Artificial Intelligence Algorithm-Based Method for Calculating the Molecular Weight of Biomacromolecular Materials

Abstract

A method for calculating the molecular weight of biopolymer substances utilizing AI-based computation employs ionic liquids as the dissolving medium, enabling biomacromolecular substances to form distinct single biomacromolecular chains upon being dissolved. Subsequently, rheological techniques are applied to gather rheological data for the biopolymer in ionic liquid solutions. The Rouse model is utilized as a preferred model for characterizing the properties of polymer solution. Through extensive data collection and AI algorithm-assisted nonlinear regression analysis, critical parameter values are derived and employed for calculating the molecular weight of the substance under examination.

Claims

1-10. (canceled)

11. A method for calculating the molecular weight of biopolymer materials based on AI algorithms and the following steps: S1: Sample Preparation: Take the biopolymer material sample to be tested, dissolve it in an ionic liquid to obtain the sample solution, which is the ionic liquid solution of the biopolymer material. S2: Sample Testing: Place the sample prepared in step S1 on a rheometer for sample testing and calculate the required data. S3: Establish AI Algorithm to Evaluate Rouse Model: Using common AI model optimization algorithms, establish an algorithm to assess the quality of the Rouse model's prediction results. S4: Fit the Rouse Model to the Biopolymer solution system. S5: Use the AI-optimized Rouse model to calculate the molecular weight of biopolymers.

12. The method for calculating the molecular weight of a biological macromolecule material based on an AI algorithm according to claim 11, in which the particular technique for sample preparation in step S1 is as follows: a) Preparation of the dissolution system: The dried biopolymer material sample is dissolved into the ionic liquid and dissolved at room temperature; b) Desiccation and water removal; c) Decompression and dissolution, in step S1, the biopolymer material includes silk fibroin, hyaluronic acid, collagen, recombinant collagen, and fibroin protein, the silk fibroin can be selected from options like silkworm silk, spider silk, or tussah silk, the ionic liquid used in step S1 is the combination of [AMIm]Cl and [HMIm]HSO.sub.4.

13. The method for calculating the molecular weight of the biomacromolecular material based on the AI algorithm according to claim 11, wherein the concentration of the silk fibroin in the ionic liquid solution of the silk protein of the biopolymer material prepared in the step S1 is 0.1%-50%, 1%-20%, or 5-15%.

14. The method for calculating the molecular weight of the biomacromolecular material based on the AI algorithm according to claim 12, wherein the concentration of the silk fibroin in the ionic liquid solution of the silk protein of the biopolymer material prepared in the step S1 is 0.1%-50%, 1%-20%, or 5-15%.

15. The method for calculating the molecular weight of a biological macromolecule material based on an AI algorithm according to claim 12, wherein the freeze-drying method is selected in the step S1 to dry and remove water. Specifically, the sample is placed in liquid nitrogen for freezing and is placed in a freeze-dryer to prepare the lyophilized powder. It was then taken out after being fully dehumidified and sealed for storage at room temperature.

16. The approach for determining the molecular weight of biopolymer materials using AI algorithms according to claim 12 is defined by the particular vacuum dissolution technique: the blend of silk fibroin and ionic liquid made in step S1(b) is warmed while being stirred in an oil bath, then undergoes vacuum distillation with an oil pump to get rid of any remaining water and to eradicate bubbles, continuing the heating process until the biopolymer is fully dissolved, the temperature of oil bath is 0-180 C., and the vacuum range is-0.01 MPa to-0.5 MPa, to remove any residual water and eliminate bubbles. The heating process persists as the biopolymer is fully dissolved. Upon ceasing the heating and mixing process, the solution is permitted to cool, subsequently transforming into a consistent and clear liquid, the obtained biopolymer ionic liquid solution is sealed, and stored in a dry environment at room temperature for later application.

17. The approach for determining the molecular weight of biopolymer materials using AI algorithms according to claim 14 is defined by the particular vacuum dissolution technique: the blend of silk fibroin and ionic liquid made in step S1(b) is warmed while being stirred in an oil bath, then undergoes vacuum distillation with an oil pump to get rid of any remaining water and to eradicate bubbles, continuing the heating process until the biopolymer is fully dissolved, the temperature of oil bath is 0-180 C., and the vacuum range is 0.01 MPa to 0.5 MPa, to remove any residual water and eliminate bubbles. The heating process persists as the biopolymer is fully dissolved. Upon ceasing the heating and mixing process, the solution is permitted to cool, subsequently transforming into a consistent and clear liquid, the obtained biopolymer ionic liquid solution is sealed, and stored in a dry environment at room temperature for later application.

