WO2023024607A1 - 基于ai算法的生物大分子材料分子量的计算方法 - Google Patents

基于ai算法的生物大分子材料分子量的计算方法 Download PDF

Info

Publication number
WO2023024607A1
WO2023024607A1 PCT/CN2022/094288 CN2022094288W WO2023024607A1 WO 2023024607 A1 WO2023024607 A1 WO 2023024607A1 CN 2022094288 W CN2022094288 W CN 2022094288W WO 2023024607 A1 WO2023024607 A1 WO 2023024607A1
Authority
WO
WIPO (PCT)
Prior art keywords
molecular weight
algorithm
model
rouse
ionic liquid
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2022/094288
Other languages
English (en)
French (fr)
Inventor
刘也卓
杨文华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Favorsun Medical Tech Suzhou Co Ltd
Original Assignee
Favorsun Medical Tech Suzhou Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Favorsun Medical Tech Suzhou Co Ltd filed Critical Favorsun Medical Tech Suzhou Co Ltd
Priority to EP22859955.1A priority Critical patent/EP4528573A4/en
Priority to CN202280097001.5A priority patent/CN119365865B/zh
Priority to US18/875,765 priority patent/US20250226062A1/en
Priority to PCT/CN2022/094288 priority patent/WO2023024607A1/zh
Priority to KR1020247041818A priority patent/KR20250023989A/ko
Publication of WO2023024607A1 publication Critical patent/WO2023024607A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N1/00—Sampling; Preparing specimens for investigation
    • G01N1/28—Preparing specimens for investigation including physical details of (bio-)chemical methods covered elsewhere, e.g. G01N33/50, C12Q
    • G01N1/40—Concentrating samples
    • G01N1/4055—Concentrating samples by solubility techniques
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N11/00—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N11/00—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties
    • G01N11/10—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties by moving a body within the material
    • G01N11/14—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties by moving a body within the material by using rotary bodies, e.g. vane
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N11/00—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties
    • G01N11/10—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties by moving a body within the material
    • G01N11/14—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties by moving a body within the material by using rotary bodies, e.g. vane
    • G01N11/142—Sample held between two members substantially perpendicular to axis of rotation, e.g. parallel plate viscometer
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N11/00—Investigating flow properties of materials, e.g. viscosity, plasticity; Analysing materials by determining flow properties
    • G01N2011/0026—Investigating specific flow properties of non-Newtonian fluids
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
    • G01N2203/0058—Kind of property studied
    • G01N2203/0089—Biorheological properties
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2203/00—Investigating strength properties of solid materials by application of mechanical stress
    • G01N2203/02—Details not specific for a particular testing method
    • G01N2203/0202—Control of the test
    • G01N2203/0212—Theories, calculations
    • G01N2203/0218—Calculations based on experimental data
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/30—Prediction of properties of chemical compounds, compositions or mixtures

Definitions

  • the invention relates to a method for measuring the molecular weight of biological macromolecular materials, in particular to calculating the molecular weight of materials difficult to measure based on rheological models, and using AI nonlinear fitting to solve the molecular weight of biological materials.
  • biomacromolecular materials have also received more extensive attention, such as silk protein, hyaluronic acid, collagen, recombinant collagen, sericin, etc., which have been used as important medical and cosmetic raw materials. .
  • the biological properties of many biomacromolecular materials are closely related to their relative molecular weights. Biomacromolecular materials with different relative molecular weights have different physiological activities, and some even have completely opposite effects.
  • hyaluronic acid with a molecular weight greater than 2000kDa has good moisturizing properties, viscoelasticity, lubrication, and anti-inflammatory functions, and can be used in ophthalmic surgery and bone joint injection preparations; It can be absorbed by the road to exert the effect of beautifying and nourishing the skin.
  • Another example is that the swelling property and light transmittance of silk protein hydrogel will decrease with the increase of centrin molecular weight. Therefore, if the molecular weight of the polymer material can be accurately obtained, the properties of the material can be fully utilized.
  • the calculation of molecular weight based on the above models is relatively complicated, and a large amount of data needs to be tested and a large amount of calculations are introduced. Then if it is calculated manually, the efficiency is relatively low, and the accuracy is not high. If the AI algorithm is introduced into the traditional calculation method of polymer molecular weight, it can solve the timeliness and accuracy of calculation, greatly improve the work efficiency of R&D personnel, and further develop innovations in the field of biopolymer calculation.
  • the problem to be solved in the present invention is to invent a method for calculating the molecular weight of biomacromolecular materials based on AI algorithm in view of the existing difficulties in molecular weight measurement of biomacromolecular materials.
  • the invention proposes a method for measuring the molecular weight of biomacromolecular materials based on rheology.
  • the present invention selects ionic liquid as the dissolution system, selects the Rouse model as the polymer solution model, and uses AI algorithm to train the model by measuring a large amount of data to obtain key parameter values, thereby calculating the molecular weight value of the measured material.
  • the present invention screens out good solvents such as ionic liquids for dissolving biomacromolecules by testing different solvent systems.
  • An ionic liquid refers to an ionic compound in a liquid state, or an ionic compound whose melting point is lower than a certain temperature.
  • ionic liquid has attracted great attention from all over the world. It has many excellent characteristics such as good thermal stability, structural designability, low vapor pressure, non-toxicity and recyclability, which is beneficial to the environment. Protection and health of operators.
