Method for quantitatively analyzing beta-folding structure in silk fibroin material based on machine learning

Abstract

The invention provides a method for quantitatively analyzing and comparing beta-folded structures in silk fibroin materials based on machine learning, and relates to the field of material characterization technologies and biomedical materials. The method comprises a small-angle X-ray scattering detection experiment and machine learning model analysis based on big data. The SAXS experiment needs to be carried out on equipment meeting the detection sensitivity requirement, and scattered light intensity I data is obtained; after q-I data of an SAXS experiment of an unknown silk protein material is input into the machine learning model, the shape and the feature size of a beta-folding structure are output. The method provided by the invention has higher convenience; sample pretreatment is not needed, and a statistical result of structural characteristics in a sample can be visually reflected; the lt can be reflected; according to the method, the light scattering result is analyzed innovatively through a machine learning method according to the nano-structure information within the scale range of 500 nm, and the detailed structure information of the material is deduced more effectively, so that the material design is optimized.

Description

Technical Field

[0001] The present invention relates to the field of material characterization and biomedical material technology, and in particular to a quantitative analysis method for the microstructure of a silk fibroin material based on machine learning. Background Art

[0002] The molecular assembly structure of proteins in vivo and in vitro has a significant impact on their performance and function. In the biomedical field, structural proteins such as collagen and silk fibroin have been developed as biomedical materials, which show excellent biocompatibility. However, there are multiple complex molecular interactions such as hydrogen bonds, van der Waals forces, and hydrophobic interactions in the protein topological structure. At the same time, the protein structure is sensitive to factors such as heat and force. The precise and quantitative analysis of its microstructure is still a bottleneck restricting its application.

[0003] Silk fibroin is a protein extracted from natural silk, with typical β-folding, helical and random coil conformational structures. Films, solutions, gels and scaffolds made of silk fibroin have become an important category of biomedical materials. Most of the amino acid sequence of silk fibroin is a GAGAGS repeating module, with a repeat count of 432 times. When two GAGAGS segments are arranged in parallel and locked by hydrogen bonds, an antiparallel β-folding structure is formed, which can be considered as the smallest repeating unit of the β-folding structure (see Appendix). Figure 1 ). In silk fibroin materials, such as solutions / gels, the size of the β-folded structure depends on the preparation conditions and will evolve over time. Different silk fibroin materials will contain β-folded structures of different sizes, which can be considered as sheets or cuboids stacked with the smallest repeating units. Figure 2 The long side of the cuboid is the a direction where hydrogen bonds are formed between molecular chains, the wide side is the b direction where the sheets are stacked, and the height is the c direction where the chains extend. Currently, information on chain spacing and interlayer spacing has been obtained by wide-angle X-ray diffraction (WAXD), but there is still a lack of quantitative analysis methods for the β-folded structures of sheets and cuboids with sizes ranging from 10nm to 500nm.

[0004] Small-angle X-ray diffraction (SAXS) involves exposing a sample to high-energy, hard X-rays with very small wavelengths, i.e., 0.07–0.2 nm, and photon energies above ~10 keV. Such high energy allows the sample to be placed 3 to 4 m away from the X-ray source, i.e., at very small scattering angles, and still obtain sufficiently strong scattered light signals. Therefore, the SAXS method has become an important tool for solving protein structures. This method is applicable to solutions, gases, and solid samples, and is a non-invasive and non-destructive characterization technique. The same sample can produce a unique scattering signal determined by its internal structure, so the microscopic structure of the sample can be accurately and effectively inferred from the scattering results. Small-angle scattering includes SAXS and small-angle neutron scattering (SANS), and is commonly used to study the shape, structure, and interaction of proteins and polymers in solution.

[0005] Machine learning is a data-driven method that has significant advantages in solving multivariate nonlinear problems. Supervised learning in machine learning includes two major algorithms: regression and classification, which are used to predict continuous and discrete results respectively, and can effectively assist research topics in the field of natural sciences. These mainstream algorithms include decision trees, random forests, support vector machines (SVM), naive Bayesian Bayes), ensemble learning and multi-layer perceptron (MLP). As a key technology in the field of artificial intelligence, machine learning has had a very significant impact in computer graphics, natural language processing, content generation and other fields. In the field of materials science, the application of machine learning models based on big data is also becoming more and more widespread. By building a machine learning model to predict the size of the characteristic β-folded structure in silk fibroin, the experimental cost can be effectively reduced and the material development and optimization process can be accelerated. Summary of the invention

[0006] The present invention intends to protect a method for quantitatively analyzing the β-folded structure in silk fibroin materials. The method uses SAXS as an experimental technical means, combined with machine learning methods and models, to obtain the shape and size information of the β-folded structure in silk fibroin solutions, gels and other materials. This quantitative method has important reference value for constructing the structure-activity relationship of silk fibroin materials, optimizing the preparation process of silk fibroin materials, and expanding its application scenarios.