18. The technique for determining the molecular weight of biopolymer substances using AI algorithms according to any of claims 11-17 is distinguished by the particular approach for sample examination in step S2: a) Storage Modulus and Loss Modulus Testing: Parallel plates are chosen, and safeguard the testing environment by introducing nitrogen gas through a temperature-regulated cover during the process; employ linear dynamic elasticity measurements where the strain amplitude is maintained below 50%, guaranteeing the linearity of the storage and loss moduli across the frequency sweep spectrum (from 110.sup.2 rad/s-3010.sup.2 rad/s), conduct frequency sweeps at varying temperatures (0 C., 10 C., 20 C., and 30 C.) to generate storage and loss modulus curves corresponding to different temperatures; b) Viscosity Assessment: Choose parallel plates and shield the test from the nitrogen gas stream at a steady temperature; perform steady-state tests by scanning the shear rate from low to high, within the range of 10.sup.5-10.sup.5 s.sup.1, and record the viscosity values by the instrument.

19. The method for calculating the molecular weight of biopolymer materials utilizing AI algorithms according to claim 11 is distinguished by the particular approach for constructing the AI algorithm to assess the Rouse model during step S3, which is outlined as follows: a) Set a suitable error threshold .sub.th, where the Rouse model's prediction error is considered acceptable if below the threshold .sub.th, suggesting that the Rouse model is adequately precise for characterizing the experimental polymer system; b) Set a reasonable initial learning rate .sub.0 to control the iteration speed of the algorithm; c) Employ gradient descent to continuously adjust the model parameters until the prediction error is below 0.01, the optimization algorithms of AI model in step S3 include Gradient Descent, Conjugate Descent, Adam, AdamGrad, and RMSProp.

20. The method for calculating the molecular weight of biopolymer materials based on AI algorithms according to claim 11 is distinguished by selecting the error threshold .sub.th in step S3 within the bracket of 0.001-0.25, and by opting for the learning rate .sub.0 within the confines of 0.0001-0.1.

21. The method for calculating the molecular weight of biopolymer materials based on AI algorithms according to claim 11, characterized in that the specific method for fitting the Rouse model to the biopolymer system in step S4 is as follows: a) It is assumed that the structure of the biomolecular system meets the Rouse model, and the molecular weight distribution of the system meets the normal distribution; The experimental system was modeled, and the mean M.sub.0 and standard deviation M.sub.0 of molecular weight distribution were initialized; b) {circumflex over (M)}.sub.0 and M.sub.0 were used to construct a normal distribution of molecular weight MN(M, M) [Formula (4)], and samples were taken according to this distribution to establish an initial simulated physical system; $\begin{array}{cc}f\left(M_i\right)=\frac{1}{M\sqrt{2}}{e}^{-\frac{\left(M_i-\stackrel{}{M}\right)}{2{}^2}}& \left(4\right)\end{array}$ c) Calculate the relaxation time .sub.ip of mode p based on the molecular weight distribution of the simulation physical system according to formula (3); d) Consider the contributions of all vibration modes p to the storage modulus G and dissipation modulus G and calculate G and G of the simulated physical system according to formulas (1) and (2), the polymer density p in solution, the zero shear viscosity .sub.0, and the solvent viscosity .sub.s used in the calculation should be consistent with experimental data; e) According to the simulated data obtained from the Rouse theoretical model, the relationship between log Glog and log Glog of the simulated physical system can be calculated. The error between the predicted values log G and log G of the Rouse model and the experimental values measured in step S2 can be solved by the root-mean-square formula. f) The error between the calculated log G and log G, which describes the degree of agreement between the Rouse model and the experimental results, the fitting process requires that the prediction results of Rouse model approximate to the experimental measurement results. Therefore, the error must be as small as possible, then the fitting of the Rouse model is an optimization problem, and the optimization objective is as follows: $\begin{array}{cc}\underset{\stackrel{\_}{M},M}{\min }\left(N\left(\stackrel{\_}{M},M\right)\right)& \left(5\right)\end{array}$ The error varies with the distribution of polymers. The AI algorithm described in Step S3 is used to optimize the objective function of formula (5). Compare the error threshold error of the error and AI algorithm, such as .sub.th, then optimize and update the normal distribution parameters M and M by AI algorithm, and return to process b) after updating, and re-calculate the Rouse model; If <.sub.th, it indicates that the Rouse model can accurately describe the experimental results.

