control signal value; F auto is the autofluorescence signal intensity; k is the autofluorescence correction coefficient; δ is the signal offset. The background correction equation adopts a subtraction form to eliminate the influence of non-specific fluorescence and autofluorescence; among them, the negative control term and the autofluorescence term respectively consider background signals from different sources, and the correction coefficient k adjusts the contribution degree of autofluorescence; this equation ensures the accurate extraction of specific fluorescence signals.

[0080] The distribution fitting equation is specifically expressed as follows:

[0081]

[0082] In the formula, P(x) is the probability density function of the fluorescence signal; w i is the weight of the i-th Gaussian component; μ i is the mean of the i-th Gaussian component; σ i is the standard deviation of the i-th Gaussian component; m is the number of Gaussian components; γ is the baseline correction term. The distribution fitting equation adopts a mixture Gaussian model to describe the probability distribution characteristics of the fluorescence signal; each Gaussian component represents a sub-population, the weight coefficient reflects the proportion of each sub-population, and the baseline correction term considers the signal offset; this equation is applicable to the analysis of the signal characteristics of multi-peak distributions.

[0083] The derivation process of the distribution fitting equation is as follows:

[0084] 1. First, assume that the fluorescence signal is composed of multiple sub-populations, and each sub-population follows a Gaussian distribution:

[0085]

[0086] 2. Consider the weight contribution of each sub-population:

[0087]

[0088] where n i is the number of cells in the i-th subpopulation;

[0089] 3. Introduce the baseline correction term γ to obtain the final mixture distribution model:

[0090]

[0091] The exponential decay function is specifically expressed as follows:

[0092] S(t) = S0e -λt + Asin(ωt + φ) + η;

[0093] where S(t) is the signal intensity; S0 is the initial signal intensity; λ is the decay coefficient; t is the time variable; A is the amplitude of periodic fluctuation; ω is the angular frequency; φ is the phase; η is the random noise term. The exponential decay function includes a decay term and a periodic term, describing the variation law of the signal intensity with time; among them, the exponential term describes the overall decay trend of the signal, the sine term describes the periodic fluctuation of the signal, and the noise term takes into account the random perturbation; this equation comprehensively considers various dynamic characteristics of the signal.

[0094] The construction process of the exponential decay function is as follows:

[0095] 1. Basic decay term: S0e -λt , describing the natural decay of the signal intensity with time;

[0096] 2. Periodic fluctuation term: Asin(ωt + φ), describing the periodic perturbation of the system;

[0097] 3. Random noise term: η ~ N(0, σ 2 ), describing the random error in the measurement process;

[0098] 4. Combine the above terms to obtain the complete model:

[0099] S(t) = S0e -λt + Asin(ωt + φ) + η.

[0100] The mathematical model of the knapsack optimization problem is specifically expressed as follows:

[0101]

[0102] x i ∈ {0, 1}, i = 1, 2,..., n;

[0103] where v i is the value coefficient of the i-th antibody sample; w i is the weight coefficient of the i-th antibody sample; xi is a decision variable; W is the total weight constraint; n is the number of antibody samples. The knapsack optimization model adopts the 0-1 integer programming form to solve the optimal detection combination scheme; the objective function maximizes the total value, the constraint condition limits the total weight, and the decision variable indicates whether to select a certain sample; this model balances the detection effect and cost constraints.

[0104] The solution process of the knapsack optimization problem uses the dynamic programming method, and the specific steps are as follows:

[0105] 1. Construct the state transition matrix:

[0106]

[0107] In the formula, dp ij represents the maximum value when considering the first i items and the weight does not exceed j;

[0108] 2. State transition equation:

[0109]

[0110] 3. Backtracking of the optimal solution:

[0111]

[0112] The calculation formula of the repeatability coefficient of variation is specifically expressed as follows:

[0113]

[0114] In the formula, CV is the coefficient of variation; x i is the i-th measurement value; is the average value; N is the number of repeated measurements. The repeatability coefficient of variation adopts the form of the ratio of the standard deviation to the mean value to evaluate the dispersion degree of the data; among them, the standard deviation reflects the fluctuation range of the data, and the mean value reflects the central tendency of the data; this equation provides a dimensionless measure of the data repeatability.

