A method, system, device, and medium for muscle fiber type analysis based on raman spectroscopy

Abstract

The application relates to the technical field of bioengineering, and discloses a muscle fiber type analysis method, system, device and medium based on Raman spectrum, which realizes non-destructive, efficient and low-cost muscle fiber type analysis by combining Raman spectrum technology with a long short-term memory network algorithm. The Raman spectrum technology does not require complex pretreatment and can preserve sample integrity, and the LSTM algorithm can accurately learn the relationship between Raman spectrum data and initial DNA copy data of muscle fiber coding genes, thereby improving the accuracy and intelligent level of muscle fiber type analysis. In addition, the application is suitable for muscle tissues of any part of mammals, effectively breaks through the limitations of traditional methods, and provides a new tool which is efficient, accurate and widely applicable for pork quality evaluation and related research.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of bioengineering, and particularly relates to a muscle fiber type analysis method, system, device and medium based on Raman spectrum. BACKGROUND

[0002] In recent years, people's living standards have gradually improved, and the requirements for pork quality are also becoming more and more strict. As the basic unit of muscle, the morphology, type and other characteristics of muscle fiber are closely related to the meat quality evaluation index concerned by consumers. Muscle fiber is highly differentiated, and can be divided into different types according to its morphology, function and physiological and biochemical characteristics. At present, the commonly used muscle fiber classification method is to divide muscle fiber into four types according to the enzyme system and its activity characteristics: slow oxidative muscle fiber (type I), fast oxidative muscle fiber (type IIa), fast glycolytic muscle fiber (type IIb) and intermediate muscle fiber (type IIx). At the molecular level, muscle fiber type is regulated by myosin heavy chain (MyHC) gene family, and different types of muscle fibers express specific MyHC subtypes. There are mainly four MyHC isoforms in the skeletal muscle of adult mammals, which are MYH7, MYH2, MYH1 and MYH4, respectively, encoding four types of muscle fibers I, IIa, IIx and IIb. The muscle fiber type composition of different parts of pig carcass is quite different, and most muscles are composed of oxidative and glycolytic muscle fibers. Different types of muscle fibers have different biochemical characteristics, mainly in myoglobin content, glycogen content, ATP, lipid content, soluble protein content, enzyme content and activity, etc., which have great differences.

[0003] Traditional muscle fiber analysis methods, such as histological staining (such as ATPase staining), immunohistochemistry (MyHC isoform antibody) or qPCR technology, have the limitations of high sample destruction, complicated steps and large reagent consumption. In recent years, the development of single-cell metabolomics technology provides a new tool for analyzing cell heterogeneity, but the complex pretreatment process and high cost of mass spectrometry technology limit its large-scale application. Therefore, there is an urgent need for a more efficient and low-cost muscle fiber type analysis method. SUMMARY

[0004] The present application provides a muscle fiber type analysis method, system, device and medium based on Raman spectrum, to solve the defects of high sample destruction, complicated steps and high cost of existing muscle fiber analysis methods.

[0005] The present application provides a method for constructing a muscle fiber type analysis model based on Raman spectrum, comprising:

[0006] Obtain Raman spectrum data and initial DNA copy data of muscle fiber coding genes of training samples, wherein the training samples are muscle tissues of any part of mammals;

[0007] According to the Raman spectrum data of the training sample and the initial DNA copy data of the muscle fiber encoding gene, the long short-term memory network algorithm is used to make the model learn the relationship between the Raman spectrum data of the training sample and the initial DNA copy data of the muscle fiber encoding gene, so as to construct the muscle fiber type analysis model.

[0008] According to the muscle fiber type analysis model based on the Raman spectrum provided by the application, the mammal can be an animal selected from the bovidae, equidae, felidae, canidae, leporidae, suidae, camelidae, rodents and primates, including but not limited to cattle, horses, goats, sheep, cats, rabbits, pigs, camels, alpacas, rats, mice, guinea pigs, non-human primates (such as apes, monkeys, baboons and chimpanzees) and humans, preferably cattle, horses, dogs, goats, sheep, pigs, camels, rats, mice, monkeys and humans. Preferably, the mammal is an experimental animal, including but not limited to mice, rats, rabbits, guinea pigs, hamsters, monkeys, dogs, cats, pigs, sheep, horses and the like. These animals have developed muscle tissue and are widely used in scientific research, medical experiments and muscle biology research.

[0009] According to the muscle fiber type analysis model based on the Raman spectrum provided by the application, the muscle fiber encoding gene includes MYH7, MYH2, MYH1 and MYH4 which encode type I, type IIa, type IIx and type IIb muscle fibers respectively.

[0010] According to the muscle fiber type analysis model based on the Raman spectrum provided by the application, the muscle fiber type analysis model based on the Raman spectrum provided by the application, the muscle fiber encoding gene includes MYH7, MYH2, MYH1 and MYH4 which encode type I, type IIa, type IIx and type IIb muscle fibers respectively.

[0011] The Raman spectrum data of the training sample is preprocessed, wherein the preprocessing includes any one or any combination of the following: SG smoothing, multiplicative scatter correction (MSC), standard normal variate transform (SNV) and second derivative (2D).

