A method and system for analyzing genetic testing data based on fluorescence spectroscopy

Abstract

The present invention discloses a method and system for analyzing genetic detection data based on fluorescence spectroscopy, which relates to the technical field of fluorescence spectrum analysis. The analysis method includes: obtaining a sample; generating a multi-channel fluorescence spectrum analysis strategy using a fluorescence multi-channel matching analysis model and a genetic algorithm; extracting a multi-channel fluorescence spectrum based on the generated strategy; processing the spectrum using a fluorescence spectrum gene analysis model to obtain a genetic detection result; and performing subsequent genetic detection analysis based on the analysis result. The present invention realizes high-precision and high-efficiency genetic detection through an innovative multi-channel matching analysis model, an improved genetic algorithm, and an efficient feature extraction and analysis module. The denoising layer based on dilated convolution and a multi-layer feature extraction structure realizes high-precision fluorescence signal denoising and feature extraction, generating accurate genetic detection analysis results.

Description

Technical Field

[0001] The present invention relates to the technical field of fluorescence spectrum analysis, and in particular to a method and system for analyzing gene detection data based on fluorescence spectrum. Background Art

[0002] Fluorescence spectroscopy-based genetic testing data analysis technologies primarily utilize changes in the intensity and wavelength of fluorescent signals to detect target genes. Common techniques include fluorescently labeled probes and fluorescence resonance energy transfer. These technologies are rapid, sensitive, and low-cost, and are widely used in pathogen detection and gene mutation analysis. However, existing technologies also have some drawbacks: first, fluorescent signals may be affected by background noise and nonspecific binding, resulting in reduced accuracy; second, signal overlap and spectral band interference during the analysis process may affect the reliability of multiplexed detection; and finally, the selection and optimization of fluorescent probes are complex and require highly precise experimental design.

[0003] Therefore, in genetic testing based on fluorescence spectroscopy, it is necessary to denoise the fluorescence signal under various environments to eliminate background noise and nonspecific interference and ensure the accuracy of the signal; at the same time, by using the fluorescence spectra of multiple channels for composite analysis, different target genes can be detected simultaneously, increasing the sensitivity and diversity of detection, and effectively improving the accuracy and efficiency of genetic testing, especially in multiple labeling and complex sample analysis, significantly improving the reliability and throughput of detection. Summary of the Invention

[0004] The present invention aims to provide a method and system for analyzing gene detection data based on fluorescence spectroscopy, which can simultaneously detect different target genes for comprehensive gene analysis.

[0005] A method for analyzing gene detection data based on fluorescence spectroscopy, comprising the following steps:

[0006] Obtain a gene sample to be analyzed; perform analysis based on the gene sample to be analyzed and a fluorescence multi-channel matching analysis model to obtain a multi-channel fluorescence spectrum analysis strategy; perform multi-channel fluorescence spectrum extraction on the gene sample to be analyzed based on the multi-channel fluorescence spectrum analysis strategy to obtain a fluorescence spectrum to be analyzed; perform fluorescence spectrum analysis strategy matching based on a genetic algorithm of diversity variation using the fluorescence multi-channel matching analysis model to extract a multi-channel fluorescence spectrum;

[0007] Based on the fluorescence spectrum to be analyzed and the fluorescence spectrum gene analysis model, analysis is performed to obtain the gene detection analysis results; the fluorescence spectrum gene analysis model includes a fluorescence signal denoising layer, a spectral feature extraction layer, a gene map analysis layer and a gene detection output layer; among them, the fluorescence signal denoising layer is used to denoise the fluorescence spectrum to be analyzed based on dilated convolution to obtain a high-precision denoised fluorescence spectrum for subsequent feature analysis; the gene map analysis layer is used to perform feature fusion analysis on the gene detection results based on low-dimensional convolution features and principal component features to obtain the gene detection features of multi-channel fluorescence spectra; subsequent gene detection analysis is performed based on the gene detection analysis results.

[0008] As a preferred technical solution of the present invention, the fluorescence multi-channel matching analysis model includes a data acquisition layer, a fluorescence channel matching layer and a strategy output layer;

[0009] The data acquisition layer is used to obtain the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye based on the gene sample to be analyzed;

[0010] The fluorescence channel matching layer is used to perform strategy analysis based on the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye in combination with the improved genetic algorithm to obtain a multi-channel fluorescence spectrum analysis strategy;

[0011] The strategy output layer is used to output multi-channel fluorescence spectrum analysis strategies.