[0115] The calculation formula of the bootstrap resampling statistic is specifically expressed as follows:

[0116]

[0117] In the formula, is the bootstrap estimator; is the statistic obtained from the i-th resampling; B is the number of resamplings; SE BIt is the bootstrap standard error. The bootstrap resampling statistics are in the form of mean and standard deviation to evaluate the uncertainty of parameter estimation; the resampled mean reflects the expected value of the parameter, and the standard error reflects the estimation precision; this equation does not rely on the data distribution assumption and provides robust statistical inference.

[0118] The calculation formula for the second derivative peak shape analysis is specifically expressed as follows:

[0119]

[0120] In the formula, f″(t) is the second derivative value; f(t) is the chromatographic signal intensity; t is the retention time; h is the time interval.

[0121] The derivation process of the second derivative peak shape analysis equation is as follows:

[0122] 1. First, calculate the first derivative of the chromatographic curve:

[0123]

[0124] 2. Then, take the derivative of the first derivative to obtain the second derivative:

[0125]

[0126] 3. Substitute the first derivative expression and simplify to obtain the final form:

[0127]

[0128] In this equation, the time interval h is determined by the sampling frequency of the chromatograph and is usually set to 0.1 second; the signal intensity f(t) is directly measured by the detector of the chromatograph, and the unit is millivolt.

[0129] The calculation formula for the chromatographic peak area is specifically expressed as follows:

[0130]

[0131] In the formula, A is the peak area; f(t) is the chromatographic signal intensity; t1 and t2 are the start and end times of integration; Δt is the sampling interval.

[0132] The chromatographic peak area is calculated using the trapezoidal integration method, and its derivation process is as follows:

[0133] 1. Theoretically, the peak area is the integral of the signal intensity with respect to time:

[0134]

[0135] 2. Considering that the actual sampling is discrete points, the trapezoidal rule is used for approximation:

[0136]

[0137] In the formula, the integration interval [t1, t2] is determined by the starting point and the ending point of the peak, and usually the position where the signal intensity exceeds three times the baseline noise is selected.

[0138] The separation equation of the stable binding component is specifically expressed as follows:

[0139] S total = S stable + S nonspecific + ξ;

[0140]

[0141] In the formula, S total is the total signal; S stable is the stable binding component; S nonspecific is the non-specific binding component; IMF i (t) is the i-th intrinsic mode function; r n (t) is the residual function; k is the component separation threshold; ξ is the noise term.

[0142] The signal component separation adopts the empirical mode decomposition method, and the specific steps are as follows:

[0143] 1. Construct the signal component vector:

[0144]

[0145] 2. Construct the intrinsic mode function matrix:

[0146]

[0147] 3. Establish the component separation equation:

[0148] S total = S stable + S nonspecific + ξ;

[0149] In the formula, the intrinsic mode function IMF i (t) is obtained through the sifting process, and the component separation threshold k is determined based on the frequency characteristics, and usually the position where the cumulative energy ratio reaches 90% is selected.

[0150] The mathematical model for repeatability analysis includes:

[0151] 1. Construct the measurement value matrix:

[0152]

[0153] In the formula, x ij is the j-th measurement value of the i-th sample;

[0154] 2. Calculate the mean vector:

[0155]

[0156] Where:

[0157] 3. Calculate the standard deviation vector:

[0158]

[0159] Where:

[0160] 4. Calculate the coefficient of variation:

[0161]

[0162] The implementation process of bootstrap resampling is as follows:

[0163] 1. Construct the original data matrix:

[0164]

[0165] 2. Resampling process: Sample with replacement from the original data to obtain a sample Calculate the statistic Repeat B times to obtain a sequence of statistics

[0166] 3. Calculate the bootstrap estimate:

[0167]

[0168] 4. Calculate the standard error:

[0169]

[0170] Specifically, the principle of the present invention is: The core principle of the present invention is to establish a complete set of signal analysis and optimization systems. First, in terms of antibody binding characteristic analysis, an improved Hilbert-Huang decomposition algorithm is used to decompose complex binding signals. This algorithm replaces the traditional envelope calculation with wavelet threshold denoising and introduces an adaptive weight matrix to achieve precise separation of signal components. This improvement overcomes the endpoint effect and mode mixing problems existing in traditional empirical mode decomposition and improves the accuracy of signal analysis.