[0012] It should be noted that the MSC can effectively eliminate noise and baseline drift caused by mirror reflection of the sample, uneven particle size and spectral scattering effect, and enhance the spectral information related to the content of the component. The SG smoothing can effectively remove noise and better preserve the original spectral information, the SNV can effectively eliminate Raman spectrum noise caused by external factors such as light source power variation and light intensity attenuation, and the 2D can remove high-frequency noise and mutual interference between adjacent components.

[0013] According to the present application, a method for constructing a muscle fiber type analysis model based on Raman spectrum is provided. According to the Raman spectrum data of the training sample and the initial DNA copy data of the muscle fiber coding gene, the long short-term memory network algorithm is used to learn the relationship between the Raman spectrum data of the training sample and the initial DNA copy data of the muscle fiber coding gene, and a muscle fiber type analysis model is constructed.

[0014] According to the Raman spectrum data of the training sample, the Competitive Adaptive Reweighted Sampling (CARS) and / or Uninformative Variables Elimination (UVE) are used to obtain the Raman spectrum characteristic wavelength data of the training sample.

[0015] According to the Raman spectrum characteristic wavelength data of the training sample and the initial DNA copy data of the muscle fiber coding gene, the long short-term memory network algorithm (LSTM) is used to learn the relationship between the Raman spectrum characteristic wavelength data of the training sample and the initial DNA copy data of the muscle fiber coding gene, and a muscle fiber type analysis model is constructed.

[0016] It should be noted that CARS is a variable selection algorithm based on Darwin's survival of the fittest theory. By adaptively weighted sampling, the points with larger absolute value of regression coefficient in the partial least squares model are reserved as a new subset, and then a partial least squares model is established based on the new subset. After multiple calculations, the wavelengths in the subset with the smallest root mean square error of cross-validation in the model are selected as the characteristic wavelengths.

[0017] UVE screens variables by analyzing the stability of regression coefficients. Specifically, UVE adds random noise variables to the original data, compares the "importance" of these noise variables and original variables, identifies the uninformative variables that contribute less to modeling, and removes them. By removing uninformative variables, the number of variables required for modeling is reduced, thereby reducing the complexity of the model.

[0018] LSTM is a time series based recurrent neural network, which adopts time reverse propagation for training. In LSTM, a storage unit is used to replace a conventional neuron, each storage unit is composed of an input gate, an output gate and its own state, and the internal information processing of a neuron is completed through the above three gate mechanisms. Through the above three gate mechanisms, the LSTM model can make information pass through the cell chain to capture the dependency between elements with a large distance between two time points, that is, it can form a memory for long-term data information in the past. The storage unit of the LSTM is more intelligent than the ordinary neural network, and has the memory function of the recent sequence, which solves the problems of gradient explosion and gradient disappearance.

[0019] According to the muscle fiber type analysis model based on the Raman spectrum provided by the application, the relationship between the Raman spectrum data of the training sample and the initial DNA copy data of the muscle fiber coding gene is learned by using the long short-term memory network algorithm, and the muscle fiber type analysis model is constructed.

[0020] During the training, the performance of the muscle fiber type analysis model is evaluated by using the model performance evaluation index until the performance of the muscle fiber type analysis model meets the preset requirements, and the muscle fiber type analysis model is obtained, wherein the model performance evaluation index includes any one or any combination of the following: the correction set determination coefficient (R c 2 ), the validation set determination coefficient (R p 2 ), and the root mean square error (Root Mean Square Error, RMSE).

[0021] The application provides a muscle fiber type analysis method based on Raman spectrum, comprising:

[0022] Obtaining the Raman spectrum data of the sample to be measured, the sample to be measured being muscle tissue of any part of a mammal;

[0023] According to the Raman spectrum data of the sample to be measured, the muscle fiber type analysis model based on the Raman spectrum is obtained by the muscle fiber type analysis model construction method, and the initial copy number of MYH7, MYH2, MYH1 and MYH4 of the four muscle fiber types of type I, type IIa, type IIx and type IIb in the sample to be measured is obtained.

[0024] The muscle fiber type analysis method based on the Raman spectrum provided by the application further comprises:

[0025] According to the initial copy numbers of MYH7, MYH2, MYH1 and MYH4 respectively encoding type I, type IIa, type IIx and type IIb muscle fiber types in the to-be-tested sample, the composition ratio of the four muscle fiber types in the to-be-tested sample is obtained.

[0026] The application further provides a muscle fiber type analysis system based on Raman spectrum, comprising:

[0027] The data acquisition module is configured to acquire Raman spectrum data of the to-be-tested sample, the to-be-tested sample being muscle tissue of any part of a mammal.

[0028] The analysis module is configured to obtain the initial copy numbers of MYH7, MYH2, MYH1 and MYH4 respectively encoding type I, type IIa, type IIx and type IIb muscle fiber types in the to-be-tested sample according to the Raman spectrum data of the to-be-tested sample and the muscle fiber type analysis model obtained by the construction method of the muscle fiber type analysis model based on Raman spectrum, and further obtain the composition ratio of the four muscle fiber types in the to-be-tested sample.