  • Ionic liquids are not only widely used as solvents and reaction media in organic synthesis reactions, but also as green solvents for some natural macromolecules. Ionic liquids also exhibit excellent properties in dissolving biopolymer materials, for example, silk protein molecules with unique structures can also exist in a single-chain form.
  • the present invention screens out the Rouse model from many polymer solution models through simulation calculation, and this model is more suitable for the calculation of the molecular weight of biological polymers.
  • the relationship between the polymer viscoelasticity data and the molecular weight function in the Rouse model is as follows:
  • rheological methods were used to test the rheological data of biomacromolecules dissolved in ionic liquids.
  • Rheological methods can well reflect the chain state and viscoelastic behavior of polymer chains in solution, and rheological data such as viscoelasticity of polymer chains in ionic liquids can be measured by rheology.
  • the invention introduces AI algorithms to solve the calculation of a large amount of rheological data, which improves the timeliness and accuracy, improves the work efficiency of R&D personnel, and further develops innovations in the field of biopolymer computing.
  • a method for calculating the molecular weight of biomacromolecular materials based on AI algorithm comprising the steps of:
  • Sample testing place the sample prepared in step S1 on a rheometer for sample testing, and test and calculate the required data;
  • the specific method for preparing the sample is:
  • the ionic liquid is 1-allyl-3-methylimidazolium chloride ([AMIm]Cl) and 1-hexyl-3-methylimidazolium bisulfate ([Hmim] HSO 4 ).
  • the biopolymer material includes: silk protein, hyaluronic acid, collagen, recombinant collagen, and sericin.
  • the silk protein is selected from mulberry silk protein, spider silk protein and tussah silk protein.
  • the mulberry silk protein is the remaining part after silkworm cocoons have taken off the sericin, also known as silk fibroin (SILK FIBROIN).
  • an ionic liquid solution of biopolymer silk protein is prepared, with a concentration of 0.1%-50%, preferably 1%-25%, and more preferably 5-20%.
  • freeze-drying is selected for drying and removing water: freeze the sample in liquid nitrogen to fully solidify the ionic liquid, place it in a freeze dryer to fully dehumidify it, take it out, and seal it at room temperature for thawing.
  • the specific method of dissolving under reduced pressure is: heating the silk protein ionic liquid mixture prepared in step b in an oil bath under agitation, and distilling under reduced pressure with an oil pump to remove trace amounts that may remain in the mixture Remove water and eliminate air bubbles, and heat until the biopolymer dissolves completely.
  • the silk protein ionic liquid mixture prepared in step S1(b) is heated in an oil bath at a temperature of 0-180° C. under stirring, and distilled under reduced pressure with an oil pump, and the vacuum degree ranges from -0.01MPa to -0.5MPa, In order to remove the trace moisture that may remain in the mixture and eliminate air bubbles, heat until the biopolymer is completely dissolved; stop heating and stirring, and obtain a uniform and transparent solution after cooling down.
  • the obtained biopolymer ionic liquid solution is stored at room temperature in a sealed and dry environment for future use.
  • the specific method of testing the sample with a rheometer is:
  • the parallel splint is selected, and the test is protected by nitrogen purge of the temperature control cover during the test.
  • the linear dynamic elastic test is adopted, that is, the strain amplitude of the oscillation mode is controlled below 50%, so as to ensure the storage modulus and loss modulus within the frequency sweep range (1 ⁇ 10 2 rad/s ⁇ 30 ⁇ 10 -2 rad/s) is linear.
  • 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.
  • the parallel splint is selected, and the test is protected by nitrogen purge of the temperature control cover during the test.
  • the steady-state test experiment is adopted: the shear rate is scanned from low to high, and the shear rate range is: 10 -5 ⁇ 10 5 s -1 . Record the viscosity value of the plateau curve.
  • the specific method of establishing the AI algorithm to evaluate the Rouse model is: based on the commonly used AI model optimization algorithm, an algorithm for evaluating the quality of the Rouse model prediction result is established, and this algorithm describes the Rouse model prediction and experimental results. degree of matching.
  • the fitting process requires that the theoretical prediction results of the Rouse model approach the experiment. Therefore, the following algorithm parameters need to be set in advance:
  • the AI model optimization algorithm includes Gradient Descent, Conjugate Descent, Adam, AdamGrad, RMSProp.
  • the selection range of the error threshold ⁇ th is 0.001-0.25.
  • the selection range of the learning rate ⁇ 0 is 0.0001-0.1.
  • the specific method of establishing a Rouse model to fit the biopolymer system is as follows: a) Assume that the structure of the biopolymer system satisfies the Rouse model, and the molecular weight distribution of the system satisfies a normal distribution. Model the experimental system and initialize the average value of the molecular weight distribution with standard deviation ⁇ M 0 .
  • the difference ⁇ varies with the distribution of polymers, and the objective function of formula (5) is optimized by using the AI algorithm described in step S3. Compare the actual ⁇ with the error threshold ⁇ th of the AI algorithm. If ⁇ th , the normal distribution parameters are corrected by the AI algorithm Optimize and update with ⁇ M, return to process b) after updating, and recalculate the Rouse model; if ⁇ th , it means that the Rouse model at this time can accurately describe the experimental results, complete this step and enter the next step.
  • the molecular weight related data are weight average molecular weight, number average molecular weight and molecular weight distribution.
  • step S5 using the optimized Rouse model obtained in the step S4, the molecular weight distribution of the simulated physical system is used The calculation of weight average molecular weight M w , number average molecular weight M n and molecular weight distribution, ie M w /M n is completed.