22. The method for calculating the molecular weight of biopolymer materials based on AI algorithms according to claim 11, characterized in that in step S5, the molecular weight data includes weight-average molecular weight, number-average molecular weight, and molecular weight distribution, in step S5, the method for calculating the molecular weight using the AI-optimized Rouse model is: using the optimized Rouse model obtained from steps S1-S4, the molecular weight distribution N(M.sub.opt, M.sub.opt) of the simulation physical system is used to calculate the weight-average molecular weight M.sub.w, number-average molecular weight M.sub.n, and molecular weight distribution M.sub.w/M.sub.n.

Description

FIGURE LEGENDS

[0052] FIG. 1. The flow chart of calculating the molecular weight of a biopolymer system by Rouse model.

[0053] FIG. 2-5. The correlation between the storage modulus and the loss modulus, along with the shear frequency relative to silk proteins of varying molecular weights, is observed in examples 1-4 of this invention when examining the silk protein ionic liquid solution in oscillatory mode.

[0054] FIG. 6. The analyzed results of the molecular weights of different silk proteins in examples 1-4.

THE SPECIFIED EXAMPLES

[0055] The present invention is embodied with the following specified examples. However, it is understood by those versed in the field that these detailed examples do not define the boundaries of the present invention's protection.

Example 1

S1: Sample Preparation

[0056] a) Dissolution: Silkworm silk fibroin prepared by sodium carbonate degumming for 15 minutes was selected, and mulberry silk protein was mixed with ionic liquid 1-allyl-3-methylimidazole chloride salt, and the concentration of silk fibroin was controlled at 20%. [0057] b) Freeze-drying and dehydration: The sample is frozen in liquid nitrogen to fully solidify the ionic liquid, and placed in a freeze-dryer to be fully dehumidified and then taken out, sealed and thawed at room temperature. [0058] c) Dissolution under reduced pressure: The silk fibroin ionic liquid mixture prepared by step b is heated in an oil bath at a temperature of 150 C. under stirring, and distilled with an oil pump under reduced pressure (vacuum range: 0.01 MPa 0.5 MPa) to remove the possible residual trace water in the mixture and eliminate air bubbles, and heat until the silk fibroin is completely dissolved. Stop heating and stirring, and get a uniform and transparent solution after cooling. The resulting silk fibroin ionic liquid solution is stored at room temperature in a sealed and dry environment for later use.

S2: Rheological Tests:

1) Energy Storage Modulus and Loss Modulus Test

[0059] Parallel splints are chosen, and the test is protected by nitrogen purge of the temperature control hood.

[0060] The linear dynamic elasticity test is adopted, that is, the strain amplitude of the oscillation mode is controlled below 50% to ensure that the storage modulus and loss modulus are linear in the frequency sweep range (110.sup.2 rad/s 3010.sup.2 rad/s).

[0061] Frequency sweeps were performed at the following temperatures (0 C., 10 C., 20 C., and 30 C.) to obtain the storage modulus and loss modulus curves of the samples at different temperatures.

2) Viscosity Test

[0062] Parallel splints are chosen, and the test is protected by nitrogen purge of the temperature control hood.

[0063] Steady-state test experiments were adopted: the shear rate was scanned from low to high, and the shear rate ranged from 10.sup.510.sup.5 s.sup.1. The viscosity value of the platform curve is recorded.

[0064] S3: Modeling with the Rouse Model: Based on the commonly used AI model optimization algorithms, an algorithm is established to evaluate the quality of the prediction results of the Rouse model. [0065] a) Set the error threshold .sub.th=0.01, when the prediction error of the Rouse model is less than .sub.th it is considered that the accuracy of the Rouse model is high enough to truly describe the experimental polymer system. [0066] b) Set the initial learning rate 0-0.05 to control the iteration speed of the algorithm. [0067] C) The gradient descent method was used to continuously modify the model parameters until the prediction error was less than 0.01.