[0029] The application further provides an electronic device comprising a processor and a memory storing a computer program, wherein the processor implements the construction method of the muscle fiber type analysis model based on Raman spectrum or the muscle fiber type analysis method based on Raman spectrum when executing the computer program.

[0030] The application further provides a non-transitory computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the construction method of the muscle fiber type analysis model based on Raman spectrum or the muscle fiber type analysis method based on Raman spectrum.

[0031] The application further provides a computer program product comprising a computer program, wherein the computer program can be stored on a non-transitory computer readable storage medium, and the computer program is executed by a processor to enable a computer to implement the construction method of the muscle fiber type analysis model based on Raman spectrum or the muscle fiber type analysis method based on Raman spectrum.

[0032] The application provides a muscle fiber type analysis method, system, device and medium based on Raman spectrum, which realizes non-destructive, efficient and low-cost muscle fiber type analysis by combining Raman spectrum technology with a long short-term memory network algorithm. The Raman spectrum technology does not need complex pretreatment and can retain sample integrity, and the LSTM algorithm can accurately learn the relationship between Raman spectrum data and initial DNA copy data of muscle fiber coding genes, thereby improving the accuracy and intelligent level of muscle fiber type analysis. In addition, the application is suitable for muscle tissues of any part of mammals, effectively breaks through the limitations of traditional methods, and provides a new tool that is efficient, accurate and widely applicable for pork quality evaluation and related research. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0034] Figure 1 A flowchart of a muscle fiber type analysis method based on Raman spectrum provided by the application.

[0035] Figure 2 A GAPDH standard curve of MYH1, MYH2, MYH4 and MYH7 is shown.

[0036] Figure 3 The original Raman spectrum of muscle tissue and the Raman spectrum after MSC+2D processing.

[0037] Figure 4 A process of screening characteristic Raman variables by CARS algorithm is shown.

[0038] Figure 5 A RMSE change graph in the LSTM iteration process.

[0039] Figure 6 A scatter plot of the calibration set and the prediction set of the optimal model.

[0040] Figure 7 A structure diagram of a muscle fiber type analysis system based on Raman spectrum provided by the application.

[0041] Figure 8 A structure diagram of an electronic device provided by the application. DETAILED DESCRIPTION

[0042] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be clearly and completely described below in combination with the drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments of the present application, and they should not be understood as limiting the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application. In the description of the present application, it should be understood that the terms used are only for the purpose of description and should not be understood as indicating or implying relative importance.

[0043] The present application provides a muscle fiber type analysis method based on Raman spectrum, which includes a muscle fiber type analysis model construction method and an application method. The execution subject of the muscle fiber type analysis method based on Raman spectrum provided by the present application can be any applicable terminal side device or network side device, such as a muscle fiber type analysis device based on Raman spectrum.

[0044] Referring to Figure 1 The muscle fiber type analysis model construction method part of the muscle fiber type analysis method based on Raman spectrum provided by the present application can include the following steps.

[0045] S110, acquiring Raman spectrum data and initial DNA copy data of muscle fiber encoding genes of the training sample, wherein the training sample is muscle tissue of any part of a mammal, and the muscle fiber encoding genes include MYH7, MYH2, MYH1 and MYH4 which encode type I, type IIa, type IIx and type IIb muscle fibers respectively.

[0046] It should be noted that the mammal can be an animal selected from the group consisting of bovids, equids, felids, canids, leporids, suids, camellids, rodents and primates, including but not limited to cattle, horses, goats, sheep, cats, rabbits, pigs, camels, alpacas, rats, mice, guinea pigs, non-human primates (such as apes, monkeys, baboons and chimpanzees) and humans, preferably cattle, horses, dogs, goats, sheep, pigs, camels, rats, mice, monkeys and humans. Preferably, the mammal is an experimental animal, including but not limited to mice, rats, rabbits, guinea pigs, hamsters, monkeys, dogs, cats, pigs, sheep, horses and the like. These animals have developed muscle tissue and are widely used in scientific research, medical experiments and muscle biology research.

[0047] In the present embodiment, the Raman spectrum data and the initial DNA copy data of the muscle fiber encoding genes of the training sample can be obtained by scanning the training sample by an experimental personnel using a Raman spectrum scanning device and by a PCR experiment, and the specific process can be as follows.

[0048] 1. Sample preparation

[0049] Duroc × Landrace × Yorkshire boar and sow, each 5, body weight 110 ± 5 kg, after slaughter, respectively, select the pig's psoas major muscle, longissimus dorsi muscle, rectus abdominis muscle, trapezius muscle, external oblique muscle, brachial triceps muscle, common extensor muscle, semimembranosus muscle, biceps femoris muscle, gastrocnemius muscle, tongue muscle, masseter muscle, toe lateral extensor muscle, etc. Muscle tissue at each site, 1 g sample into a frozen tube, frozen in liquid nitrogen and transferred to a -80℃ refrigerator. At the same time, use a portable Raman spectrometer (E785PRO, Anfitek (Beijing) Technology Co., Ltd.) to collect spectral information.