  • Figure 1 is a flow chart of calculating the molecular weight of a biopolymer system based on Rouse model fitting.
  • 2-5 are the relationship between storage modulus and loss modulus of silk protein ionic liquid solution and shear frequency corresponding to different molecular weight silk proteins in Examples 1-4 of the present invention under oscillation mode.
  • Fig. 6 is the calculation results of different silk protein molecular weights in Examples 1-4 of the present invention.
  • mulberry silk protein fiber prepared by degumming with sodium carbonate for 15 minutes, mix mulberry silk protein with ionic liquid 1-allyl-3-methylimidazolium chloride, and control the concentration of silk protein to 20%.
  • Freeze-drying to remove water freeze the sample in liquid nitrogen to fully solidify the ionic liquid, place it in a freeze dryer to fully dehumidify it, take it out, and seal and thaw at room temperature.
  • step b Dissolving under reduced pressure: under stirring, the silk protein ionic liquid mixture prepared in step b was heated in an oil bath at a temperature of 150° C., and distilled under reduced pressure with an oil pump (vacuum degree range: -0.01MPa ⁇ -0.5MPa), to Remove trace moisture that may remain in the mixture and eliminate air bubbles, and heat until the silk protein is completely dissolved. Stop heating and stirring, and obtain a uniform and transparent solution after cooling down. The obtained silk protein ionic liquid solution is stored at room temperature in a sealed and dry environment for future use.
  • an oil pump vacuum degree range: -0.01MPa ⁇ -0.5MPa
  • the parallel splint is selected, and the test is protected by nitrogen purge of the temperature control cover during the test.
  • the linear dynamic elastic test is adopted, that is, the strain amplitude of the oscillation mode is controlled below 50%, so as to ensure the storage modulus and loss modulus within the frequency sweep range (1 ⁇ 10 2 rad/s ⁇ 30 ⁇ 10 -2 rad/s) is linear.
  • 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.
  • the parallel splint is selected, and the test is protected by nitrogen purge of the temperature control cover during the test.
  • the steady-state test experiment is adopted: the shear rate is scanned from low to high, and the shear rate range is: 10 -5 ⁇ 10 5 s -1 . Record the viscosity value of the plateau curve.
  • the difference ⁇ varies with the distribution of polymers, and the objective function of formula (5) is optimized by using the AI algorithm described in step S3. Compare the actual ⁇ with the error threshold ⁇ th of the AI algorithm. If ⁇ th , the normal distribution parameters are corrected by the AI algorithm Optimize and update with ⁇ M, return to process b) after updating, and recalculate the Rouse model; if ⁇ th , it means that the Rouse model at this time can accurately describe the experimental results, complete this step and enter the next step.
  • step S5 Use the Rouse model optimized by the AI algorithm to calculate molecular weight related data: use the optimized Rouse model obtained in step S4 to use the molecular weight distribution of the simulated physical system A weight average molecular weight M w of 266 kDa, a number average molecular weight M n of 253 kDa and a molecular weight distribution of 1.05 were achieved.
  • mulberry silk protein fiber prepared by degumming with sodium carbonate for 30 minutes, mix mulberry silk protein with ionic liquid 1-allyl-3-methylimidazolium chloride, and control the concentration of silk protein to 20%.
  • Freeze-drying to remove water freeze the sample in liquid nitrogen to fully solidify the ionic liquid, place it in a freeze dryer to fully dehumidify it, take it out, and seal and thaw at room temperature.
  • step b Dissolving under reduced pressure: under stirring, the silk protein ionic liquid mixture prepared in step b was heated in an oil bath at a temperature of 150° C., and distilled under reduced pressure with an oil pump (vacuum degree range: -0.01MPa ⁇ -0.5MPa), In order to remove the trace moisture that may remain in the mixture and eliminate air bubbles, heat until the silk protein is completely dissolved. Stop heating and stirring, and obtain a uniform and transparent solution after cooling down. The obtained silk protein ionic liquid solution is stored at room temperature in a sealed and dry environment for future use.
  • an oil pump vacuum degree range: -0.01MPa ⁇ -0.5MPa
  • step S5 Use the Rouse model optimized by the AI algorithm to calculate molecular weight related data: use the optimized Rouse model obtained in step S4 to use the molecular weight distribution of the simulated physical system A weight average molecular weight M w of 181 kDa, a number average molecular weight M n of 97 kDa and a molecular weight distribution of 1.87 were achieved.
  • Freeze-drying to remove water freeze the sample in liquid nitrogen to fully solidify the ionic liquid, place it in a freeze dryer to fully dehumidify it, take it out, and seal and thaw at room temperature.
  • step b Dissolving under reduced pressure: under stirring, the silk protein ionic liquid mixture prepared in step b was heated in an oil bath at a temperature of 120° C., and distilled under reduced pressure with an oil pump (vacuum degree range: -0.01MPa ⁇ -0.5MPa), In order to remove the trace moisture that may remain in the mixture and eliminate air bubbles, heat until the silk protein is completely dissolved. Stop heating and stirring, and obtain a uniform and transparent solution after cooling down. The obtained silk protein ionic liquid solution is stored at room temperature in a sealed and dry environment for future use.
  • an oil pump vacuum degree range: -0.01MPa ⁇ -0.5MPa
  • step S5 Use the Rouse model optimized by the AI algorithm to calculate molecular weight related data: use the optimized Rouse model obtained in step S4 to use the molecular weight distribution of the simulated physical system A weight average molecular weight M w of 145 kDa, a number average molecular weight M n of 71 kDa and a molecular weight distribution of 2.05 were achieved. .