[0012] As a preferred technical solution of the present invention, the specific steps of performing strategy analysis in the fluorescence channel matching layer include:

[0013] Construct a fluorescence matching digital twin model based on the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye;

[0014] Construct K fluorescence channel matching samples , k=1, 2, ..., K; each fluorescence channel matches the sample Represents a fluorescence spectrum analysis strategy for the gene sample to be analyzed; set the fluorescence channel to match the sample The fitness of ; for The reciprocal of; set the maximum number of iterations, t is the current number of iterations, t=1, 2, ..., T; T is the specific value of the maximum number of iterations; match K fluorescence channels to samples Combine to obtain an iterative population of fluorescence channel matching samples; the maximum number of iterations is set by professional technicians based on actual conditions;

[0015] Setting Function

[0016]

[0017] Matching samples to fluorescent channels The fitness function of Represented as fluorescence channel matching samples The matching degree in the fluorescence matching digital twin model, for The norm of Matching samples to fluorescent channels Emission wavelength range in the fluorescence matching digital twin model, =[ , ], Indicates the minimum value of the emission wavelength range, Indicates the maximum value of the emission wavelength range; is the corresponding weight coefficient; Matching samples to fluorescent channels Emission spectra in fluorescence matching digital twins; is the corresponding weight coefficient;

[0018] set up is the population similarity index, i=1, 2,…,K, j=1, 2,…,K, i≠j; For the fluorescence channel matching sample when the number of iterations is t Match samples to fluorescent channels Similarity between individuals;

[0019] When performing population iteration, if the population similarity index When it is less than the preset population diversity index, the population mutation probability is increased; otherwise, the population mutation probability is reduced; the population mutation probability formula is: ; W is the variation weight coefficient, is the population mutation probability;

[0020] When the maximum number of iterations is reached, the fluorescence channel matching sample corresponding to the current maximum fitness is output, which is the optimal fluorescence channel matching sample; based on the optimal fluorescence channel matching sample, a multi-channel fluorescence spectrum analysis strategy is output.

[0021] As a preferred technical solution of the present invention, the fluorescence spectrum gene analysis model includes a fluorescence signal denoising layer, a spectrum feature extraction layer, a gene map analysis layer and a gene detection output layer;

[0022] The fluorescence signal denoising layer is used to denoise the fluorescence spectrum to be analyzed and obtain the denoised fluorescence spectrum;

[0023] The spectral feature extraction layer is used to extract features from the denoised fluorescence spectrum to obtain channel fluorescence spectrum features. , m=1, 2, …, M; M represents the total number of channel fluorescence spectrum features after feature extraction from the denoised fluorescence spectrum;

[0024] The gene map analysis layer is used to analyze the fluorescence spectrum based on the channel Perform feature analysis to obtain gene map detection results ;

[0025] The gene detection output layer is used to detect all gene maps The results of gene detection and analysis are obtained.

[0026] As a preferred technical solution of the present invention, the specific steps of performing denoising in the fluorescence signal denoising layer include:

[0027] Signal analysis layer, signal enhancement layer and signal output layer in the signal denoising layer;

[0028] The signal analysis layer is used to perform preliminary feature analysis on the fluorescence spectrum to be analyzed and obtain the characteristics of the fluorescence spectrum to be analyzed;

[0029] The signal enhancement layer consists of N convolution kernels The fluorescence spectrum features to be analyzed are denoised layer by layer to obtain the denoised fluorescence spectrum;

[0030] Among them, the convolution kernel is an expanded convolution, and the convolution kernel expansion rate is , n=1, 2, …, N;

[0031] in, , is the convolution kernel Denoising strength; is the convolution kernel The denoising strength, also expressed as denoising strength Target denoising strength; Represents the convolution kernel The convolution step size is Represents the convolution kernel The convolution kernel size;

[0032] In the convolution kernel In the denoising intensity, the fluorescence spectrum feature matching is used to analyze the , using the convolution kernel expansion rate De-noise the fluorescence spectrum features to be analyzed to obtain the layered denoised fluorescence spectrum ; Denoised fluorescence spectra Perform feature extraction to obtain layered fluorescence spectrum features ;

[0033] In the convolution kernel In the present study, based on the hierarchical fluorescence spectral characteristics Matching denoising strength , using the convolution kernel expansion rate De-noise the fluorescence spectrum features to be analyzed to obtain the layered denoised fluorescence spectrum ; Denoised fluorescence spectra Perform feature extraction to obtain layered fluorescence spectrum features ;

[0034] Until all convolution kernels are traversed, the denoised fluorescence spectrum is obtained;

[0035] The signal output layer is used to output the denoised fluorescence spectrum.

[0036] As a preferred technical solution of the present invention, the specific steps of performing feature analysis in the gene map analysis layer include:

[0037] The gene map analysis layer includes a local feature extraction layer, a principal component analysis layer, and a feature analysis layer;

[0038] In the local feature extraction layer, the channel fluorescence spectrum features are extracted Perform low-dimensional feature extraction to obtain low-dimensional channel fluorescence spectral features ;

[0039] In the principal component analysis layer, PCA is used to analyze the channel fluorescence spectrum characteristics. Perform principal component feature extraction to obtain the principal component channel fluorescence spectrum characteristics ;

[0040] In the feature analysis layer, the fluorescence spectrum features of low-dimensional channels are analyzed. and main component channel fluorescence spectral characteristics Perform feature recognition and obtain gene map detection results .