[0050] 2. Spectral acquisition

[0051] Raman spectrum scanning range 32-3650 cm -1 , laser intensity 785 mW, resolution 2 cm -1 , scan 7 times. In order to make the spectrum representative, all samples are repeated 3 times, and the average spectrum is obtained.

[0052] 3. Muscle fiber type detection

[0053] 1) Total RNA extraction

[0054] Take a grinding tube, add 1 mL of RNA extraction solution, add grinding beads, and pre-cool on ice. Take 5-20 mg of tissue and add it to the grinding tube. Grind the grinder thoroughly until no visible tissue pieces. Centrifuge at 12000 rpm for 10 min at 4℃ to take the supernatant. Add 100 μL of chloroform substitute, mix well, and stand for 3 min. Centrifuge at 12000 rpm for 10 min at 4℃, transfer 400 μL of supernatant to a new centrifuge tube, add 550 μL of isopropanol, and mix well. Place at -20℃ for 15 min. Centrifuge at 12000 rpm for 10 min at 4℃, and the white precipitate at the bottom of the tube is the RNA. Remove the liquid, add 1 mL of 75% ethanol, mix well, and wash the precipitate, centrifuge at 12000 rpm for 5 min at 4℃. Remove the liquid again, add 1 mL of 75% ethanol, mix well, and wash the precipitate, centrifuge at 12000 rpm for 5 min at 4℃. Remove the liquid completely, and place the centrifuge tube on the clean bench and blow for 3-5 min. Add 15 μL of RNA dissolving solution to dissolve the RNA. Use Nanodrop 2000 to detect the concentration and purity of RNA, and dilute the RNA with a proper proportion to make its final concentration 200 ng / μL.

[0055] 2) RNA reverse transcription to synthesize cDNA

[0056] RNA samples were reverse transcribed into cDNA using a one-tube genomic DNA removal and reverse transcription kit from Wuhan Saivier Biotechnology Co., Ltd. according to the instructions. Store in a -20°C refrigerator for standby.

[0057] 3) Primer sequence and synthesis

[0058] All primer sequences were designed through the NCBI website and synthesized by Shanghai Shengong Bioengineering Co., Ltd. The primer sequence design is shown in Table 1.

[0059]

[0060] 4) Preparation of plasmid standard

[0061] The target gene fragment was synthesized, and the amplification product was detected by 3% agarose gel electrophoresis. The gel recovery kit was used for purification and recovery. The recovery product was ligated with pMD18-T vector, and the ligation product was transformed into E. coli Top10 competent cells, take positive colonies, inoculate into liquid medium, 37°C shaking culture, use bacterial solution as template, PCR identification. Reaction system (total volume is 50 μL): 2×Fast Pfus PCR Master Mix 25 μL; 10 μM Forward Primer 1 μL; 10 μM Reverse Primer 1 μL; Bacterial solution 2 μL; Water Nuclease-Free 21 μL. Reaction conditions: 98 ℃ pre-denaturation 2 min; 98 ℃ denaturation 10 s, 55 ℃ annealing 10 s, 72 ℃ extension 15 s, a total of 30 cycles, finally 72 ℃ final extension 5 min, 16 ℃ cooling 2 min.

[0062] The qualified plasmid was sent to Wuhan Jin Kai Rui Biological Engineering Co., Ltd. for sequencing comparison. The above sequencing correct plasmid was extracted with kit, and its concentration was measured with spectrophotometer. According to the formula plasmid copy number (copies / μL)=(6.02 ×10 23 )×(concentration ng / μL ×10 -9 ) / molecular weight, the plasmid copy number was obtained, and 10-fold gradient dilution was made as standard for standby.

[0063] 5) Real-time fluorescent quantitative PCR

[0064] The sample and different dilutions (10 4 , 10 5 , 10 6 , 10 7 , 10 8The standard sample of the standard substance (the standard sample of the standard substance) was set as a template, and real-time fluorescent quantitative PCR reaction was performed. The reaction system (total volume was 15 μL): 2×Universal Blue SYBR Green qPCR Master Mix, 7.5 μL; 2.5 μM gene primers (upstream + downstream), 1.5 μL; reverse transcription product (cDNA), 2.0 μL; Water Nuclease-Free, 4.0 μL. Reaction conditions: 95 ℃ pre-denaturation for 30 s; 95 ℃ denaturation for 15 s, 60 ℃, annealing / extension for 30 s, a total of 40 cycles. After amplification, 65-95 ℃, every time 0.5 ℃, collect fluorescence signal once. The amplification curve and Ct value were obtained by PCR amplification, and the standard curve was drawn with the log value of the starting copy number (SQ, Starting Quantity, unit: copies / μL) of the standard sample as the abscissa and the corresponding Ct value as the ordinate.

[0065] In an embodiment, after acquiring the Raman spectrum data of the training sample, the Raman spectrum data of the training sample can be preprocessed, wherein the preprocessing includes any one or any combination of the following: SG smoothing, multivariate scatter correction, standard normal variable variation, and second derivative.