  • Freeze-drying to remove water freeze the sample in liquid nitrogen to fully solidify the ionic liquid, place it in a freeze dryer to fully dehumidify it, take it out, and seal and thaw at room temperature.
  • step b Dissolving under reduced pressure: under stirring, the silk protein ionic liquid mixture prepared in step b was heated in an oil bath at a temperature of 120° C., and distilled under reduced pressure with an oil pump (vacuum degree range: -0.01MPa ⁇ -0.5MPa), In order to remove the trace moisture that may remain in the mixture and eliminate air bubbles, heat until the silk protein is completely dissolved. Stop heating and stirring, and obtain a uniform and transparent solution after cooling down. The obtained silk protein ionic liquid solution is stored at room temperature in a sealed and dry environment for future use.
  • an oil pump vacuum degree range: -0.01MPa ⁇ -0.5MPa
  • step S5 Use the Rouse model optimized by the AI algorithm to calculate molecular weight related data: use the optimized Rouse model obtained in step S4 to use the molecular weight distribution of the simulated physical system A weight average molecular weight M w of 115 kDa, a number average molecular weight M n of 58 kDa and a molecular weight distribution of 1.97 were achieved. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Chemical & Material Sciences (AREA)
  • Medical Informatics (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • Pathology (AREA)
  • Immunology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Databases & Information Systems (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Biophysics (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Computer Hardware Design (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Biomedical Technology (AREA)
  • Geometry (AREA)
  • Epidemiology (AREA)
  • Bioethics (AREA)
  • Public Health (AREA)
  • Biotechnology (AREA)
  • Evolutionary Biology (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Investigating Or Analysing Biological Materials (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Peptides Or Proteins (AREA)

Abstract

本发明涉及一种基于AI算法的生物大分子材料分子量的计算方法。部分生物高分子材料在溶解成水溶液后难以形成单分子链，本发明选用了离子液体为溶解体系，大分子材料溶解后能很好地形成单分子链，然后使用流变学方法测试生物大分子溶解于离子液体后的流变学数据。本发明选用了Rouse模型为高分子溶液模型，通过测量大量数据，用AI算法对模型进行训练，并进行非线性拟合，得到了关键参数值，从而计算得到所测材料的分子量数值。该计算方法具有精确，快速的特点，可以很好地推广到高分子材料的计算机研究领域。

Description

基于AI算法的生物大分子材料分子量的计算方法 技术领域