[0007] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0008] The specific steps are as follows (Attachment Figure 3 ):

[0009] (1) Data collection and processing. The present invention measures the scattering curve data (i.e., data on the change of scattering intensity I with scattering vector q) of a large number of silk fibroin solutions through SAXS experiments, and performs preliminary processing on the scattering data. Then, the β-pleated structure size of the measured sample is characterized based on an independent method. Assuming that the β-pleated structure is a spherical particle, in the small scattering angle region, the scattering signal satisfies Guinier's law. Therefore, the present invention uses Guinier's law to calculate the β-pleated structure size of the sample, and uses this calculated value as the benchmark size of the β-pleated structure. Based on this method, the data set required for training the machine learning model is obtained.

[0010] (2) Feature extraction and size calibration. The present invention extracts characteristic indicators of SAXS scattering curves through a variety of numerical analysis methods. These characteristics are related to the physical and chemical processes of protein β-folding in silk protein solution, and also reflect the morphological characteristics of β-folding structure to a certain extent. Related features include but are not limited to:

[0011] a) Scattering curve characteristics: such as the slope of the scattering curve, maximum / minimum scattering intensity, etc.

[0012] b) Radial distribution function: The characteristic peak value is obtained by Fourier transform calculation.

[0013] c) Structure factor S(q): The structure factor reflects the medium- and short-range order of atoms or molecules in the system and can provide information about the correlation between atoms.

[0014] Through a variety of methods, all features involved in modeling are standardized to ensure the quality and consistency of the features.

[0015] (3) AI model training. Select appropriate traditional machine learning models for training. Optional models include:

[0016] a) Random Forest: Random Forest is an ensemble learning algorithm composed of multiple decision trees. It can construct different decision trees through multiple random sampling and take the average value for prediction. Random Forest has the advantages of resisting overfitting, processing high-dimensional data and nonlinear relationships, and is suitable for the problem of predicting the size of β-folded structure.

[0017] b) Support Vector Machine (SVM): Support vector machine is a machine learning model for classification and regression. Support vector machine achieves efficient prediction by finding the optimal hyperplane, especially by using kernel functions (such as RBF kernel) to deal with nonlinear problems, support vector machine can capture the complex relationship between β-folding structure size and structure.

[0018] When training the model, first, all the collected data is divided into training set and test set according to a certain ratio (such as 8:2); then, the model is fitted using the training set data, and the test set is used for preliminary evaluation and parameter adjustment. During the model training process, cross-validation or Bayesian optimization methods are used to search for the optimal combination of model hyperparameters. Model hyperparameters include: the number and depth of decision trees in random forests, and the kernel function selection and penalty parameter C in SVM.

[0019] The model is evaluated by the mean square error (MSE) and goodness of fit (R2).

[0020]

[0021]

[0022] Where n represents the total number of protein sample datasets, and i are the model prediction value and theoretical calculation value of the i-th data, respectively. All predicted values The average value of .

[0023] (4) AI model evaluation and optimization. The machine learning model obtained through training can be used to predict the size of the β-folded structure in the silk protein solution. In order to evaluate the effect of the model, it is necessary to conduct SAXS experiments on new solution samples, process the scattering curve data of the new samples in the same way as the training data set, and use the extracted features and the trained model to predict the size of the β-folded structure of the new samples. Compare the model prediction results with the results calibrated by Guinier's law. If the results reach a certain level of accuracy, it means that the AI ​​model has achieved an industrial application level. The next step is to deploy the trained model in production environment applications to effectively assist the structural design and synthesis experiments of silk protein and reduce the cost and time of experimental measurements. If the accuracy of the results does not reach a usable level, it is necessary to continuously improve the model by modifying the model training parameters, increasing the amount of data, adjusting feature extraction, etc.

[0024] The present invention proposes a method for quantitatively analyzing the shape and size information of β-folded structure in silk fibroin material based on machine learning. The method provides the shape and size information of β-folded structure based on SAXS experimental data and a machine learning model.