[0041] As a preferred technical solution of the present invention, the specific steps of training the gene map analysis layer include:

[0042] Collecting several groups of gene map analysis training samples, each group of gene map analysis training samples contains target gene analysis values ​​and fluorescence spectrum characteristics; combining the several groups of gene map analysis training samples to obtain a gene map analysis training set;

[0043] The gene map analysis training set is input into the gene map analysis layer for training to obtain an initial gene map analysis layer; a model evaluation is performed on the initial gene map analysis layer to obtain an initial gene map analysis layer model evaluation result; if the initial gene map analysis layer model evaluation result is passed, the initial gene map analysis layer is used as the gene map analysis layer in the fluorescence spectrum gene analysis model; otherwise, the model training is continued using the gene map analysis training set.

[0044] A gene detection data analysis system based on fluorescence spectroscopy, comprising:

[0045] The fluorescence spectrum matching module includes a sample acquisition unit and a channel matching unit; the sample acquisition unit is used to acquire the gene sample to be analyzed; the channel matching unit is used to analyze the gene sample to be analyzed and the fluorescence multi-channel matching analysis model to obtain a multi-channel fluorescence spectrum analysis strategy; based on the multi-channel fluorescence spectrum analysis strategy, the multi-channel fluorescence spectrum of the gene sample to be analyzed is extracted to obtain the fluorescence spectrum to be analyzed; the fluorescence multi-channel matching analysis model uses a genetic algorithm based on diversity variation to match the fluorescence spectrum analysis strategy to extract the multi-channel fluorescence spectrum;

[0046] The gene detection and analysis module includes a gene analysis unit and a subsequent analysis unit; the gene analysis unit is used to perform analysis based on the fluorescence spectrum to be analyzed and the fluorescence spectrum gene analysis model to obtain the gene detection analysis results; the fluorescence spectrum gene analysis model includes a fluorescence signal denoising layer, a spectral feature extraction layer, a gene map analysis layer and a gene detection output layer; among them, the fluorescence signal denoising layer is used to denoise the fluorescence spectrum to be analyzed based on dilated convolution to obtain a high-precision denoised fluorescence spectrum for subsequent feature analysis; the gene map analysis layer is used to perform feature fusion analysis of the gene detection results based on low-dimensional convolution features and principal component features to obtain gene detection features of multi-channel fluorescence spectra; the subsequent analysis unit is used to perform subsequent gene detection analysis based on the gene detection analysis results.

[0047] The present invention has the following advantages:

[0048] 1. The present invention uses a multi-channel fluorescence spectral analysis strategy combined with genetic algorithm optimization matching to more comprehensively capture the fluorescence signal characteristics of the sample to be analyzed, thereby improving the accuracy and reliability of genetic testing; multi-channel data fusion helps to extract more representative gene features from spectral information of different wavelengths, improving the sensitivity to gene markers; and the dilated convolution method is used to denoise the fluorescence spectrum to be analyzed, which can effectively filter out background noise, environmental interference and other factors, significantly improving the signal-to-noise ratio of the signal, thereby ensuring data quality during the genetic testing process.

[0049] 2. Through the layered denoising design, the signal denoising layer of the present invention can effectively eliminate the noise in the fluorescence spectrum to be analyzed, especially when the signal is weak or the background noise is high. Each layer of convolution kernel adopts different denoising strengths and expansion rates to finely remove noise of different frequencies. This layered and step-by-step denoising process can retain useful genetic information to the greatest extent and improve signal quality. By matching the denoising strength and utilizing the convolution kernel expansion rate, the denoising strength can be adaptively adjusted according to different signal characteristics. This design enables the denoising processing of each layer to be optimized according to the specific situation of the current signal, avoiding the limitations of unified parameter settings in traditional denoising methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a schematic diagram of the structure of a fluorescence spectroscopy-based gene detection data analysis system used in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0052] Example 1, a method for analyzing gene detection data based on fluorescence spectroscopy, comprising the following steps:

[0053] Obtain a gene sample to be analyzed; perform analysis based on the gene sample to be analyzed and a fluorescence multi-channel matching analysis model to obtain a multi-channel fluorescence spectrum analysis strategy; perform multi-channel fluorescence spectrum extraction on the gene sample to be analyzed based on the multi-channel fluorescence spectrum analysis strategy to obtain a fluorescence spectrum to be analyzed; perform fluorescence spectrum analysis strategy matching based on a genetic algorithm of diversity variation using the fluorescence multi-channel matching analysis model to extract a multi-channel fluorescence spectrum;

[0054] The genetic sample to be analyzed can be obtained by obtaining a blood sample through venous or fingertip blood collection, or by collecting saliva from the mouth using a dedicated saliva collection tool, or by extracting a tissue sample from the human body or animal body through surgery, biopsy, or needle aspiration, or by extracting hair roots from the head or body hair for genetic analysis.