[0171] In terms of signal optimization, the present invention constructs a complete set of signal optimization equations. The signal gain equation achieves precise correction of the original fluorescence signal by considering the voltage gain coefficient and channel sensitivity of the instrument. The signal compensation equation eliminates the spectral overlap effect in multiplex staining by analyzing the crosstalk relationship between different fluorescence channels. The background correction equation effectively removes non-specific fluorescence interference by introducing negative control signals. The distribution fitting equation ensures the accuracy of the detection results by analyzing the signal distribution characteristics of the effective cell population.

[0172] In terms of the screening strategy, the present invention adopts a multi-level optimization method. The stability of antibody binding is evaluated by covariance matrix decomposition, the optimal configuration of multiplex detection combinations is achieved by the 0-1 knapsack optimization algorithm, and the repeatability of the screening results is ensured by bootstrap resampling verification. This multi-level optimization strategy ensures the scientificity and reliability of the screening process.

[0173] A specific Example 1 of the present invention is provided below. The specific implementation of each step in this Example 1 is described in detail as follows.

[0174] The specific implementation of step S01 is to perform antibody purity analysis based on high performance liquid chromatography. A reverse phase chromatographic column is used for separation, with a methanol-water system as the mobile phase, the flow rate is controlled within the range of 0.5 to 2 milliliters per minute, the column temperature is set at 25 to 40 degrees Celsius, the absorbance signal of the protein is detected at a wavelength of 214 nanometers by an ultraviolet detector, chromatogram data is collected, and the chromatogram is recorded at a frequency of 4 to 10 sampling points per second. The chromatographic data is converted into digital signals through a data acquisition system for analysis. First, baseline correction and noise filtering are performed on the chromatogram. The Savitzky-Golay smoothing algorithm is used to smooth the original chromatographic data, with the window width set at 5 to 15 data points, and high-frequency noise is eliminated by moving average to improve the signal-to-noise ratio. Then, second derivative analysis is performed on the smoothed chromatogram, and the formula is used to calculate the second derivative value, where f″(t) is the second derivative value, f(t) is the chromatographic signal intensity, t is the retention time, and h is the time interval. Overlapping peaks and shoulder peaks can be accurately identified through the second derivative peak shape. Then, the peak area method is used to calculate the content of each component, using the formula Calculate the peak area, where A is the peak area, f(t) is the chromatographic signal intensity, t1 and t2 are the start and end times of integration, and Δt is the sampling interval. Calculate the sample purity by the ratio of the main peak area to the total peak area, and select monoclonal antibody samples with a purity greater than 95% and a protein concentration in the range of 1 to 5 mg / mL. The purpose of this step is to ensure that the monoclonal antibody samples used for screening have high purity and an appropriate concentration range, providing a reliable sample basis for subsequent binding analysis. Among them, the reference value of sample purity is 95% to 99.9%, and the reference value of protein concentration is 1.5 to 4.5 mg / mL.

[0175] The specific implementation of step S02 is to prepare a concentration gradient by the method of serial dilution. First, dilute the monoclonal antibody sample with phosphate buffer to an initial concentration of 10 μg per 1 million cells, and then sequentially dilute it at a ratio of 1:2 to prepare 6 concentration gradient points, with the lowest concentration being 1 μg per 1 million cells. Using the antibody titration experimental method of flow cytometry, incubate the prepared monoclonal antibody samples with different concentrations with target antigen-positive cells and target antigen-negative cells at 2 to 8 °C for 30 to 60 minutes. Add phosphate buffer containing 0.1% bovine serum albumin as a blocking agent to the incubation system to prevent non-specific binding. After incubation, use the centrifugal washing method to remove the unbound antibody. The centrifugal speed for each time is 300 to 500 g, and the time is 5 minutes. Repeat the washing 2 to 3 times. After washing, resuspend the cells in phosphate buffer containing 0.1% sodium azide and adjust the cell concentration to 1 million per milliliter. Use flow cytometry to detect the cell samples labeled with antibodies at different concentrations, collect 10,000 to 50,000 effective cell events, and the collected data is used for subsequent binding analysis. The purpose of this step is to investigate the binding characteristics of monoclonal antibodies to cell surface target antigens through concentration gradient experiments and provide experimental data for subsequent affinity analysis. Among them, the reference value of the antibody concentration gradient is 10, 5, 2.5, 1.25, and 1 μg per 1 million cells, the reference value of the incubation temperature is 4 °C, and the reference value of the incubation time is 45 minutes.