本发明涉及一种测量生物大分子材料分子量的方法，尤其涉及基于流变学模型计算难以测定分子量的材料，并用AI非线性拟合求解生物材料的分子量。
背景技术
随着生物分子学的不断发展，生物大分子材料也得到更加广泛的关注，如丝蛋白、透明质酸、胶原蛋白、重组胶原、丝胶蛋白等，已经被作为重要的医疗、美容原料进行应用。而许多生物大分子材料的生物学性质与其相对分子量大小有着密切的关系，不同相对分子量的生物大分子材料，其生理活性也不同，有些甚至具有完全相反的作用。如分子量大于2000kDa的透明质酸具有较好的保湿性、粘弹性、润滑、抑制炎性等功能，可用于眼科手术、骨关节注射制剂；分子量10kDa-80kDa的透明质酸可以直接口服，被肠道吸收，发挥美容养颜的功效。再如丝蛋白水凝胶的溶胀性、透光率会着丝蛋白分子量的增大而减小。因此，如果能准确得到高分子材料的分子量大小，就能更加充分地利用材料的性能。
传统的测定蛋白分子量的方法包括：凝胶电泳(SDS-PAGE)、光散射(SLS)、小角中子衍射(SANS)、凝胶渗透色谱(GPC)等测试方法。但传统测量方法在使用时往往存在一些局限性，即对于一些在溶液中存在聚集行为的高分子材料无法准确测定。如丝蛋白分子的独特结构(亲疏水片段相间的类多嵌段共聚结构)，将造成再生丝蛋白在溶液中极易聚集成胶束等聚集体，而不是以“自由”的单分子链形式存在。所以需要先找到一种良溶剂，在良溶剂中丝蛋白可以实现分子链以“自由”单链形式存在。再找到一个合适的测试方法，可以定量的描述高分子链在溶液中链状态及其黏弹行为。
此外还需要找到一个合适的高分子溶液模型。研究者们为了研究高分子溶液的结构和动力学特征，建立了几种粗粒化几何模型，如珠-棍模型(bead-stick model)、珠-簧模型(bead-spring model)、珍珠项链模型(pearl-necklace model)和蛇形模型(reptation model)等。到底哪个模型更合适用于丝蛋白的分子量计算需要做进一步的探索。
基于上述模型的分子量计算都比较复杂，需要测试大量的数据，引入大量的计算。那么如果是人工计算，效率比较低，而且精准度不高。如果将AI算法引入到传统高分子分子量的计算方法中，可以解决计算的时效性和精准度，将大大提高研发人员的工作效率，也可以进一步开拓生物高分子计算领域的创新。
发明内容
本发明要解决的问题是针对现有生物高分子材料分子量测量中存在的难题，发明了一种基于AI算法的生物大分子材料分子量的计算方法。
本发明提出了一种基于流变学的生物大分子材料分子量测量方法。本发明选用了离子液体为溶解体系，选用Rouse模型为高分子溶液模型，通过测量大量数据，用AI算法对模型进行训练，得到关键参数值，从而计算得到所测材料的分子量数值。
首先本发明通过测试不同的溶剂体系，筛选出离子液体这种良溶剂，用于溶解生物大分子。
离子液体是指液态时的离子化合物，也可指熔点低于一定温度的离子化合物。近年来，离子液体作为一种新型绿色溶剂已经引起世界各国的高度重视，它具备良好热稳定性和结构可设计性、蒸汽压低、无毒且可回收循环利用等诸多优异的特性，有利于环境保护和操作人员的健康。离子液体不仅被广泛用做有机合成反应的溶剂和反应媒介，而且被视为一些天然大分子的绿色溶剂。在溶解生物高分子材料方面，离子液体也表现出优异的性质，如存在独特结构的丝蛋白分子也能在其中以单链形式存在。
其次，本发明通过模拟计算从众多高分子溶液模型中筛选出Rouse模型，这一个模型更适用于生物高分子分子量的计算。Rouse模型中高分子粘弹性数据与分子量函数之间的关系如下：
Figure PCTCN2022094288-appb-000001
Figure PCTCN2022094288-appb-000002
Figure PCTCN2022094288-appb-000003
然后使用流变学方法测试生物大分子溶解于离子液体后的流变学数据。流变学方法能够很好地反映高分子链在溶液中的链状态及其黏弹行为,采用流变学可以测定高分子在离子液体中的粘弹性等流变学数据。
最后，本发明引入AI算法，以解决计算大量的流变学数据，提高了时效性和精准度，提高了研发人员的工作效率，也进一步开拓生物高分子计算领域的创新。
本发明通过如下技术方案进行:
一种基于AI算法的生物大分子材料分子量的计算方法，包括如下步骤:
S1.制备样品：取待测生物高分子材料样品，将其溶解于离子液体中，得到待测样品溶液；
S2.样品测试：将步骤S1制备的样品置于流变仪上进行样品测试，测试计算所需数据；
S3.建立AI算法评估Rouse模型；
S4.建立Rouse模型拟合生物高分子体系；
S5.使用AI算法优化的Rouse模型计算分子量相关数据；
优选的，所述步骤S1中，制备样品的具体方法为：
a)溶解：将干燥的生物高分子材料加入离子液体中混合，常温下溶解；
b)干燥除水；
c)减压溶解。
优选的，所述步骤S1中，所述离子液体为1-烯丙基-3-甲基咪唑氯盐([AMIm]Cl)和1-己基-3-甲基咪唑硫酸氢盐([Hmim]HSO 4)。
优选的，所述步骤S1中，生物高分子材料包括：丝蛋白、透明质酸、胶原蛋白、重组胶原、丝胶蛋白。优选地，所述丝蛋白选自桑蚕丝蛋白，蜘蛛丝蛋白，柞蚕丝蛋白。所述桑蚕丝蛋白为蚕茧脱去丝胶后剩余的部分，又称丝素蛋白(SILK FIBROIN)。优选的，所述步骤S1中，制备生物高分子材料丝蛋白的离子液体溶液，浓度为0.1％-50％，优选1％-25％，再优选为5-20％。
优选的，所述步骤S1中，干燥除水选择冷冻干燥：将样品放置于液氮中冷冻，使离子液体充分固化，并置于冷冻干燥机中充分抽湿后取出，室温条件密封解冻。