[0066] It should be noted that the MSC can effectively eliminate noise and baseline drift caused by mirror reflection of the sample, uneven particle size, and spectral scattering effect, and enhance the spectral information related to the content of the component. The SG smoothing can effectively remove noise and better preserve the original spectral information, the SNV can effectively eliminate Raman spectrum noise caused by external factors such as light source power variation and light intensity attenuation, and the 2D can remove high-frequency noise and mutual interference between adjacent components.

[0067] S120, according to the Raman spectrum data of the training sample and the initial DNA copy data of the myofiber encoding gene, using a long short-term memory network algorithm, the model learns the relationship between the Raman spectrum data of the training sample and the initial DNA copy data of the myofiber encoding gene, and a myofiber type analysis model is constructed.

[0068] The original Raman spectrum collected has a large number (1849 in this embodiment) of variables, and there can be a large amount of redundant information in the variables, which can affect the accuracy of the model and reduce the operation speed. Therefore, in one embodiment, for the preprocessed Raman spectrum data, S120 can use the competitive adaptive reweighted algorithm and / or non-informative variable elimination to obtain the Raman spectrum characteristic wavelength data of the training samples, so as to reduce the collinearity of the original Raman spectrum data and improve the operation speed of the model and optimize the performance of the model; and then according to the Raman spectrum characteristic wavelength data of the training samples and the initial DNA copy data of the myofiber coding gene, the long short-term memory network algorithm is used to make the model learn the relationship between the Raman spectrum characteristic wavelength data of the training samples and the initial DNA copy data of the myofiber coding gene, and a myofiber type analysis model is constructed. In this embodiment, the Raman spectrum characteristic wavelength data and the initial DNA copy data of the myofiber coding gene are randomly divided into a calibration set and a prediction set according to a 3:1 ratio, and a long short-term memory network is used to establish a quantitative myofiber type analysis model to predict the initial copy number of MYH1, MYH2, MYH4 and MYH7 in pork.

[0069] It should be noted that CARS is a variable selection algorithm based on Darwin's survival of the fittest theory. It retains the points with larger absolute value weights of regression coefficients in the partial least squares model as a new subset through adaptive weighted sampling, and then establishes a partial least squares model based on the new subset. After multiple calculations, the wavelengths in the subset with the smallest root mean square error of cross-validation in the model are selected as the characteristic wavelengths.

[0070] UVE screens variables by analyzing the stability of regression coefficients. Specifically, UVE adds random noise variables to the original data, compares the "importance" of these noise variables and the original variables, identifies the uninformative variables that contribute less to modeling, and eliminates them. By removing uninformative variables, the number of variables required for modeling is reduced, thereby reducing the complexity of the model.

[0071] LSTM is a recurrent neural network based on time series, which is trained using time backpropagation. In LSTM, the storage unit is used to replace the conventional neuron, and each storage unit is composed of an input gate, an output gate and its own state. Through the mechanism of the above three gates, the internal information processing of a neuron is completed. Through the above three gating mechanisms, the LSTM model can make information transmitted between cells to capture the dependency between elements with a large distance between two time points, i.e., it can form a memory for long-term data information in the past. The storage unit of LSTM is more intelligent than the ordinary neural network, and has the memory function of the recent sequence, solving the problem of gradient explosion and gradient disappearance.

[0072] In an embodiment, during the training, S120 can evaluate the performance of the muscle fiber type analysis model using a model performance evaluation index, wherein the model performance evaluation index includes a validation set determination coefficient and a root mean square error. The determination coefficient R 2 is the ratio of the mean square error to the variance, and the larger the result indicates the higher the model accuracy. The root mean square error is the square root of the sum of the square of the deviation between the observed value and the true value and the number of observations n, and the smaller the root mean square error, the better the modeling effect. The processed spectral data is modeled, and the determination coefficient and the root mean square error of the obtained model are compared to test the model accuracy.

[0073] Referring to Figure 2 , the MYH1, MYH2, MYH4, and MYH7 standard samples with a concentration of 100 ng / μL are diluted 10,000 times as the first gradient, and then sequentially diluted 10 times to obtain the last four gradients. The five concentrations are used as standard samples for fluorescence quantitative PCR reaction to obtain standard curves. The standard curve regression equations of MYH1, MYH2, MYH4, and MYH7 are y=-3.3891x+33.832, y=-3.5007x+35.931, y=-3.5717x+35.799, and y=-3.5612x+36.494, respectively. The curve fitting degrees (R 2 ) are 0.9979, 0.9971, 0.9984, and 0.9989, respectively, all of which are above 0.99, indicating that the logarithmic value of the initial copy number of the standard sample has a good linear relationship with the corresponding cq value. It is generally considered that a good amplification efficiency should be between 90% and 110%, and the amplification efficiencies (E) of the four standard curves are 97.27%, 93.04%, 90.54%, and 90.90%, respectively, meeting the experimental requirements.

[0074] According to the regression equation, the initial copy number of MYH1, MYH2, MYH4, and MYH7 in each sample is calculated, and the SQ value is converted to log for convenience of Raman spectrum modeling as the target variable of the model. Table 2 shows the range and standard deviation of the log (SQ) value of MYH1, MYH2, MYH4, and MYH7 in the sample.