[0068] S4: The specific method of establishing the Rouse model to fit the biopolymer system is as follows: [0069] a) It is assumed that the structure of the biopolymer system satisfies the Rouse model, and the molecular weight distribution of the system satisfies the normal distribution. The experimental system was modeled, and the mean M.sub.0 and standard deviation M.sub.0 of the molecular weight distribution were initialized. [0070] b) M.sub.0 and M.sub.0 were used to construct a normal distribution of molecular weight MN(M,M) [Equation (4)], and the initial simulation physics system was established by sampling according to this distribution

[00004]$\begin{array}{cc}f\left(M_i\right)=\frac{1}{M\sqrt{2}}{e}^{-\frac{\left(M_i-\stackrel{}{M}\right)}{2{}^2}}& \left(4\right)\end{array}$ [0071] c) According to the molecular weight distribution of the simulated physical system, the relaxation time .sub.ip under mode p is calculated according to equation (3). [0072] d) Considering the contribution of all vibration modes p to the storage modulus G and the dissipative modulus G, the G and G of the simulated physical system are calculated according to equations (1) and (2). The density p of the polymer in the solution, the zero-cut viscosity 0 of the solution and the viscosity s of the solvent used in the calculation are consistent with the experimental data. [0073] e) According to the simulation data obtained from the Rouse theoretical model, the relationship between log Glog and log Glog of the simulated physical system can be calculated. The error between the predicted values log G and log G of the Rouse model and the experimental value measured in step S2 can be solved by the root mean square formula. [0074] f) The calculated error of the log G and log G, which describes the degree to which the Rouse model agrees with the experimental results. The fitting process requires that the prediction results of the Rouse model tend to be close to the experimental measurements. Therefore, the error must be as small as possible, and the fitting of the Rouse model evolves into an optimization problem with the optimization objectives as:

[00005]$\begin{array}{cc}\underset{\stackrel{\_}{M},M}{\min }\left(N\left(\stackrel{\_}{M},M\right)\right)& \left(5\right)\end{array}$

[0075] The error varies with the distribution of the polymer, and the objective function of equation (5) is optimized by using the AI algorithm described in step S3. Compare the error threshold .sub.th by the AI algorithm of the actual , for example, .sub.th, the normal distribution parameters M and M are optimized and updated by the AI algorithm, and the process b is returned after the update), and the Rouse model is recalculated. For example, <.sub.th means that the Rouse model at this time has been able to accurately describe the experimental results, complete this step and move on to the next step.

[0076] S5: Calculate molecular weight-related data using the Rouse model optimized by AI algorithm: The optimized Rouse model obtained in step S4 is used to simulate the molecular weight distribution N(M.sub.opt, M.sub.opt) of the physical system to achieve a weight-average molecular weight M.sub.w of 266 kDa, a number-average molecular weight M.sub.n of 253 kDa, and a molecular weight distribution of 1.05.

Example 2

[0077] S1: Sample Preparation: [0078] a) Dissolution: Silkworm silk fibroin prepared by sodium carbonate degumming for 30 minutes was selected, and Silkworm silk fibroin was mixed with ionic liquid 1-allylenyl-3-methylimidazole chloride salt, and the concentration of silk fibroin was controlled at 20%. [0079] b) Freeze-drying and dehydration: The sample is frozen in liquid nitrogen to fully solidify the ionic liquid, and placed in a freeze-dryer to be fully dehumidified and then taken out, sealed and thawed at room temperature. [0080] c) Dissolution under reduced pressure: The silk fibroin ionic liquid mixture prepared by step b is heated in an oil bath at a temperature of 150 C. under stirring and distilled under reduced pressure with an oil pump (vacuum range: 0.01 MPa 0.5 MPa) to remove the possible residual trace water in the mixture and eliminate air bubbles, and heat until the silk fibroin is completely dissolved. Stop heating and stirring, and get a uniform and transparent solution after cooling. The resulting silk fibroin ionic liquid solution is stored at room temperature in a sealed and dry environment for later use.