优选的，所述步骤S1中，减压溶解的具体方法为：将步骤b制备的丝蛋白离子液体混合物在搅拌状态下，油浴加热，并用油泵减压蒸馏，以除去混合物中可能残余的微量水分并消除气泡，加热至生物高分子溶解完全即可。
优选地，将步骤S1(b)制备的丝蛋白离子液体混合物在搅拌状态下，油浴加热，温度为0-180℃，并用油泵减压蒸馏，真空度范围为-0.01MPa—-0.5MPa，以除去混合物中可 能残余的微量水分并消除气泡，加热至生物高分子溶解完全；停止加热和搅拌，降温后得到均匀透明的溶液。
更优选，所得生物高分子离子液体溶液在密封干燥的环境下室温保存备用。
优选的，所述步骤S2中，用流变仪测试样品的具体方法为：
a)储能模量及损耗模量测试
选择平行夹板，测试过程中通过控温罩的氮气吹扫对测试进行保护。
采用线性动态弹性测试，即震荡模式的应变振幅控制在50％以下，以确保在频率扫描范围(1×10 2rad/s～30×10 -2rad/s)内存储模量和损耗模量为线性。
在以下几个温度下(0℃、10℃、20℃和30℃)进行频率扫描，得到不同温度下样品的储能模量和损耗模量曲线。
b)粘度测试
选择平行夹板，测试过程中通过控温罩的氮气吹扫对测试进行保护。
采取稳态测试实验：剪切速率由低到高进行扫描，剪切速率范围为：10 -5～10 5s -1。记录平台曲线的粘度数值。
优选的，所述步骤S3中，所述建立AI算法评估Rouse模型的具体方式为：基于常用的AI模型优化算法，建立评估Rouse模型预测结果质量的算法，此算法描述了Rouse模型预测与实验结果的吻合程度。拟合过程要求Rouse模型的理论预测结果趋近于实验。因此，需提前设置以下算法参数：
a)设置一个合理的误差阈值ε th，当Rouse模型的预测误差ε小于ε th，则认为Rouse模型精度已足够高，可真实描述实验高分子体系的情况；
b)设置一个合理的初始学习速率α 0，控制算法的迭代速度；
C)采用梯度下降法对模型参数进行不断修正，直到预测误差小于0.01。
优选的，所述步骤S3中，所述AI模型优化算法包括梯度下降法(Gradient Descent)，共轭梯度法(Conjugate Descent)，Adam，AdamGrad，RMSProp。
优选的，所述步骤S3中，所述误差阈值ε th的选取范围在0.001-0.25。
优选的，所述步骤S3中，所述学习速率α 0的选取范围在0.0001-0.1。
优选的，所述步骤S4中，所述建立Rouse模型拟合生物高分子体系的具体方法为：a) 假定生物高分子体系的结构满足Rouse模型，且体系的分子量分布满足正态分布。对实验体系进行建模，初始化分子量分布的平均值
Figure PCTCN2022094288-appb-000004
与标准差ΔM 0。
b)使用
Figure PCTCN2022094288-appb-000005
和ΔM 0构建分子量的正态分布
Figure PCTCN2022094288-appb-000006
[公式(4)]，依据此分布进行采样，建立初始的模拟物理体系。
Figure PCTCN2022094288-appb-000007
c)根据模拟物理体系的分子量分布，按照公式(3)计算模式p下的弛豫时间τ ip。
d)考虑所有振动模式p对储能模量G′和耗散模量G″的贡献，按照公式(1)和(2)计算模拟物理体系的G′和G″。计算中所用到的高分子在溶液中的密度ρ、溶液的零切黏度η 0和溶剂的粘度η s与实验数据一致。
e)按照Rouse理论模型得到的模拟数据，可以计算出模拟物理体系的log G′～logω和log G″～logω关系。Rouse模型的预测值log G′、log G″与步骤S2测量实验值的差值ε，可通过均方根公式求解得到。
f)计算的log G′与log G″数据的差值ε，描述了Rouse模型与实验结果的符合程度。拟合过程要求Rouse模型的预测结果趋近于实验测量结果。因此，差值ε须尽可能小，则Rouse模型的拟合演化为一个优化问题，优化目标为
Figure PCTCN2022094288-appb-000008
差值ε随着高分子的分布情况而变化，利用步骤S3所述的AI算法，对公式(5)的目标函数进行优化。比较实际ε与AI算法的误差阈值ε th，如ε≥ε th，则通过AI算法对正态分布参数
Figure PCTCN2022094288-appb-000009
和ΔM进行优化更新，更新后返回过程b)，重新计算Rouse模型；如ε＜ε th，则说明此时的Rouse模型已经能够准确描述实验结果，完成此步骤并进入下一步骤。
优选的，所述步骤S5中，使用AI算法优化的Rouse模型计算分子量中，所述分子量相关数据为重均分子量、数均分子量和分子量分布。
进一步的，所述步骤S5中，利用步骤S4得到的经优化的Rouse模型，使用模拟物理体系的分子量分布
Figure PCTCN2022094288-appb-000010
完成重均分子量M w、数均分子量M n和分子量分布，即M w/M n的计算。
附图说明
图1是基于Rouse模型拟合计算生物高分子体系的分子量流程。
图2-5是分别对应按本发明实施例1-4中不同分子量丝蛋白在震荡模式下丝蛋白离子液体溶液的储能模量以及损耗模量与剪切频率的关系。
图6是本发明实施例1-4中对不同丝蛋白分子量的计算结果。
具体实施方式
本发明通过如下具体实施方式。但本领域技术人员知悉，下述具体实施方式不对本发明保护范围的限制。
实施例1：
S1：样品制备
a)溶解：选用碳酸钠脱胶15分钟制得的桑蚕丝蛋白纤维，将桑蚕丝蛋白与离子液体1-烯丙基-3-甲基咪唑氯盐混合，丝蛋白的浓度控制为20％。