[0176] The specific implementation of step S03 is to calculate the binding affinity constant of the monoclonal antibody sample to target antigen-positive cells based on nonlinear fitting, and use the Langmuir isothermal adsorption model to describe the binding equilibrium between the antibody and the antigen. First, calculate the binding affinity constant according to the formula where K a is the binding affinity constant, k on is the binding rate constant, k off is the dissociation rate constant, [AB] is the concentration of the antigen-antibody complex, [A] is the concentration of free antibody, and [B] is the concentration of free antigen. Then use the Langmuir isothermal adsorption equation Perform non - linear fitting, where Y is the measured binding signal value, B max is the maximum number of binding sites, X is the antibody concentration, K d is the dissociation constant, and ε is the measurement error. Non - linear fitting of the experimental data is performed using the Levenberg - Marquardt algorithm, iteratively optimizing the fitting parameters to minimize the sum of squared residuals to obtain the binding affinity constant. First, the binding signal values at different concentrations are normalized to remove the influence of the background signal, and then the Langmuir isothermal adsorption equation is established. The least - squares method is used to optimize the fitting parameters to calculate the binding affinity constant. The purpose of this step is to quantitatively evaluate the binding strength between the monoclonal antibody and the target antigen, providing a basis for screening high - affinity antibodies. Among them, the reference value of the binding affinity constant is 10 -7 to 10 -10 moles per liter.

[0177] The specific implementation of step S04 is to analyze the binding signal between the monoclonal antibody sample and the target antigen - positive cells using an improved Hilbert - Huang transform algorithm. First, perform the Hilbert transform according to the formula where x(t) is the original signal, H[x(t)] is the signal after Hilbert transform, t is the time variable, and τ is the integration variable. Then, use the empirical mode decomposition method to decompose the signal into multiple intrinsic mode functions and a residual function according to the formula where x(t) is the original signal, c i (t) is the i - th intrinsic mode function, r n (t) is the residual function, and n is the decomposition level. By screening the frequency characteristics of the function, identify the stable binding component and the non - specific binding component. Perform spectral analysis on the decomposed intrinsic mode functions to establish the binding strength matrix where M is the binding strength matrix, m ij represents the binding strength value of the i - th sample in the j - th characteristic dimension, p is the number of samples, and n is the number of characteristic dimensions. The purpose of this step is to separate the true antigen - antibody binding signal and the non - specific background signal from the complex binding signal, providing data support for evaluating the specificity of the antibody. Among them, the reference value of the dimension of the binding strength matrix is 20 to 30 dimensions.

[0178] The specific implementation of step S05 is to calculate the binding stability coefficient matrix based on the stable binding component and the non - specific binding component using the covariance matrix decomposition method. First, calculate the covariance matrix according to the formula where C is the covariance matrix, X is the original data matrix, is the mean matrix, U is the left singular matrix, ∑ is the diagonal matrix of singular values, and V is the right singular matrix. Then, use the principal component analysis method to calculate the principal component contribution rate through the formula where η kis the contribution rate of the k-th principal component, λ k is the k-th eigenvalue, and n is the total number of eigenvalues. The eigenvalues are arranged in descending order, and the contribution rates of the eigenvalues are accumulated. When the cumulative contribution rate reaches more than 0.9, the corresponding eigenvectors are selected to construct the combined stability coefficient matrix. The purpose of this step is to evaluate the stability of antibody binding by the principal component analysis method and screen out monoclonal antibody samples with good binding stability. Among them, the reference value of the principal component contribution rate is 0.9 to 0.95, and the reference value of the eigenvalue threshold is 0.1.

[0179] The specific implementation of step S06 is to detect the cell suspension labeled with monoclonal antibody by flow cytometry, enrich the living cells by density gradient centrifugation of the cell suspension, adjust the cell concentration to 1 million per milliliter, and filter out cell clumps through a 200-mesh cell sieve. Set the cell population acquisition area on the flow cytometry according to the characteristics of the forward scatter signal and the side scatter signal. The forward scatter signal reflects the cell size, and the side scatter signal reflects the internal granularity of the cell. Based on the scatter characteristics of normal cells, set the forward scatter threshold to 5000 to 50000, and the side scatter threshold to 2000 to 20000. Perform quality control calibration on the detection system, and use fluorescent microspheres to adjust the laser power and the photomultiplier tube voltage to ensure the linear range and resolution of the detection system. Collect 50,000 effective cell events at a collection rate of 200 to 500 cells per second. During the collection process, monitor the stability of the sample flow rate and the laser power in real time to ensure the data quality. The purpose of this step is to obtain high-quality flow cytometry detection data and provide a reliable experimental basis for subsequent signal analysis. Among them, the reference value of the forward scatter threshold is 10000, the reference value of the side scatter threshold is 5000, and the reference value of the collection rate is 300 cells per second.