[0055] The analysis is performed based on the fluorescence spectrum to be analyzed and the fluorescence spectrum gene analysis model to obtain the gene detection analysis results. The fluorescence spectrum gene analysis model includes a fluorescence signal denoising layer, a spectral feature extraction layer, a gene map analysis layer, and a gene detection output layer. Among them, the fluorescence signal denoising layer is used to denoise the fluorescence spectrum to be analyzed based on dilated convolution to obtain a high-precision denoised fluorescence spectrum for subsequent feature analysis. The gene map analysis layer is used to perform feature fusion analysis on the gene detection results based on low-dimensional convolution features and principal component features to obtain gene detection features of multi-channel fluorescence spectra. Subsequent gene detection analysis is performed based on the gene detection analysis results.

[0056] The fluorescence multi-channel matching analysis model includes a data acquisition layer, a fluorescence channel matching layer, and a strategy output layer;

[0057] The data acquisition layer is used to obtain the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye based on the gene sample to be analyzed. The data acquisition layer can extract detailed characteristic information from the gene sample to be analyzed, and at the same time, combine the spectral characteristics of the fluorescent dye to ensure that the sample characteristics and fluorescent signals are fully captured during the analysis process. This process provides high-quality input data for subsequent fluorescence channel matching analysis, ensuring the accuracy and reliability of the analysis.

[0058] The fluorescence channel matching layer is used to perform strategy analysis based on the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye, combined with an improved genetic algorithm, to obtain a multi-channel fluorescence spectrum analysis strategy. Through the fluorescence channel matching layer, the improved genetic algorithm can be used to optimize the multi-channel fluorescence spectrum extraction strategy based on the comprehensive information of the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye. The introduction of the genetic algorithm enables the spectrum extraction strategy to be adaptively adjusted, thereby maximally matching the characteristics of the sample to be analyzed, improving the efficiency and accuracy of multi-channel fluorescence spectrum extraction.

[0059] The strategy output layer is used to output multi-channel fluorescence spectrum analysis strategy;

[0060] Through precise fluorescence channel matching and strategy optimization, the model can effectively reduce noise and interference in fluorescence spectra, ensuring high-quality spectral data. This is particularly important for genetic testing, as high-quality spectral data is the basis for accurately identifying genetic markers, mutations, and other key features.

[0061] The specific steps for performing strategy analysis in the fluorescence channel matching layer include:

[0062] Construct a fluorescence matching digital twin model based on the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye;

[0063] Construct K fluorescence channel matching samples , k=1, 2, ..., K; each fluorescence channel matches the sample Represents a fluorescence spectrum analysis strategy for the gene sample to be analyzed; set the fluorescence channel to match the sample The fitness of ; for The reciprocal of; set the maximum number of iterations, t is the current number of iterations, t=1, 2, ..., T; T is the specific value of the maximum number of iterations; match K fluorescence channels to samples Combine to obtain an iterative population of fluorescence channel matching samples; the maximum number of iterations is set by professional technicians based on actual conditions;

[0064] Setting Function

[0065]

[0066] Matching samples to fluorescent channels The fitness function of Represented as fluorescence channel matching samples The matching degree in the fluorescence matching digital twin model, for The norm of Matching samples to fluorescent channels Emission wavelength range in the fluorescence matching digital twin model, =[ , ], Indicates the minimum value of the emission wavelength range, Indicates the maximum value of the emission wavelength range; is the corresponding weight coefficient, and the value is set manually; Matching samples to fluorescent channels Emission spectra in fluorescence matching digital twins; is the corresponding weight coefficient, and the value is set manually;

[0067] In the fitness function, the first term Matching samples to fluorescent channels The measure of the matching degree between the fluorescence matching digital twin models. Its mathematical meaning is the size of the matching degree. The optimization goal of this item is to minimize the matching degree error.

[0068] The second item Involves matching samples in fluorescent channels The emission spectrum in the fluorescence matching digital twin model is The minimum light intensity over the emission wavelength range in the fluorescence matching digital twin model is integrated to represent the total spectral intensity. By minimizing this term, the optimization goal is to make the light intensity value of the emission spectrum within the wavelength range as close to the expected target as possible, avoiding excessively low light intensity in undesirable wavelength ranges.

[0069] When the first and second terms are considered as fitness functions, it means that the fluorescence channels need to match the samples. Maximize the matching degree in the digital twin model while ensuring that the sample's emission spectrum meets the target requirements. By minimizing the matching error and the low-value part of the spectral intensity, more accurate fluorescence matching and higher-quality spectral output are achieved.