b)冷冻干燥除水：将样品放置于液氮中冷冻，使离子液体充分固化，并置于冷冻干燥机中充分抽湿后取出，室温条件密封解冻。
c)减压溶解：将经过步骤b制备的丝蛋白离子液体混合物在搅拌状态下，油浴加热，温度150℃，并用油泵减压蒸馏(真空度范围：-0.01MPa～-0.5MPa)，以除去混合物中可能残余的微量水分并消除气泡，加热至丝蛋白溶解完全即可。停止加热和搅拌，降温后得到均匀透明的溶液。所得丝蛋白离子液体溶液在密封干燥的环境下室温保存备用。
S2：流变方法测试：
1)储能模量及损耗模量测试
选择平行夹板，测试过程中通过控温罩的氮气吹扫对测试进行保护。
采用线性动态弹性测试，即震荡模式的应变振幅控制在50％以下，以确保在频率扫描范围(1×10 2rad/s～30×10 -2rad/s)内存储模量和损耗模量为线性。
在以下几个温度下(0℃、10℃、20℃和30℃)进行频率扫描，得到不同温度下样品的储能模量和损耗模量曲线。
2)粘度测试
选择平行夹板，测试过程中通过控温罩的氮气吹扫对测试进行保护。
采取稳态测试实验：剪切速率由低到高进行扫描，剪切速率范围为：10 -5～10 5s -1。记录平台曲线的粘度数值。
S3：用rouse模型进行建模：基于常用的AI模型优化算法，建立评估Rouse模型预测结果质量的算法：
a)设置误差阈值ε th＝0.01，当Rouse模型的预测误差ε小于ε th，则认为Rouse模型精度已足够高，可真实描述实验高分子体系的情况。
b)设置初始学习速率α 0＝0.05，控制算法的迭代速度。
C)采用梯度下降法对模型参数进行不断修正，直到预测误差小于0.01。
S4：建立Rouse模型拟合生物高分子体系的具体方法为：a)假定生物高分子体系的结构满足Rouse模型，且体系的分子量分布满足正态分布。对实验体系进行建模，初始化分子量分布的平均值
Figure PCTCN2022094288-appb-000011
与标准差ΔM 0。
b)使用
Figure PCTCN2022094288-appb-000012
和ΔM 0构建分子量的正态分布
Figure PCTCN2022094288-appb-000013
[公式(4)]，依据此分布进行采样，建立初始的模拟物理体系。
Figure PCTCN2022094288-appb-000014
c)根据模拟物理体系的分子量分布，按照公式(3)计算模式p下的弛豫时间τ ip。
d)考虑所有振动模式p对储能模量G′和耗散模量G″的贡献，按照公式(1)和(2)计算模拟物理体系的G′和G″。计算中所用到的高分子在溶液中的密度ρ、溶液的零切黏度η 0和溶剂的粘度η s与实验数据一致。
e)按照Rouse理论模型得到的模拟数据，可以计算出模拟物理体系的log G′～logω和log G″～logω关系。Rouse模型的预测值log G′、log G″与步骤S2测量实验值的差值ε，可通过均方根公式求解得到。
f)计算的log G′与log G″数据的差值ε，描述了Rouse模型与实验结果的符合程度。拟合过程要求Rouse模型的预测结果趋近于实验测量结果。因此，差值ε须尽可能小，则Rouse模型的拟合演化为一个优化问题，优化目标为
Figure PCTCN2022094288-appb-000015
差值ε随着高分子的分布情况而变化，利用步骤S3所述的AI算法，对公式(5)的目标函数进行优化。比较实际ε与AI算法的误差阈值ε th，如ε≥ε th，则通过AI算法对正态分布参数
Figure PCTCN2022094288-appb-000016
和ΔM进行优化更新，更新后返回过程b)，重新计算Rouse模型；如ε＜ε th，则说明此时的Rouse模型已经能够准确描述实验结果，完成此步骤并进入下一步骤。
S5：使用AI算法优化的Rouse模型计算分子量相关数据：将步骤S4中得到的优化Rouse模型，使用模拟物理体系的分子量分布
Figure PCTCN2022094288-appb-000017
完成重均分子量M w为266kDa、数均分子量M n为253kDa和分子量分布为1.05。
实施例2
S1：样品制备：
a)溶解：选用碳酸钠脱胶30分钟制得的桑蚕丝蛋白纤维，将桑蚕丝蛋白与离子液体1-烯丙基-3-甲基咪唑氯盐混合，丝蛋白的浓度控制为20％。
b)冷冻干燥除水：将样品放置于液氮中冷冻，使离子液体充分固化，并置于冷冻干燥机中充分抽湿后取出，室温条件密封解冻。
c)减压溶解：将经过步骤b制备的丝蛋白离子液体混合物在搅拌状态下，油浴加热，温度为150℃，并用油泵减压蒸馏(真空度范围：-0.01MPa～-0.5MPa)，以除去混合物中可 能残余的微量水分并消除气泡，加热至丝蛋白溶解完全即可。停止加热和搅拌，降温后得到均匀透明的溶液。所得丝蛋白离子液体溶液在密封干燥的环境下室温保存备用。
S2：流变方法测试：同实施例1中的步骤S2。
S3：用rouse模型进行建模：同实施例1中的步骤S3。
S4：建立Rouse模型拟合生物高分子体系：同实施例1中的步骤S4。
S5：使用AI算法优化的Rouse模型计算分子量相关数据：将步骤S4中得到的优化Rouse模型，使用模拟物理体系的分子量分布
Figure PCTCN2022094288-appb-000018
完成重均分子量M w为181kDa、数均分子量M n为97kDa和分子量分布为1.87。
实施例3
S1：样品制备：
a)溶解：选用碳酸钠脱胶45分钟制得的可溶性大分子量喷干桑蚕丝蛋白粉末，将桑蚕丝蛋白粉末与离子液体1-烯丙基-3-甲基咪唑氯盐混合，丝蛋白的浓度控制为20％。
b)冷冻干燥除水：将样品放置于液氮中冷冻，使离子液体充分固化，并置于冷冻干燥机中充分抽湿后取出，室温条件密封解冻。