[0180] The specific implementation of step S07 is to construct a fluorescence intensity contribution matrix and perform calculations using the signal optimization equations. According to the number of channels and signal characteristics of the flow cytometry, establish a fluorescence intensity contribution matrix of 20 to 100 dimensions. First, use the signal gain equation F corrected =F raw ·e -αv ·S + β, where F corrected is the corrected true fluorescence intensity value, F raw is the original fluorescence signal intensity, V is the voltage gain coefficient, α is the gain correction coefficient, S is the channel sensitivity parameter, and β is the baseline drift correction term. Then use the signal compensation equation

[0181] where is the fluorescence signal after compensation for the i-th channel, F i is the original fluorescence signal of the i-th channel, s ijis the crosstalk coefficient of channel i to channel j, and n is the number of fluorescence channels. Then, the background correction equation F specific = F compensated - (F control + k·F auto ) + δ is used, where F specific is the specific fluorescence signal after background subtraction, F compensated is the compensated fluorescence signal, F control is the negative control signal value, F auto is the autofluorescence signal intensity, k is the autofluorescence correction coefficient, and δ is the signal offset. Finally, the distribution fitting equation is used, where P(x) is the probability density function of the fluorescence signal, w i is the weight of the i-th Gaussian component, μ i is the mean of the i-th Gaussian component, σ i is the standard deviation of the i-th Gaussian component, m is the number of Gaussian components, and γ is the baseline correction term. The purpose of this step is to comprehensively optimize the flow cytometry detection signal and improve the accuracy and reliability of the detection. Among them, the reference value of the dimension of the fluorescence intensity contribution matrix is 50 dimensions, the reference value of the signal compensation coefficient is from 0.01 to 0.1, and the reference value of the background fluorescence threshold is from 100 to 1000.

[0182] The specific implementation of step S08 is to construct a fluorescence intensity contribution matrix based on the optimal signal distribution parameters and calculate the signal stability contribution value using an exponential decay function. The exponential decay function S(t) = S0e -λt + Asin(ωt + φ) + η is used, where S(t) is the signal intensity, S0 is the initial signal intensity, λ is the decay coefficient, t is the time variable, A is the amplitude of the periodic fluctuation, ω is the angular frequency, φ is the phase, and η is the random noise term. First, the fluorescence intensity contribution matrix is sorted according to the signal intensity size, an exponential decay model is established to describe the decay characteristics of the signal over time, and the decay coefficient is determined by the maximum likelihood estimation method. The stability contribution value of each signal region is calculated, and the region with a stability contribution value greater than 0.8 and a decay rate less than 10% is selected as the effective detection region. The purpose of this step is to determine the signal region most suitable for flow cytometry detection and ensure the stability and repeatability of the detection results. Among them, the reference range of the signal stability contribution value is from 0.8 to 1.0, the reference value of the signal decay rate is from 5% to 10%, and the reference value of the signal intensity range of the effective detection region is from 1000 to 100000.

[0183] The specific implementation of step S09 is to optimize the multiplex detection combination of monoclonal antibody samples in the effective detection region. First, the specificity of the monoclonal antibody sample is used as the value coefficient and the detection cost is used as the weight coefficient to establish a knapsack optimization model where vi is the value coefficient of the i-th antibody sample, w i is the weight coefficient of the i-th antibody sample, x i is the decision variable, W is the total weight constraint, and n is the number of antibody samples. The dynamic programming algorithm is used to solve the optimal detection combination scheme. By iteratively calculating the total value and total cost of different combination schemes, the combination scheme with the maximum value under the cost constraint is selected. A state transition matrix is constructed Use the state transition equation to update the dynamic programming table. The optimal solution is determined by the backtracking method to judge whether each antibody sample is selected. The purpose of this step is to achieve the optimal combination of multiple detections under the limited detection cost, improving the detection efficiency and economy. Among them, the reference range of the value coefficient is from 0 to 1, the reference range of the weight coefficient is from 1 to 10, and the reference value of the number of combination schemes is from 3 to 8