[0075]

[0076] From Figure 3 it can be seen that the Raman characteristic peaks of the muscle fibers of the pork are at 828 cm -1 , 850 cm -1 , 935 cm -1 , 1002 cm -1 , 1028 cm -1 , 1122 cm -1 , and 1312 cm -11447 cm -1 1647 cm -1 2926 cm -1 At the point of displacement. 828 cm. -1 and 850 cm -1 The vibration of the benzene ring of tyrosine is 1002 cm⁻¹. -1 This is the respiratory vibration of the benzene ring of phenylalanine, 1312 cm. -1 It is amide III, 1647 cm -1 It is amide I, 935 cm -1 CC key extension, 1028 cm -1 For the C-C bond in -(CH2)n-, 1122 cm -1 For C and C bonds stretched in the same direction, 1447 cm -1 The CH bond is bent, 2926 cm. -1 The CH bond in the methylene CH2 group. Different muscle fiber types exhibit significant differences in their metabolic pathways, including glycolysis, lipogenesis, fatty acid synthesis, and the tricarboxylic acid (TCA) cycle. This leads to substantial differences in the metabolic products (including amino acids, phospholipids, carbohydrates, steroids, fatty acids and lipids, and carboxylates) of muscle tissue from different muscle fiber types. These characteristic peaks are closely related to metabolic precursors and metabolites in muscle fibers. Figure 3 The images show the Raman spectra after MSC and 2D preprocessing. It can be seen that the signals of each characteristic peak are significantly enhanced after fluorescence background subtraction.

[0077] The collected spectral information was preprocessed, and the prediction results of the model are shown in Table 2. The model performed best when the MSC+2D preprocessing method was used for type I muscle. The calibration set and prediction set R... 2 The accuracy and accuracy of the models reached 0.97 and 0.88 respectively, and the RMSE reached 0.15 and 0.31 respectively. The model established using the 2D preprocessing method for type IIa muscle showed the best performance, with the calibration set and prediction set R... 2 The accuracy and RMSE values ​​reached 0.93 and 0.85 respectively, and 0.24 and 0.29 respectively. The model established using the MSC+2D preprocessing method for type IIb muscle showed the best performance, with the calibration set and prediction set R... 2 The accuracy and accuracy of the models reached 0.98 and 0.88, respectively, and the RMSE reached 0.15 and 0.30, respectively. The model established using the MSC+2D preprocessing method for type IIx muscle showed the best performance, with the calibration set and prediction set R... 20.95 and 0.86, and the RMSE reached 0.19 and 0.26, respectively. The prediction models of the four muscle fiber types were established, and the 2D method was used for spectral pretreatment. The established models were relatively accurate because the 2D method can effectively eliminate baseline drift and effectively distinguish overlapping peaks in the original spectrum, thereby improving the identification ability of the characteristic peaks.

[0078] The CARS algorithm and the UVE algorithm were applied to the pretreated Raman spectrum data for waveband optimization. Taking the MYH7 Log(SQ) value prediction model as an example, the variable number of Raman spectrum, RMSE, and the change of regression coefficient of each Raman shift feature during the operation of the CARS algorithm are shown in Figure 4 With the increase of the number of sample operations, the number of Raman shift variables retained by the model gradually decreased, and the reduction rate decreased from high to low, indicating that the variable screening process was from coarse screening to fine screening. When the operation number reached 37, a large number of irrelevant Raman shift variables of type I muscle fiber Raman spectrum were eliminated, and the RMSE value was the smallest, indicating that the prediction ability of the model was the strongest. Finally, the variable subset and the characteristic Raman shift variable related to type I muscle fiber prediction were selected, which were 154 Raman shift variables, accounting for 8.32% of the total variable number.

[0079] The wavebands screened by the CARS and UVE methods were used for LSTM modeling, and the results are shown in Table 3. After spectral pretreatment, the modeling effect was improved, and the modeling effect of CARS was better, with Rp2 of the prediction set reaching 0.88 and RMSEP of 0.08.

[0080]

[0081] The LSTM algorithm was used to establish the prediction model of the initial copy number of MYH1, MYH2, MYH4, and MYH7 in the sample. During the modeling process, the RMSE change graph of LSTM in the iteration process is shown in Figure 5 The model gradually stabilized in the iteration process, and the optimal modeling effect was obtained. Table 4 shows the modeling effect of different pretreatment methods after CARS screening wavebands. It can be seen that the prediction effect of the initial copy number of the four muscle fiber type-specific marker genes is good, with R p 2 The range of R is 0.85-0.88, and the range of RMSEP is 0.26-0.31. It is shown that the initial copy number of the four muscle fiber type-specific marker genes can be accurately predicted by Raman spectroscopy, so as to calculate the composition ratio of the four muscle fiber types in the sample.