[0025] The silk protein material includes but is not limited to macroscopic forms such as silk fibroin solution, gel, film, powder, sponge, block, fiber, etc., and a suitable SAXS sample stage / injection system is selected according to the material form.

[0026] In the SAXS experiment, the scattering angle range is 0.1-5°, the scattering vector is q=4πsinθ / λ, and the β-folded structure in the silk fibroin material serves as a scatterer.

[0027] The machine learning model takes the SAXS experimental results, i.e., the Iq data curve, as input, and predicts the shape and size of the β-folded structure in the silk fibroin material by extracting the mathematical features of the curve.

[0028] The machine learning model is based on a training set containing >100 sets of data, and one machine learning model with the highest prediction accuracy is selected from the AI ​​models;

[0029] The >100 data sets include Iq curve data obtained by performing SAXS experimental tests on >100 samples, the gyration radius of the scatterer calculated based on Guinier's law, and the power law of q and I in a specific range. This is specifically embodied in the embodiments of the present invention.

[0030] The machine learning model extracts characteristic data for model training through the scattering curve characterized by the SAXS experiment, and uses the gyration radius fitted according to Guinier's law as the target of model prediction.

[0031] The machine learning model uses k-fold cross validation to improve the generalization ability of the model and uses Bayesian optimization to search for the optimal model hyperparameters.

[0032] The machine learning model described above uses mean square error and R 2 Values ​​are used to evaluate the effectiveness of machine learning models.

[0033] The invention has a wide range of applications, including obtaining the mechanical structure-activity relationship of a material or the relationship between the biological performance and structure of a material or assisting in optimizing a process.

[0034] For example, for silk fibroin materials prepared under specific process conditions, after completing the quantitative analysis of the β-folded structure, their mechanical performance indicators, such as modulus, can be further measured to obtain the mechanical structure-activity relationship of the material; for silk fibroin materials prepared under specific process conditions, after completing the quantitative analysis of the β-folded structure, their biological performance indicators, such as the in vivo degradation / absorption rate based on the rat subcutaneous animal model, can be further measured to obtain the relationship between the biological performance and structure of the material; in the process of new material research and development, the relationship between process conditions, such as solution concentration, temperature, etc., and β-folded structure parameters is constructed to assist in optimizing the process.

[0035] Beneficial effects of the present invention:

[0036] (1) The present invention proposes a submicron-scale measurement method that can quantify the size of the β-fold structure in the silk fibroin solution / gel. The present invention uses SAXS as a detection method, and ultimately can obtain the stacking of β-folds in the material and the corresponding size when the SAXS signal is obtained.

[0037] (2) After accumulating a large amount of SAXS detection data of various silk fibroin solutions / gels, the present invention obtains the β-folding size of the corresponding silk fibroin solution / gel according to the corresponding law, and can establish a deep learning model based on the accumulated data, and can infer the β-folding size of the silk fibroin solution / gel to be tested based on the model.

[0038] (3) After the β-folding structure size of the silk fibroin solution / gel system is obtained through simulation and calculation based on the machine learning model, the physicochemical properties of the related materials can be inferred based on this information, providing powerful assistance for the efficient design of silk fibroin materials.

[0039] (4) Compared with traditional experimental methods, the machine learning algorithm of the present invention can complete the prediction of the β-fold structure size of a large number of protein sequences within a few minutes, greatly shortening the research cycle and reducing the overall cost of research investment. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is the repeating sequence GAGAGS in silk fibroin, which is the smallest repeating unit of the β-folding structure;

[0041] Figure 2 Schematic diagram of the sheet-like and rectangular β-folded structure formed by stacking of the minimum repeating units in silk fibroin;

[0042] Figure 3 A flowchart for the application of machine learning-assisted β-sheet structure size prediction model;

[0043] Figure 4 Iq signal diagram of SAXS test for pure mixed solvent system;

[0044] Figure 5 Iq signal diagram of SAXS test for pure mixed solvent system;

[0045] Figure 6 Iq signal diagram of two scaffolds subjected to SAXS test;

[0046] Figure 7 Silk fibroin sample pictures;

[0047] Figure 8 Iq signal diagram of SAXS test of silk fibroin aqueous solution. DETAILED DESCRIPTION

[0048] The present invention is further described below in conjunction with specific embodiments, but the protection scope of the present invention is not limited thereto:

[0049] Embodiment 1:

[0050] After using hexafluoroisopropanol (HFIP) to dissolve silk fibroin to form a completely transparent homogeneous solution, water was added to induce conformational change in the silk protein chain. The mass ratio of the three phases in the solution was: SF: HFIP: H2O = 0.4%: 61.3%: 38.3%. After 12 hours of physical gelation, a micro-sized β-folding structure was generated in the system. Relying on the Xeuss 3.0 platform of Xenocs, a liquid window was selected to place the sample, and two detectors were set at a distance of 1798mm and 300mm to cover a wide range of q values. SAXS tests were performed on the heterogeneous solution system containing β-folding structure and the simple mixed solvent system, and Iq signals were obtained (see Appendix). Figure 4 Assuming that the β-folded structure is a spherical particle, in the low q value (small scattering angle) region, the SAXS curve is in the Guinier region (qR g <1.3) Follow Guinier's law:

[0051]

[0052] By comparing logI to q 2 The linear fitting method is used to obtain the slope and obtain the radius of gyration R of the particle. g ~0.5nm, which is converted into a particle diameter D~1.3nm, which is consistent with the size of the β-folded structure of the minimum repeating structure predicted by the amino acid sequence mentioned above.

[0053] This embodiment proves through testing that the size of the β-folded structure can be obtained from the structural information of chemical bonds in theory, and at the same time, it follows Guinier's law.

[0054] Embodiment 2:

[0055] To explore the effect of concentration, different concentrations of silk fibroin HFIP solutions were prepared (as shown in Table 1) to obtain samples 2-1, 2-2, and 2-3. Relying on the Xeuss 3.0 platform of Xenocs, the liquid window was selected to place the sample, and the distances between the two detectors were set to 1798 mm and 300 mm to cover a wide range of q values. SAXS tests were performed on heterogeneous solution systems of different concentrations and pure mixed solvent systems to obtain Iq signals (see Appendix). Figure 5 ). Assuming that the β-sheet structure is a spherical particle, the Guinier law of the SAXS curve in the small scattering angle region is used to calculate the size of the β-sheet structure in each sample. Select a suitable low q value space qR g<1.3, and ensure that the data does not contain noise or areas with strong background scattering signals; then, the Iq curve is processed and lnI(q) and q are calculated. 2 Third, use the least squares method to find the relationship between lnI(q) and q 2 Perform linear fitting and calculate the slope m and intercept c of the fitting line; finally, according to the formula Calculate the radius of gyration R from the slope obtained from the fit g .

[0056] Table 1 β-sheet size corresponding to different concentrations of silk fibroin solution

[0057] Serial number name concentration SAXS theoretical calculation of β-sheet size 2-1 Silk fibroin solution 0.4％ 2.1nm 2-2 Silk fibroin solution 1.9％ 3.4nm 2-3 Silk fibroin solution 3.6％ 4.8nm

[0058] This example proves that: by characterizing silk fibroin solutions of different concentrations through SAXS, the β-folding size data is obtained by calculation through the Guinier formula, and the β-folding sizes corresponding to silk fibroin solutions of different concentrations are different.

[0059] Embodiment 3:

[0060] A silk protein scaffold with a physical and chemical double cross-linked structure was prepared by cryogel method using a small molecule cross-linking agent and compared with a pure silk protein scaffold sample. Relying on the 1W2A beamline of the Beijing Synchrotron Radiation Light Source, SAXS tests were performed on the two scaffolds to obtain Iq signals (see Appendix Figure 6 ). The β-folding structure size is calculated by combining SAXS theory, namely Guinier's law, as shown in Table 2.

[0061] Table 2 β-sheet size of silk fibroin scaffolds prepared by different cross-linking methods

[0062]

[0063]

[0064] This example demonstrates that the β-folding sizes of silk fibroin scaffolds prepared by different cross-linking methods are different, as characterized by SAXS and calculated by Guinier's formula.

[0065] Example 4: This example establishes a machine learning method and model, characterizes different silk protein materials through SAXS, inputs the scattering curve of SAXS into the trained machine learning model, and obtains the size information of the β-folded structure.

[0066] (1) Prepare 120 silk fibroin samples, including solutions, gels, films, powders, sponges, blocks, fibers (e.g. Figure 7 ).

[0067] (2) The scattering curve data (including scattering intensity I and scattering vector q) of all silk protein samples were measured through SAXS experiments. These Iq curves are used to characterize the internal structure of the samples and provide microscopic information about particle size, shape, distribution, etc.