[0184] The specific implementation of step S10 is to use the bootstrap resampling method to verify the repeatability of the optimal detection combination scheme. First, use the repeatability coefficient of variation calculation formula where CV is the coefficient of variation, x i is the i-th measured value, is the average value, and N is the number of repeated measurements. Then use the bootstrap resampling statistic calculation formula and where is the bootstrap estimator, is the statistic obtained from the i-th resampling, B is the number of resamplings, and SE B is the bootstrap standard error. Construct the original data matrix Randomly sample from it for repeated experiments. Conduct statistical analysis on the signal stability contribution value of each repeated experiment to evaluate the dispersion and stability of the data. Select monoclonal antibody samples with a repeatability coefficient of variation less than 10% and a stable signal stability contribution value as antibodies for flow cytometry detection. The purpose of this step is to verify the reliability of the screening results through statistical methods to ensure that the selected antibodies have good repeatability and stability. Among them, the reference value of the repeatability coefficient of variation is from 5% to 10%, and the standard deviation reference value of the signal stability contribution value is less than 0.05

[0185] To better understand and implement the present invention, the following provides an embodiment 2 of a specific application scenario of the present invention: During the development of a monoclonal antibody drug targeting the CD47 target by a certain research institute, it was found that traditional antibody screening methods had problems such as inaccurate specificity evaluation and incomplete binding force analysis, which affected the drug development efficiency. The research institute decided to optimize using the technical solution provided by the present invention, and the specific implementation process is as follows

[0186] First, high-performance liquid chromatography (HPLC) was used for antibody purity analysis. The researchers used Waters' Alliance HPLC system equipped with a reversed-phase C18 column. The mobile phase was a mixed solution of acetonitrile and water, and the gradient elution program was as shown in Table 1:

[0187] Table 1 HPLC gradient elution conditions table

[0188] Time (min) Mobile Phase A (Water, %) Mobile Phase B (Acetonitrile, %) Flow Rate (mL / min) 0 95 5 1.0 10 65 35 1.0 20 35 65 1.0 30 5 95 1.0 35 95 5 1.0

[0189] Figure 2 The high-performance liquid chromatogram of the monoclonal antibody sample MAb-003 and the results of its second derivative analysis are shown. The blue solid line represents the original chromatographic signal, with a main peak appearing at 15.3 minutes, showing good symmetry; the red dashed line represents the second derivative curve, which can clearly show the changes in the peak profile and helps to identify overlapping peaks and impurity peaks. The abscissa in the figure is the retention time (minutes), the left ordinate is the signal intensity (mAU×10 3 ), and the right ordinate is the second derivative value (d 2 A / dt 2 ). The chromatographic separation was carried out at a column temperature of 30 °C, the detection wavelength was set at 214 nm, and the sampling frequency was 10 Hz. The second derivative peak shape analysis method was used to calculate the sample purity, and the calculation was carried out through the formula where the time interval h was set at 0.1 s. After analysis, the purity data of the obtained monoclonal antibody samples are shown in Table 2:

[0190] Table 2 Monoclonal antibody sample purity analysis results table

[0191]

[0192] Next, concentration gradient experiments were carried out on the antibody samples that met the purity requirements. The researchers selected two samples, MAb-003 and MAb-005, prepared concentration gradients and carried out binding experiments with the Raji cell line expressing CD47. The specific concentration gradients and binding signal values are shown in Table 3:

[0193] Table 3 Antibody concentration gradient binding experiment results table

[0194] Antibody Concentration (μg / 1 million cells) MAb-003 Fluorescence Intensity (MFI) MAb-005 Fluorescence Intensity (MFI) 10.0 15680 18450 5.0 12340 14560 2.5 8920 10230 1.25 5460 6340 1.0 4120 4890

[0195] Based on the experimental data, the researchers used the Langmuir isothermal adsorption equation for non-linear fitting to calculate the binding affinity constant. The fitting results showed that the dissociation constants K d of MAb-003 and MAb-005 were 3.2×10 -8 and 2.1×10-8 mol / L