[0082]

[0083]

[0084] After the muscle fiber type analysis model is trained, a muscle fiber type analysis method based on Raman spectroscopy can be formed, comprising:

[0085] S130, acquiring Raman spectrum data of a to-be-tested sample, the to-be-tested sample being muscle tissue of any part of a mammal;

[0086] S140, obtaining initial copy numbers of MYH7, MYH2, MYH1 and MYH4 respectively encoding four types of muscle fiber types I, IIa, IIx and IIb in the to-be-tested sample according to the Raman spectrum data of the to-be-tested sample through the muscle fiber type analysis model obtained above, and further, obtaining the composition ratio of the four types of muscle fiber types in the to-be-tested sample according to the initial copy numbers of MYH7, MYH2, MYH1 and MYH4 respectively encoding four types of muscle fiber types I, IIa, IIx and IIb in the to-be-tested sample.

[0087] The muscle fiber type analysis method, system, device and medium based on Raman spectroscopy provided by the application realize non-destructive, efficient and low-cost muscle fiber type analysis by using Raman spectroscopy technology combined with a long short-term memory network algorithm. The Raman spectroscopy technology does not require complex pretreatment and can preserve the integrity of the sample, and the LSTM algorithm can accurately learn the relationship between the Raman spectrum data and the initial DNA copy data of the muscle fiber encoding gene, thereby improving the accuracy and intelligent level of muscle fiber type analysis. In addition, the application is suitable for muscle tissue of any part of a mammal, effectively breaking through the limitations of traditional methods, and providing a new tool that is efficient, accurate and widely applicable for pork quality evaluation and related research.

[0088] The initial copy numbers of four muscle fiber type-specific marker genes are predicted by the model, and the composition ratio of the four muscle fiber types in the sample is calculated, which has important scientific research and clinical application value. This can be used for muscle physiology and pathology research, helping to understand muscle function characteristics and diagnose muscle diseases; in sports science, optimizing training plans for athletes and evaluating sports performance; in rehabilitation medicine, evaluating rehabilitation effects and developing personalized rehabilitation programs; in drug research and development, evaluating the mechanism of action and efficacy of drugs on muscles; in aging and metabolism research, exploring the effects of aging on muscle function and the role of muscles in energy metabolism; in gene expression and regulation research, revealing the role of genes in muscle development and function; in personalized medicine, developing personalized treatment plans for patients.

[0089] The muscle fiber type analysis system based on Raman spectroscopy provided by the application is described below, and the muscle fiber type analysis system based on Raman spectroscopy described below can be mutually corresponding with the muscle fiber type analysis method based on Raman spectroscopy described above.

[0090] The application provides a muscle fiber type analysis system based on Raman spectrum.

[0091] The data acquisition module is used for acquiring Raman spectrum data of the sample to be measured, the sample to be measured being muscle tissue of any part of a mammal.

[0092] The analysis module is used for obtaining initial copy numbers of MYH7, MYH2, MYH1 and MYH4 respectively encoding four muscle fiber types of type I, type IIa, type IIx and type IIb in the sample to be measured according to the Raman spectrum data of the sample to be measured and the muscle fiber type analysis model obtained by the construction method of the muscle fiber type analysis model based on Raman spectrum, and further obtaining a composition ratio of the four muscle fiber types in the sample to be measured.

[0093] Figure 7 An example of an entity structure diagram of an electronic device is shown in the figure. Figure 7 As shown in the figure, the electronic device can include a processor 810, a communications interface 820, a memory 830 and a communications bus 840, wherein the processor 810, the communications interface 820 and the memory 830 complete mutual communication through the communications bus 840. The processor 810 can call logical instructions in the memory 830 to perform the following steps:

[0094] The data acquisition module is used for acquiring Raman spectrum data of the sample to be measured, the sample to be measured being muscle tissue of any part of a mammal.

[0095] The analysis module is used for obtaining initial copy numbers of MYH7, MYH2, MYH1 and MYH4 respectively encoding four muscle fiber types of type I, type IIa, type IIx and type IIb in the sample to be measured according to the Raman spectrum data of the sample to be measured and the muscle fiber type analysis model obtained by the construction method of the muscle fiber type analysis model based on Raman spectrum, and further obtaining a composition ratio of the four muscle fiber types in the sample to be measured.

[0096] Further, the logic instructions in the memory 830 described above can be implemented in the form of software functional units and sold or used as standalone products, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or partially contribute to the prior art, or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0097] In another aspect, the present application also provides a computer program product, which comprises a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program can be executed by a processor to enable a computer to perform the following steps:

[0098] Obtaining Raman spectrum data of a to-be-tested sample, the to-be-tested sample being muscle tissue of any part of a mammal;

[0099] According to the Raman spectrum data of the to-be-tested sample, the initial copy numbers of MYH7, MYH2, MYH1 and MYH4 respectively encoding four types of muscle fiber types of type I, type IIa, type IIx and type IIb in the to-be-tested sample are obtained by using the muscle fiber type analysis model obtained by the method for constructing a muscle fiber type analysis model based on Raman spectrum, and then the composition ratio of the four types of muscle fiber types in the to-be-tested sample is obtained.