[0068] (3) The acquired Iq curve data needs to be preprocessed so that it can be subsequently input into the machine learning model. The specific preprocessing steps include:

[0069] -Normalization processing: Normalize all Iq curve data to make the I values ​​of different samples within the same magnitude range to prevent unstable model training caused by too large a value range.

[0070] -Since the number and interval of Iq curve data points may be different for different samples, the q value needs to be uniformed by interpolation method to make the q value of each sample uniform under the same range and step size.

[0071] (4) According to Guinier's law, select the Iq curve data in the low q region and obtain the radius of gyration R by fitting the Guinier formula g , this parameter is an important physical quantity that characterizes the geometric dimensions of the β-fold scatterer.

[0072] (5) Based on the preprocessed Iq curve and gyration radius R g , construct features for machine learning modeling. The features of each sample include:

[0073] -Iq curve characteristics: I values ​​corresponding to a set of standardized q values.

[0074] - Scattering curve characteristics: maximum and minimum values ​​of scattering intensity.

[0075] -Radial distribution function: The positions of all characteristic peaks are obtained by Fourier transforming the Iq curve.

[0076] -Structure factor: The structure factor S(q) of the sample is separated from the experimental data. This feature can provide information about the correlation between atoms.

[0077] (6) The obtained silk protein sample database was split into 80% training set and 20% test set, and a machine learning model was established using the following five methods: random forest, support vector machine, naive Bayes, multilayer perceptron, and XGBoost. During the model training process, the cross-validation method was used to update the training set and test set to improve the generalization ability of the model. The model hyperparameters were adjusted to ensure that the model achieved the best results.

[0078] (7) Calculate the performance indicators of each model on the test set, including:

[0079] -Mean Squared Error (MSE), which is used to measure the difference between the model prediction value and the value calculated by Guinier formula.

[0080] -R 2 Value: reflects the goodness of fit of the model to the data, R 2 The closer the value is to 1, the better the model fit is.

[0081] (8) Among the five machine learning models, the model with better model performance indicators is selected as the model used in the prediction stage. In this embodiment, the random forest model shows the highest prediction accuracy, with the lowest MSE and R 2 The value is close to 1, so the random forest model is selected as the final model.

[0082] Table 3 β-folding size range of silk fibroin materials in solution, gel, film, powder, sponge, bulk and fiber forms using machine learning models

[0083] Serial number name β-sheet size range 1 Solution 1-8nm 2 gel 3-10nm 3 membrane 30-130nm 4 powder 1-2nm 5 Sponge bracket 10-14nm 6 Block 50-130nm 7 fiber 100-300nm

[0084] This example demonstrates that the β-folding size of silk fibroin materials in the form of solutions, gels, films, powders, sponges, blocks, and fibers can be effectively measured using a machine learning model.

[0085] Embodiment 5:

[0086] This example verifies the machine learning method and model. A 2.4% silk fibroin aqueous solution was prepared to obtain the Iq curve of SAXS, and a theoretical calculation was performed based on Guinier's law to obtain R g =4.8nm(Attached Figure 8 ). The Iq curve was input into the trained machine learning model, and the size of the β-folded structure was obtained to be 4.9nm, with a prediction accuracy of 98%.

Claims

1. A method for quantitatively analyzing the shape and size information of β-folded structure in silk fibroin material based on machine learning, wherein the method provides the shape and size information of β-folded structure based on SAXS experimental data and machine learning model, wherein SAXS represents small angle X-ray scattering.

2. According to claim 1, a method for quantitatively analyzing the shape and size information of β-folded structure in silk fibroin material based on machine learning, the silk fibroin material includes but is not limited to macroscopic forms such as silk fibroin solution, gel, film, powder, sponge, block, fiber, etc., and a suitable SAXS sample stage / injection system is selected according to the material form.

3. According to a method for quantitatively analyzing the shape and size information of β-folded structures in silk fibroin materials based on machine learning in claim 1, the scattering angle θ in the SAXS experiment ranges from 0.1 to 5°, the scattering vector is q=4Πsinθ / λ, λ is the wavelength of x-rays, and the β-folded structure in the silk fibroin material serves as a scatterer.

4. According to the method for quantitatively analyzing the shape and size information of β-folded structure in silk fibroin material based on machine learning in claim 1, the machine learning model is based on SAXS experimental results, i.e., Iq, Iq 2 and other data structure information as input, and by extracting the mathematical features of the curve, the shape and size of the β-folded structure in the silk protein material are predicted.