[0196] Figure 3 Shows the concentration gradient binding curves and Langmuir fitting results of two antibody samples, MAb-003 and MAb-005. The scatter points represent the experimentally measured data points, and the dashed lines represent the fitting curves of the Langmuir equation. It can be seen from the figure that MAb-005 (red) has a higher maximum binding signal and a lower dissociation constant, indicating its better binding characteristics. The abscissa in the figure is the antibody concentration (μg / 10 6 cells), and the ordinate is the mean fluorescence intensity (MFI). To further evaluate the stability of antibody binding, the researchers performed Hilbert-Huang decomposition analysis on the binding signal. Using the formula for signal transformation, and decomposing it into intrinsic mode functions according to the formula The data of the stable binding component and non-specific binding component obtained after decomposition are shown in Table 4:

[0197] Table 4 Signal Decomposition Results Table

[0198]

[0199] The researchers used a flow cytometer to detect antibody-labeled cells. Using a BD FACSCanto II flow cytometer, the forward scatter threshold was set at 15000, and the side scatter threshold was set at 8000. The acquisition rate was controlled at 250 cells per second, and a total of 40000 effective cell events were collected. The results of the flow cytometry data after compensation and correction are shown in Table 5:

[0200] Table 5 Analysis Results of Flow Cytometry Data

[0201] Detection Parameters MAb-003 Value MAb-005 Value Average Fluorescence Intensity (MFI) 12580 15240 Coefficient of Variation (%) 15.3 12.8 Positive Rate (%) 92.5 94.8 Signal Stability Contribution Value 0.85 0.89

[0202] Finally, the researchers optimized the multiplex detection combination and verified the repeatability of the selected antibodies. Using the value coefficient and weight coefficient to construct a knapsack optimization model to obtain the optimal detection combination scheme. The bootstrap resampling method was used for 3 repeated experiments, and the calculation results of the coefficient of variation are shown in Table 6:

[0203] Table 6 Results of Repeatability Verification

[0204] Replicate Experiment Number MAb-003 Coefficient of Variation (%) MAb-005 Coefficient of Variation (%) 1 6.8 5.2 2 7.2 5.5 3 6.5 4.9 Mean 6.83 5.20

[0205] Traditional monoclonal antibody screening methods mainly rely on single binding force measurement or simple flow cytometry analysis, and have the following problems: First, purity analysis usually only focuses on the main peak area ratio, ignoring the peak shape characteristics and the influence of minor impurity peaks; Second, binding force analysis only considers endpoint data and fails to reflect the dynamic characteristics of the binding process; Third, the data processing method for flow cytometry detection is too simple and fails to effectively remove the interference of non-specific signals.

[0206] In contrast, the technical solution adopted by the present invention has the following advantages: First, the sensitivity of purity analysis is improved by second derivative peak shape analysis, and impurity peaks that are difficult to detect by conventional methods can be detected; Second, the Hilbert-Huang decomposition algorithm is introduced to achieve fine analysis of binding signals, and specific binding and non-specific binding can be accurately distinguished; Third, a multi-dimensional signal optimization strategy is adopted, significantly improving the reliability of flow cytometry detection data. Experimental data show that the antibody samples screened by the method of the present invention have better specificity and binding stability, and the coefficient of variation is generally lower than 7%, while the coefficient of variation of traditional methods is usually above 15%. These improvements provide more reliable technical support for the research and development and quality control of monoclonal antibodies.

[0207] It should be noted that the detailed explanations of the variables involved in the present invention are shown in Table 7 below.

[0208] Table 7 Variable Explanation Table

[0209]

[0210]

[0211] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention.

Claims

1. An antibody screening method for flow cytometry, characterized in that, The steps include: Perform purity analysis on the monoclonal antibodies to be screened. Determine the antibody purity using high performance liquid chromatography, calculate the sample purity through second derivative peak shape analysis, measure the protein concentration according to the peak area method, and select monoclonal antibody samples with a purity greater than 95% and a protein concentration in the range of 1 to 5 mg / mL. Prepare concentration gradients of the monoclonal antibody samples using the equal ratio dilution method, and incubate them with target antigen-positive cells and target antigen-negative cells at 2 to 8 °C for 30 to 60 minutes. Calculate the binding affinity constant of the monoclonal antibody samples with the target antigen-positive cells using non-linear fitting. Analyze the binding signal of the monoclonal antibody samples with the target antigen-positive cells using the improved Hilbert-Huang decomposition algorithm, establish a binding intensity matrix, and obtain the stable binding component and non-specific binding component. Based on the stable binding component and non-specific binding component, calculate the binding stability coefficient matrix using the covariance matrix decomposition method. Detect the cell suspension labeled with the monoclonal antibody using a flow cytometer. Construct a fluorescence intensity contribution matrix and perform calculations using the signal optimization equation set. Based on the optimal signal distribution parameters, construct a fluorescence intensity contribution matrix and calculate the signal stability contribution value using an exponential decay function. Optimize the multiple detection combinations of the monoclonal antibody samples within the effective detection region. Use the bootstrap resampling method to verify the repeatability of the optimal detection combination scheme, and screen the antibodies for flow cytometry detection.