[0070] set up is the population similarity index, i=1, 2,…,K, j=1, 2,…,K, i≠j; For the fluorescence channel matching sample when the number of iterations is t Match samples to fluorescent channels Similarity between individuals;

[0071] When performing population iteration, if the population similarity index When it is less than the preset population diversity index, the population mutation probability is increased; otherwise, the population mutation probability is reduced; the population mutation probability formula is: ; W is the variation weight coefficient, the value is set by humans, is the population mutation probability;

[0072] When the maximum number of iterations is reached, the fluorescence channel matching sample corresponding to the current maximum fitness is output, which is the optimal fluorescence channel matching sample; based on the optimal fluorescence channel matching sample, a multi-channel fluorescence spectrum analysis strategy is output;

[0073] By calculating population similarity indicators and individual similarity, the similarity between matching samples in different fluorescence channels can be quantified, thereby adjusting the diversity of the population. When population similarity is too high, the mutation probability is increased to promote sample diversification; conversely, the mutation probability is reduced to enhance the stability of the strategy. This adaptive diversity adjustment mechanism improves the algorithm's exploration and convergence, and can find the optimal solution in a wider solution space. The dynamic adjustment of the population mutation probability enables the algorithm to flexibly adjust the mutation probability according to the population state in the current iteration process, avoiding falling into the local optimal solution. In particular, when population diversity decreases, by increasing the mutation probability, the problem of premature convergence is effectively avoided, ensuring the algorithm's exploration ability and innovation throughout the search process.

[0074] The fluorescence spectrum gene analysis model includes a fluorescence signal denoising layer, a spectrum feature extraction layer, a gene map analysis layer, and a gene detection output layer;

[0075] The fluorescence signal denoising layer is used to denoise the fluorescence spectrum to be analyzed and obtain the denoised fluorescence spectrum;

[0076] The spectral feature extraction layer is used to extract features from the denoised fluorescence spectrum to obtain channel fluorescence spectrum features. , m=1, 2, …, M; M represents the total number of channel fluorescence spectrum features after feature extraction from the denoised fluorescence spectrum;

[0077] The gene map analysis layer is used to analyze the fluorescence spectrum based on the channel Perform feature analysis to obtain gene map detection results ;

[0078] The gene detection output layer is used to detect all gene maps Combine and obtain the gene detection analysis results;

[0079] The fluorescence signal denoising layer denoises the fluorescence spectra to be analyzed, effectively eliminating noise interference and improving signal quality. This avoids misidentification of gene markers or mutations under the influence of noise, thereby ensuring the accuracy of genetic testing. The spectral feature extraction layer extracts channel fluorescence spectral features from the denoised fluorescence spectra to accurately extract effective information between different channels. The gene map analysis layer performs feature analysis based on channel fluorescence spectral features, which can deeply explore important information in genetic samples. Through a hierarchical design involving denoising, feature extraction, and gene map analysis, data quality and analysis results can be gradually optimized, significantly improving the accuracy of genetic testing.

[0080] The specific steps of denoising in the fluorescence signal denoising layer include:

[0081] Signal analysis layer, signal enhancement layer and signal output layer in the signal denoising layer;

[0082] The signal analysis layer is used to perform preliminary feature analysis on the fluorescence spectrum to be analyzed and obtain the characteristics of the fluorescence spectrum to be analyzed;

[0083] The signal enhancement layer consists of N convolution kernels The fluorescence spectrum features to be analyzed are denoised layer by layer to obtain the denoised fluorescence spectrum;

[0084] Among them, the convolution kernel is an expanded convolution, and the convolution kernel expansion rate is , n=1, 2, …, N;

[0085] in, , is the convolution kernel Denoising strength; is the convolution kernel The denoising strength, also expressed as denoising strength Target denoising strength; Represents the convolution kernel The convolution step size is Represents the convolution kernel The convolution kernel size;

[0086] In the convolution kernel In the denoising intensity, the fluorescence spectrum feature matching is used to analyze the , using the convolution kernel expansion rate De-noise the fluorescence spectrum features to be analyzed to obtain the layered denoised fluorescence spectrum ; Denoised fluorescence spectra Perform feature extraction to obtain layered fluorescence spectrum features ;

[0087] In the convolution kernel In the present study, based on the hierarchical fluorescence spectral characteristics Matching denoising strength , using the convolution kernel expansion rate De-noise the fluorescence spectrum features to be analyzed to obtain the layered denoised fluorescence spectrum ; Denoised fluorescence spectra Perform feature extraction to obtain layered fluorescence spectrum features ;

[0088] Until all convolution kernels are traversed, the denoised fluorescence spectrum is obtained;

[0089] The signal output layer is used to output the denoised fluorescence spectrum;

[0090] Through the design of layered denoising, the signal denoising layer can effectively eliminate the noise in the fluorescence spectrum to be analyzed, especially when the signal is weak or the background noise is high. Each layer of convolution kernel adopts different denoising strengths and expansion rates to finely remove noise of different frequencies. This layered and step-by-step denoising process can retain useful genetic information to the greatest extent and improve signal quality. By matching the denoising strength and utilizing the convolution kernel expansion rate, the denoising strength can be adaptively adjusted according to different signal characteristics. This design enables the denoising process of each layer to be optimized according to the specific situation of the current signal, avoiding the limitations of unified parameter settings in traditional denoising methods. Multi-level denoising allows each convolution kernel to focus more on signals in a specific frequency band during denoising. Each layer of convolution kernel processes the fluorescence spectrum characteristics layer by layer, refines the denoising process, and gradually removes interference noise, so that the processing results of each layer are more in line with downstream analysis requirements, and the final output denoised fluorescence spectrum is clearer and more accurate.