5. According to the method for quantitatively analyzing the shape and size information of β-folded structure in silk fibroin material based on machine learning in claim 4, the machine learning model is based on a training set containing >100 groups of data, and a machine learning model with the highest prediction accuracy is selected from 5 machine learning models; The >100 data sets include Iq curve data obtained by SAXS experimental testing on >100 samples, the gyration radius of the scatterer calculated based on Guinier's law, and the power law of q and I in a specific range.

6. According to a method for quantitatively analyzing the shape and size information of β-folded structures in silk fibroin materials based on machine learning in claim 1, the machine learning model extracts characteristic data for model training through the scattering curve characterized by SAXS experiment, and uses the gyration radius fitted according to Guinier's law as the target of model prediction.

7. According to the method for quantitatively analyzing the shape and size information of β-folded structure in silk fibroin material based on machine learning in claim 1, the machine learning model uses k-fold cross-validation to improve the generalization ability of the model, and uses Bayesian optimization to search for the optimal model hyperparameters.

8. According to the method for quantitatively analyzing the shape and size information of β-folded structure in silk fibroin material based on machine learning in claim 1, the machine learning model uses mean square error and R 2 Values ​​are used to evaluate the effectiveness of machine learning models.

9. The method for quantitatively analyzing the shape and size information of β-folded structure in silk fibroin material based on machine learning according to claim 1, comprising the following steps: (1) Data collection and processing: SAXS experiments are used to measure the scattering curve data of a large number of silk fibroin solutions, that is, the data on the change of scattering intensity I with the scattering vector q. The scattering data are preliminarily processed; then, the β-folding structure size of the measured sample is characterized based on an independent method. (2) Feature extraction and size calibration: The characteristic indicators of SAXS scattering curves are extracted through various numerical analysis methods. The relevant features include but are not limited to: a) Scattering curve characteristics: including the slope of the scattering curve and the maximum / minimum value of the scattering intensity; b) Radial distribution function: The characteristic peak value is obtained by Fourier transform calculation; c) Structure factor S(q): The structure factor reflects the medium- and short-range order of atoms or molecules in the system and can provide information about the correlation between atoms; Standardize all features involved in modeling to ensure the quality and consistency of features; (3) AI model training, including selecting appropriate traditional machine learning models for training. Optional models include: a) Random Forest: Random Forest is an ensemble learning algorithm consisting of multiple decision trees. It can construct different decision trees through multiple random sampling and take the average value for prediction; b) Support Vector Machine (SVM): Support vector machine is a machine learning model for classification and regression. Support vector machine achieves efficient prediction by finding the optimal hyperplane, especially by using kernel functions (such as RBF kernel) to deal with nonlinear problems, support vector machine can capture the complex relationship between β-folding structure size and structure. When training the model, first, all the collected data is divided into a training set and a test set according to a certain ratio (such as 8:2); then, the model is fitted using the training set data, and the test set is used for preliminary evaluation and parameter adjustment. During the model training process, cross-validation or Bayesian optimization methods are used to search for the optimal model hyperparameter combination; model hyperparameters include: the number and depth of decision trees in the random forest, and the kernel function selection and penalty parameter C in the SVM; The model was evaluated by MSE and goodness of fit R 2 Evaluate the trained machine learning model: Where n represents the total number of protein sample datasets, and i are the model prediction value and theoretical calculation value of the i-th data, respectively. All predicted values The average value of (4) AI model evaluation and optimization; the machine learning model obtained through training can be used to predict the size of the β-folded structure in the silk protein solution; to evaluate the model effect, it is necessary to conduct SAXS experiments on new solution samples, process the scattering curve data of the new samples in the same way as the training data set, and use the extracted features and the trained model to predict the size of the β-folded structure of the new samples. Compare the model prediction results with the results of Guinier's law calibration. If the results reach a certain level of accuracy, it means that the AI ​​model has achieved the level of industrial application. The next step is to deploy the trained model in the production environment, effectively assist the structural design and synthesis experiments of silk protein, and reduce the cost and time of experimental measurement. If the accuracy of the results does not reach the usable level, it is necessary to continuously improve the model by modifying the model training parameters, increasing the amount of data, adjusting feature extraction, etc.

10. The method for quantitatively analyzing the shape and size information of β-folded structure in silk fibroin material based on machine learning according to any one of claims 1 to 9, wherein the use of the method comprises: Obtain the mechanical structure-activity relationship of materials, the relationship between the biological properties of materials and their structure, or assist in optimizing the process.

Images