2. The antibody screening method for flow cytometry according to claim 1, wherein The specific steps for preparing concentration gradients of the monoclonal antibody samples using the equal ratio dilution method are as follows: First, dilute the monoclonal antibody samples with phosphate buffer to an initial concentration of 10 μg per 1 million cells, and then sequentially dilute them at a ratio of 1:2 to prepare 6 concentration gradient points, with the lowest concentration being 1 μg per 1 million cells.

3. The antibody screening method for flow cytometry according to claim 2, characterized in that The specific steps for analyzing the binding signal of the monoclonal antibody samples with the target antigen-positive cells using the improved Hilbert-Huang decomposition algorithm are as follows: First, perform a Hilbert transform on the original binding signal to obtain the instantaneous frequency and instantaneous amplitude of the signal, and then use the empirical mode decomposition method to decompose the signal into multiple intrinsic mode functions and a residual function. Identify the stable binding component and non-specific binding component by screening the frequency characteristics of the functions, and perform a spectral analysis on the decomposed intrinsic mode functions to establish a 10- to 50-dimensional binding intensity matrix.

4. The antibody screening method for flow cytometry according to claim 3, wherein The specific steps for detecting the cell suspension labeled with the monoclonal antibody using a flow cytometer are as follows: Adjust the cell suspension to a concentration of 1 million cells per milliliter, filter out cell clumps through a 200-mesh cell sieve, set the forward scatter threshold at 5000 to 50000 and the side scatter threshold at 2000 to 20000 on the flow cytometer, and collect 50000 effective cell events at a collection rate of 200 to 500 cells per second.

5. The antibody screening method for flow cytometry according to claim 4, wherein The signal optimization equations include a signal gain equation, a signal compensation equation, a background correction equation, and a distribution fitting equation. The signal gain equation is used to correct the influence of the instrument gain coefficient on the fluorescence signal. The signal compensation equation is used to eliminate spectral overlap between different fluorescence channels. The background correction equation is used to subtract the sample background and non-specific fluorescence. The distribution fitting equation is used to fit the distribution characteristics of the fluorescence signal.

6. The antibody screening method for flow cytometry according to claim 5, wherein The input data of the signal gain equation includes the original fluorescence signal intensity collected by the flow cytometer, the voltage gain coefficient of the flow cytometer, and the channel sensitivity parameter of the flow cytometer. The output data of the signal gain equation is the corrected true fluorescence intensity value.

7. The antibody screening method for flow cytometry according to claim 6, wherein, The input data of the signal compensation equation includes the corrected true fluorescence intensity value, the inter-channel crosstalk coefficient matrix of the flow cytometer, and the single-staining control data of the monoclonal antibody sample. The output data of the signal compensation equation is the pure fluorescence signal after removing crosstalk.

8. The antibody screening method for flow cytometry according to claim 7, wherein The input data of the background correction equation includes the pure fluorescence signal after removing crosstalk, the control signal value of the target antigen-negative cells, and the spontaneous fluorescence signal intensity of the target antigen-negative cells. The output data of the background correction equation is the specific fluorescence signal after subtracting the background.

9. The antibody screening method for flow cytometry according to claim 8, wherein The input data of the distribution fitting equation includes the specific fluorescence signal after subtracting the background, the number distribution of effective cell events, the forward scatter threshold, and the side scatter threshold. The output data of the distribution fitting equation is the optimal signal distribution parameter.

10. The antibody screening method for flow cytometry according to claim 9, characterized in that, The specific steps for optimizing the multiplex detection combination of the monoclonal antibody sample within the effective detection region are as follows: taking the specificity of the monoclonal antibody sample as the value coefficient and the detection cost of the monoclonal antibody sample as the weight coefficient, and using the 0-1 knapsack optimization method to calculate the optimal detection combination scheme.

Images