[0091] The specific steps for feature analysis in the gene map analysis layer include:

[0092] The gene map analysis layer includes a local feature extraction layer, a principal component analysis layer, and a feature analysis layer;

[0093] In the local feature extraction layer, the channel fluorescence spectrum features are extracted Perform low-dimensional feature extraction to obtain low-dimensional channel fluorescence spectral features ;

[0094] In the principal component analysis layer, PCA is used to analyze the channel fluorescence spectrum characteristics. Perform principal component feature extraction to obtain the principal component channel fluorescence spectrum characteristics ;PCA is principal component analysis;

[0095] In the feature analysis layer, the fluorescence spectrum features of low-dimensional channels are analyzed. and main component channel fluorescence spectral characteristics Perform feature recognition and obtain gene map detection results ;

[0096] The specific steps for training the gene map analysis layer include:

[0097] Collecting several groups of gene map analysis training samples, each group of gene map analysis training samples contains target gene analysis values ​​and fluorescence spectrum characteristics; combining the several groups of gene map analysis training samples to obtain a gene map analysis training set;

[0098] Inputting the gene map analysis training set into the gene map analysis layer for training to obtain an initial gene map analysis layer; performing model evaluation on the initial gene map analysis layer to obtain an initial gene map analysis layer model evaluation result; if the initial gene map analysis layer model evaluation result is passed, the initial gene map analysis layer is used as the gene map analysis layer in the fluorescence spectrum gene analysis model; otherwise, continuing model training using the gene map analysis training set;

[0099] Through the local feature extraction layer and the principal component analysis layer, the gene map analysis layer can effectively extract important features from the fluorescence spectrum. The local feature extraction layer reduces redundant information and retains key features through low-dimensional feature extraction. The principal component analysis layer further reduces the dimension through PCA and optimizes the feature representation. The combination of the two can effectively enhance the feature expression ability and improve the accuracy of gene map analysis. The combined use of low-dimensional feature extraction and principal component analysis can effectively reduce the dimension of the data, avoid data redundancy, and improve the speed and efficiency of subsequent analysis. PCA can extract the most discriminative features, reduce noise and irrelevant information, and improve the efficiency and performance of model training.

[0100] Subsequent genetic testing and analysis are performed based on the results of genetic testing and analysis. For example, in cancer genome analysis, different gene mutations are detected through multi-channel fluorescence. For the detection of EGFR gene and KRAS gene in lung cancer, multi-channel fluorescence spectroscopy can simultaneously identify two gene mutations, which is of great significance for personalized cancer treatment and can determine whether to use EGFR inhibitors or other targeted therapeutic drugs. In genetic testing for hereditary diseases, multi-channel fluorescence is used to simultaneously detect mutations in multiple genes. In HIV viral load testing, multi-channel fluorescence spectroscopy can simultaneously detect multiple gene fragments of the virus and perform quantitative analysis.

[0101] Example 2, a gene detection data analysis system based on fluorescence spectroscopy, see Figure 1 Shown, including:

[0102] The fluorescence spectrum matching module includes a sample acquisition unit and a channel matching unit; the sample acquisition unit is used to acquire the gene sample to be analyzed; the channel matching unit is used to analyze the gene sample to be analyzed and the fluorescence multi-channel matching analysis model to obtain a multi-channel fluorescence spectrum analysis strategy; based on the multi-channel fluorescence spectrum analysis strategy, the multi-channel fluorescence spectrum of the gene sample to be analyzed is extracted to obtain the fluorescence spectrum to be analyzed; the fluorescence multi-channel matching analysis model uses a genetic algorithm based on diversity variation to match the fluorescence spectrum analysis strategy to extract the multi-channel fluorescence spectrum;

[0103] The gene detection and analysis module includes a gene analysis unit and a subsequent analysis unit; the gene analysis unit is used to perform analysis based on the fluorescence spectrum to be analyzed and the fluorescence spectrum gene analysis model to obtain the gene detection analysis results; the fluorescence spectrum gene analysis model includes a fluorescence signal denoising layer, a spectral feature extraction layer, a gene map analysis layer and a gene detection output layer; among them, the fluorescence signal denoising layer is used to denoise the fluorescence spectrum to be analyzed based on dilated convolution to obtain a high-precision denoised fluorescence spectrum for subsequent feature analysis; the gene map analysis layer is used to perform feature fusion analysis of the gene detection results based on low-dimensional convolution features and principal component features to obtain gene detection features of multi-channel fluorescence spectra; the subsequent analysis unit is used to perform subsequent gene detection analysis based on the gene detection analysis results.

[0104] It should be understood that those skilled in the art may make improvements or modifications based on the above description, and all such improvements and modifications shall fall within the scope of protection of the appended claims. Any portion of this specification not described in detail is prior art known to those skilled in the art.