[0100] In another aspect, the present application also provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to perform the following steps:

[0101] Obtaining Raman spectrum data of a to-be-tested sample, the to-be-tested sample being muscle tissue of any part of a mammal;

[0102] According to the Raman spectrum data of the to-be-tested sample, the initial copy numbers of MYH7, MYH2, MYH1 and MYH4 respectively encoding four types of muscle fiber types of type I, type IIa, type IIx and type IIb in the to-be-tested sample are obtained by using the muscle fiber type analysis model obtained by the method for constructing a muscle fiber type analysis model based on Raman spectrum, and then the composition ratio of the four types of muscle fiber types in the to-be-tested sample is obtained.

[0103] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0104] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and the necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0105] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for constructing a muscle fiber type analysis model based on Raman spectroscopy, characterized in that, include: Raman spectral data and initial DNA copy data of myofiber encoding genes were obtained from training samples. The training samples were muscle tissue from any part of pig. The myofiber encoding genes included MYH7, MYH2, MYH1, and MYH4, which encode four myofiber types: type I, type IIa, type IIx, and type IIb, respectively. The Raman spectral data included characteristic peaks at one or more of the following Raman shifts: 828 cm⁻¹ -1 850cm -1 935cm -1 1002cm -1 1028cm -1 1122cm -1 1312cm -1 1447cm -1 1647cm -1 2926cm -1 ; Based on the Raman spectral data of the training samples and the initial DNA copy data of the myofibril coding gene, the long short-term memory network algorithm is used to enable the model to learn the relationship between the Raman spectral data of the training samples and the initial DNA copy data of the myofibril coding gene, thereby constructing a myofibril type analysis model. The acquisition of Raman spectral data of training samples and initial DNA copy data of myofibril-encoding genes includes: Raman spectral data of type I muscle were preprocessed using multivariate scattering correction and second derivative. The Raman spectral data of type IIa muscle were preprocessed using the second derivative. Raman spectral data of type IIb muscle were preprocessed using multivariate scattering correction and second derivative. Raman spectral data of type IIx muscle were preprocessed using multivariate scattering correction and second derivative. The step of constructing a myofiber type analysis model by using a long short-term memory network algorithm to teach the model the relationship between the Raman spectral data of the training samples and the initial DNA copy data of the myofiber encoding gene, based on the Raman spectral data of the training samples and the initial DNA copy data of the myofiber encoding gene, includes: Based on the preprocessed Raman spectral data of the training samples, the Raman spectral characteristic wavelength data of the training samples are obtained by using a competitive adaptive reweighting algorithm and / or removal of non-information variables. Based on the Raman spectral characteristic wavelength data of the training samples and the initial DNA copy data of the myofiber encoding gene, the long short-term memory network algorithm is used to enable the model to learn the relationship between the Raman spectral characteristic wavelength data of the training samples and the initial DNA copy data of the myofiber encoding gene, thereby constructing a myofiber type analysis model.

2. The method for constructing a muscle fiber type analysis model based on Raman spectroscopy according to claim 1, characterized in that, The method involves using Raman spectral data of training samples and initial DNA copy data of myofibril-encoding genes, employing a long short-term memory network algorithm to teach the model the relationship between the Raman spectral data of training samples and the initial DNA copy data of their myofibril-encoding genes, thereby constructing a myofibril type analysis model, including: During training, the performance of the muscle fiber type analysis model is evaluated using model performance evaluation metrics, including the validation set determination coefficient and / or root mean square error.

3. A method for analyzing muscle fiber types based on Raman spectroscopy, characterized in that, include: Obtain Raman spectral data of the sample to be tested, which is muscle tissue from any part of a pig; Based on the Raman spectral data of the sample to be tested, the initial copy number of MYH7, MYH2, MYH1, and MYH4, which encode four types of muscle fibers (type I, type IIa, type IIx, and type IIb) respectively, is obtained by constructing the muscle fiber type analysis model based on Raman spectroscopy as described in claim 1 or 2.

4. The method for analyzing muscle fiber type based on Raman spectroscopy according to claim 3, characterized in that, Also includes: Based on the initial copy numbers of MYH7, MYH2, MYH1, and MYH4, which encode the four types of muscle fibers (type I, type IIa, type IIx, and type IIb) respectively, in the sample to be tested, the composition ratio of the four types of muscle fibers in the sample to be tested is obtained.

5. A muscle fiber type analysis system based on Raman spectroscopy, characterized in that, include: The data acquisition module is used to acquire Raman spectral data of the sample to be tested, which is muscle tissue from any part of a pig. The analysis module is used to: obtain the initial copy number of MYH7, MYH2, MYH1, and MYH4, which encode four types of muscle fibers (type I, type IIa, type IIx, and type IIb), respectively, in the sample based on the Raman spectral data of the sample to be tested, using the muscle fiber type analysis model obtained by the construction method of the Raman spectroscopy-based muscle fiber type analysis model as described in claim 1 or 2.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for constructing a muscle fiber type analysis model based on Raman spectroscopy as described in claim 1 or 2 and / or the method for muscle fiber type analysis based on Raman spectroscopy as described in claim 3 or 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for constructing a muscle fiber type analysis model based on Raman spectroscopy as described in claim 1 or 2 and / or the method for muscle fiber type analysis based on Raman spectroscopy as described in claim 3 or 4.

Images