[0081] S2: Rheological method test: Same as step S2 in example 1.

[0082] S3: modeling with a Rouse model: Same as step S3 in example 1.

[0083] S4: Establish a Rouse model to fit the biopolymer system: Same as step S4 in example 1.

[0084] S5: Calculate molecular weight-related data using the Rouse model optimized by AI algorithm:

[0085] The optimized Ruse model obtained in step S4 is used to simulate the molecular weight distribution N(M.sub.opt, M.sub.opt) of the physical system to achieve a weight-average molecular weight M.sub.w of 181 kDa, a number-average molecular weight M.sub.n of 97 kDa, and a molecular weight distribution of 1.87.

Example 3

[0086] S1: Sample Preparation: [0087] a) Dissolution: The soluble large molecular weight spray-dried silkworm silk fibroin powder prepared by degumming with sodium carbonate for 45 minutes is used. The silkworm silk fibroin powder is mixed with ionic liquid 1-allyl-3-methylimidazolium chloride, and the concentration of silk fibroin is controlled to 20%. [0088] b) Freeze-drying and dehydration: The sample is frozen in liquid nitrogen to fully solidify the ionic liquid, and placed in a freeze-dryer to be fully dehumidified and then taken out, sealed and thawed at room temperature. [0089] c) Dissolution under reduced pressure: The silk fibroin ionic liquid mixture prepared by step b is heated in an oil bath at a temperature of 120 C. under stirring, and distilled under reduced pressure with an oil pump (vacuum range: 0.01 MPa0.5 MPa) to remove the possible residual trace water in the mixture and eliminate air bubbles, and heat until the silk fibroin is completely dissolved. Stop heating and stirring, and get a uniform and transparent solution after cooling. The resulting silk fibroin ionic liquid solution is stored at room temperature in a sealed and dry environment for later use.

[0090] S2: Rheological method test: Same as step S2 in example 1.

[0091] S3: modeling with a rouse model: Same as step S3 in example 1.

[0092] S4: Establish a Rouse model to fit the biopolymer system: Same as step S4 in example 1.

[0093] S5: Calculate molecular weight-related data using the Rouse model optimized by AI algorithm: The optimized Rouse model obtained in step S4 is used to simulate the molecular weight distribution N(M.sub.opt, M.sub.opt) of the physical system to achieve a weight-average molecular weight M.sub.w of 145 kDa, a number-average molecular weight M.sub.n of 71 kDa, and a molecular weight distribution of 2.05.

Example 4

[0094] S1: Sample Preparation: [0095] a) Dissolution: The soluble large molecular weight freeze-dried silkworm silk fibroin powder prepared by sodium carbonate degumming for 60 minutes was selected, and the silkworm silk fibroin powder was mixed with ionic liquid 1-allyl-3-methylimidazole chloride salt, and the concentration of silk fibroin was controlled at 20%. [0096] b) Freeze-drying and dehydration: The sample is frozen in liquid nitrogen to fully solidify the ionic liquid, and placed in a freeze-dryer to be fully dehumidified and then taken out, sealed and thawed at room temperature. [0097] c) Dissolution under reduced pressure: The silk fibroin ionic liquid mixture prepared by step b is heated in an oil bath at a temperature of 120 C. under stirring, and distilled under reduced pressure with an oil pump (vacuum range: 0.01 MPa0.5 MPa) to remove the possible residual trace water in the mixture and eliminate air bubbles, and heat until the silk fibroin is completely dissolved. Stop heating and stirring, and get a uniform and transparent solution after cooling. The resulting silk fibroin ionic liquid solution is stored at room temperature in a sealed and dry environment for later use.

[0098] S2: Rheological method test: Same as step S2 in example 1.

[0099] S3: modeling with a rouse model: Same as step S3 in example 1.

[0100] S4: Establish a Rouse model to fit the biopolymer system: Same as step S4 in example 1.

[0101] S5: Molecular weight-related data are calculated using the Rouse model optimized by AI algorithm: The optimized Rouse model obtained in step S4 is used to simulate the molecular weight distribution N(M.sub.opt, M.sub.opt) of the physical system to achieve a weight-average molecular weight M.sub.w of 115 kDa, a number-average molecular weight M.sub.n of 58 kDa, and a molecular weight distribution of 1.97.