Claims

1. A method for analyzing gene detection data based on fluorescence spectroscopy, characterized in that: The following steps are involved: Obtaining genetic samples to be analyzed; Based on the gene sample to be analyzed and the fluorescence multi-channel matching analysis model, a multi-channel fluorescence spectrum analysis strategy is obtained; Based on the multi-channel fluorescence spectrum analysis strategy, the multi-channel fluorescence spectrum of the gene sample to be analyzed is extracted to obtain the fluorescence spectrum to be analyzed; the fluorescence multi-channel matching analysis model is based on the genetic algorithm of diversity variation to match the fluorescence spectrum analysis strategy to extract the multi-channel fluorescence spectrum; The analysis is performed based on the fluorescence spectrum to be analyzed and the fluorescence spectrum gene analysis model to obtain the gene detection analysis results. The fluorescence spectrum gene analysis model includes a fluorescence signal denoising layer, a spectral feature extraction layer, a gene map analysis layer, and a gene detection output layer. Among them, the fluorescence signal denoising layer is used to denoise the fluorescence spectrum to be analyzed based on dilated convolution to obtain a high-precision denoised fluorescence spectrum for subsequent feature analysis. The gene map analysis layer is used to perform feature fusion analysis on the gene detection results based on low-dimensional convolution features and principal component features to obtain the gene detection features of the multi-channel fluorescence spectrum. Conduct subsequent genetic testing and analysis based on the genetic testing and analysis results; The fluorescence multi-channel matching analysis model includes a data acquisition layer, a fluorescence channel matching layer, and a strategy output layer; The data acquisition layer is used to obtain the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye based on the gene sample to be analyzed; The fluorescence channel matching layer is used to perform strategy analysis based on the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye in combination with the improved genetic algorithm to obtain a multi-channel fluorescence spectrum analysis strategy; The strategy output layer is used to output multi-channel fluorescence spectrum analysis strategy; The specific steps for performing strategy analysis in the fluorescence channel matching layer include: Construct a fluorescence matching digital twin model based on the characteristics of the gene sample to be analyzed and the spectral characteristics of the fluorescent dye; Construct K fluorescence channels to match sample Y k , k = 1, 2, ..., K; each fluorescence channel matches the sample Y k Represents a fluorescence spectrum analysis strategy for the gene sample to be analyzed; set the fluorescence channel to match the sample Y k The fitness is S k ;S k is the reciprocal of F(k); set the maximum number of iterations, t is the current number of iterations, t = 1, 2, ..., T; T is the specific value of the maximum number of iterations; match K fluorescence channels to sample Y k Combine to obtain an iterative population of fluorescence channel matching samples; the maximum number of iterations is set by professional technicians based on actual conditions; Setting Function Match sample Y to the fluorescence channel k The fitness function of k Represented as fluorescence channel matching sample Y k Matching degree in the fluorescence matching digital twin model,|X k | for X k The norm of B k Match sample Y to the fluorescence channel k Emission wavelength range in the fluorescence matching digital twin model, B k =[λ min ,λ max ],λ min Indicates the minimum value of the emission wavelength range, λ max Indicates the maximum value of the emission wavelength range; α is the corresponding weight coefficient; G k Match sample Y to the fluorescence channel k Emission spectrum in the fluorescence matching digital twin model; β is the corresponding weight coefficient; set up is the population similarity index, i=1,2,…,K,j=1,2,…,K,i≠j;γ(Y i , Y j ) is the fluorescence channel matching sample Y when the number of iterations is t i and fluorescence channel matching sample Y j Similarity between individuals; When performing population iteration, if the population similarity index D t When it is less than the preset population diversity index, the population mutation probability is increased; otherwise, the population mutation probability is reduced; the population mutation probability formula is C t =W*(1-D t ); W is the variation weight coefficient, C t is the population mutation probability; When the maximum number of iterations is reached, the fluorescence channel matching sample corresponding to the current maximum fitness is output, which is the optimal fluorescence channel matching sample; based on the optimal fluorescence channel matching sample, a multi-channel fluorescence spectrum analysis strategy is output.

2. The method for analyzing gene detection data based on fluorescence spectroscopy according to claim 1, characterized in that: The fluorescence spectrum gene analysis model includes a fluorescence signal denoising layer, a spectrum feature extraction layer, a gene map analysis layer, and a gene detection output layer; The fluorescence signal denoising layer is used to denoise the fluorescence spectrum to be analyzed and obtain the denoised fluorescence spectrum; The spectral feature extraction layer is used to extract the features of the denoised fluorescence spectrum and obtain the channel fluorescence spectrum feature U m , m=1,2,…,M; M represents the total number of channel fluorescence spectrum features after feature extraction from the denoised fluorescence spectrum; The gene map analysis layer is used to analyze the fluorescence spectrum of the channel m Perform feature analysis to obtain the gene map detection result E m ; The gene detection output layer is used to convert all gene map detection results E m The results of gene detection and analysis are obtained.

3. The method for analyzing gene detection data based on fluorescence spectroscopy according to claim 2, characterized in that: The specific steps of denoising in the fluorescence signal denoising layer include: Signal analysis layer, signal enhancement layer and signal output layer in the signal denoising layer; The signal analysis layer is used to perform preliminary feature analysis on the fluorescence spectrum to be analyzed and obtain the characteristics of the fluorescence spectrum to be analyzed; The signal enhancement layer consists of N convolution kernels P n The fluorescence spectrum features to be analyzed are denoised layer by layer to obtain the denoised fluorescence spectrum; Among them, the convolution kernel is dilated convolution, and the convolution kernel dilation rate is Z n , n=1,2,…,N; Among them, Z n =(H n+1 -H n ) / A n (L n-1 ), H n is the convolution kernel P n Denoising strength; H n+1 is the convolution kernel P n+1 The denoising strength is also expressed as denoising strength H n The target denoising strength is A n Represents the convolution kernel P n The convolution step size, L n Represents the convolution kernel P n The convolution kernel size; In the convolution kernel P1, based on the matching denoising intensity H2 of the fluorescence spectrum to be analyzed, the convolution kernel expansion rate Z1 is used to denoise the fluorescence spectrum to be analyzed, and a layered denoised fluorescence spectrum R1 is obtained; the layered denoised fluorescence spectrum R1 is feature extracted to obtain a layered fluorescence spectrum feature Q1; In the convolution kernel P n In the experiment, based on the layered fluorescence spectral characteristics Q n-1 Matching denoising strength H n+1 , using the convolution kernel expansion rate Z n The fluorescence spectrum features to be analyzed are denoised to obtain the layered denoised fluorescence spectrum R n ; For the hierarchical denoised fluorescence spectrum R n Perform feature extraction to obtain the layered fluorescence spectrum feature Q n ; Until all convolution kernels are traversed, the denoised fluorescence spectrum is obtained; The signal output layer is used to output the denoised fluorescence spectrum.

4. The method for analyzing gene detection data based on fluorescence spectroscopy according to claim 3, characterized in that: The specific steps for feature analysis in the gene map analysis layer include: The gene map analysis layer includes a local feature extraction layer, a principal component analysis layer, and a feature analysis layer; In the local feature extraction layer, the channel fluorescence spectrum feature U m Perform low-dimensional feature extraction to obtain low-dimensional channel fluorescence spectrum features U m '; In the principal component analysis layer, PCA is used to analyze the channel fluorescence spectrum feature U m Extract the principal component features and obtain the principal component channel fluorescence spectrum feature U m ″; In the feature analysis layer, the low-dimensional channel fluorescence spectrum feature U m ′ and the main component channel fluorescence spectral characteristics U m ″Perform feature recognition and obtain gene map detection result E m .

5. The method for analyzing gene detection data based on fluorescence spectroscopy according to claim 4, characterized in that: The specific steps for training the gene map analysis layer include: Collect several groups of gene map analysis training samples, each group of gene map analysis training samples contains target gene analysis values ​​and fluorescence spectrum characteristics; combine several groups of gene map analysis training samples to obtain a gene map analysis training set; input the gene map analysis training set into the gene map analysis layer for training to obtain an initial gene map analysis layer; perform model evaluation on the initial gene map analysis layer to obtain an initial gene map analysis layer model evaluation result; if the initial gene map analysis layer model evaluation result is passed, use the initial gene map analysis layer as the gene map analysis layer in the fluorescence spectrum gene analysis model; otherwise, continue model training using the gene map analysis training set.

6. A gene detection data analysis system based on fluorescence spectroscopy, characterized in that: The system applies the method for analyzing genetic testing data based on fluorescence spectroscopy according to any one of claims 1 to 5, comprising: The fluorescence spectrum matching module includes a sample acquisition unit and a channel matching unit; the sample acquisition unit is used to acquire the gene sample to be analyzed; the channel matching unit is used to analyze the gene sample to be analyzed and the fluorescence multi-channel matching analysis model to obtain a multi-channel fluorescence spectrum analysis strategy; based on the multi-channel fluorescence spectrum analysis strategy, the multi-channel fluorescence spectrum of the gene sample to be analyzed is extracted to obtain the fluorescence spectrum to be analyzed; the fluorescence multi-channel matching analysis model uses a genetic algorithm based on diversity variation to match the fluorescence spectrum analysis strategy to extract the multi-channel fluorescence spectrum; The gene detection and analysis module includes a gene analysis unit and a subsequent analysis unit; the gene analysis unit is used to perform analysis based on the fluorescence spectrum to be analyzed and the fluorescence spectrum gene analysis model to obtain the gene detection analysis results; the fluorescence spectrum gene analysis model includes a fluorescence signal denoising layer, a spectral feature extraction layer, a gene map analysis layer and a gene detection output layer; among them, the fluorescence signal denoising layer is used to denoise the fluorescence spectrum to be analyzed based on dilated convolution to obtain a high-precision denoised fluorescence spectrum for subsequent feature analysis; the gene map analysis layer is used to perform feature fusion analysis of the gene detection results based on low-dimensional convolution features and principal component features to obtain gene detection features of multi-channel fluorescence spectra; the subsequent analysis unit is used to perform subsequent gene detection analysis based on the gene detection analysis results.

Images