c)减压溶解：将经过步骤b制备的丝蛋白离子液体混合物在搅拌状态下，油浴加热，温度为120℃，并用油泵减压蒸馏(真空度范围：-0.01MPa～-0.5MPa)，以除去混合物中可能残余的微量水分并消除气泡，加热至丝蛋白溶解完全即可。停止加热和搅拌，降温后得到均匀透明的溶液。所得丝蛋白离子液体溶液在密封干燥的环境下室温保存备用。
S2：流变方法测试：同实施例1中的步骤S2。
S3：用rouse模型进行建模：同实施例1中的步骤S3。
S4：建立Rouse模型拟合生物高分子体系：同实施例1中的步骤S4。
S5：使用AI算法优化的Rouse模型计算分子量相关数据：将步骤S4中得到的优化Rouse模型，使用模拟物理体系的分子量分布
Figure PCTCN2022094288-appb-000019
完成重均分子量M w为145kDa、数均分子量M n为71kDa和分子量分布为2.05。。
实施例4
S1：样品制备：
a)溶解：选用碳酸钠脱胶60分钟制得的可溶性大分子量冻干桑蚕丝蛋白粉末，将桑蚕丝蛋白粉末与离子液体1-烯丙基-3-甲基咪唑氯盐混合，丝蛋白的浓度控制为20％。
b)冷冻干燥除水：将样品放置于液氮中冷冻，使离子液体充分固化，并置于冷冻干燥机中充分抽湿后取出，室温条件密封解冻。
c)减压溶解：将经过步骤b制备的丝蛋白离子液体混合物在搅拌状态下，油浴加热，温度为120℃，并用油泵减压蒸馏(真空度范围：-0.01MPa～-0.5MPa)，以除去混合物中可能残余的微量水分并消除气泡，加热至丝蛋白溶解完全即可。停止加热和搅拌，降温后得到均匀透明的溶液。所得丝蛋白离子液体溶液在密封干燥的环境下室温保存备用。
S2：流变方法测试：同实施例1中的步骤S2。
S3：用rouse模型进行建模：同实施例1中的步骤S3。
S4：建立Rouse模型拟合生物高分子体系：同实施例1中的步骤S4。
S5：使用AI算法优化的Rouse模型计算分子量相关数据：将步骤S4中得到的优化Rouse模型，使用模拟物理体系的分子量分布
Figure PCTCN2022094288-appb-000020
完成重均分子量M w为115kDa、数均分子量M n为58kDa和分子量分布为1.97。。

Claims (10)

  1. 一种基于AI算法的生物大分子材料分子量的计算方法，其特征在于，包括以下步骤：
    S1：制备样品：取待测的生物高分子材料样品，将其溶解于离子液体中，得到待测样品溶液，即为生物高分子材料的离子液体溶液；
    S2：样品测试：将步骤S1制备的样品置于流变仪上进行样品测试，测试计算所需数据；
    S3：建立AI算法评估Rouse模型:基于常用的AI模型优化算法，建立评估Rouse模型预测结果质量的算法；
    S4：建立Rouse模型拟合生物高分子体系；
    S5：使用AI算法优化的Rouse模型计算分子量。
  2. 如权利要求1所述的基于AI算法的生物大分子材料分子量的计算方法，其特征在于，所述步骤S1中制备样品的具体方法为：
    a)溶解：将干燥的生物高分子材料样品加入离子液体中混合，常温下溶解；b)干燥除水；c)减压溶解。
    优选的，所述步骤S1中，生物高分子材料包括：丝蛋白、透明质酸、胶原蛋白、重组胶原、丝胶蛋白。
    优选地，所述丝蛋白选自桑蚕丝蛋白，蜘蛛丝蛋白，柞蚕丝蛋白。
    优选地，所述步骤S1中的离子液体选自[AMIm]Cl和[HMIm]HSO 4。
  3. 如权利要求1或2所述的基于AI算法的生物大分子材料分子量的计算方法，其特征在于，所述步骤S1制备的生物高分子材料丝蛋白的离子液体溶液中，丝蛋白的浓度为0.1％-50％，优选为1％-15％，还优选为5-20％。
  4. 如权利要求2所述的基于AI算法的生物大分子材料分子量的计算方法，其特征在于，所述步骤S1中选择冷冻干燥法干燥除水，具体地，将样品放置于液氮中冷冻，使离子液体充分固化，并置于冷冻干燥机中充分抽湿后取出，室温条件密封解冻。
  5. 如权利要求2或4所述的基于AI算法的生物大分子材料分子量的计算方法，其特征在于，所述步骤S1中制备样品中减压溶解的具体方法为：将步骤S1(b)制备的丝蛋白离子 液体混合物在搅拌状态下，油浴加热，并用油泵减压蒸馏，以除去混合物中可能残余的微量水分并消除气泡，加热至生物高分子溶解完全即可。
    优选地，将步骤S1(b)制备的丝蛋白离子液体混合物在搅拌状态下，油浴加热，温度为0-180℃，并用油泵减压蒸馏，真空度范围为-0.01MPa—-0.5MPa，以除去混合物中可能残余的微量水分并消除气泡，加热至生物高分子溶解完全；停止加热和搅拌，降温后得到均匀透明的溶液。
    更优选，将所得生物高分子离子液体溶液在密封干燥的环境下室温保存备用。
  6. 如权利要求1-5任一项所述的基于AI算法的生物大分子材料分子量的计算方法，其特征在于，所述步骤S2中样品测试的具体方法为：
    a)储能模量及损耗模量测试：选择平行夹板，测试过程中通过控温罩的氮气吹扫对测试进行保护；
    采用线性动态弹性测试，即震荡模式的应变振幅控制在50％以下，以确保在频率扫描范围(1×10 2rad/s-30×10 -2rad/s)内存储模量和损耗模量为线性。
    在以下几个不同温度(0℃、10℃、20℃和30℃)下进行频率扫描，得到不同温度下样品的储能模量和损耗模量曲线；
    b)粘度测试：选择平行夹板，测试过程中通过控温罩的氮气吹扫对测试进行保护；
    采取稳态测试实验：剪切速率由低到高进行扫描，剪切速率范围为：10 -5-10 5s -1，记录平台曲线的粘度数值。
  7. 如权利要求1-6任一项所述的基于AI算法的生物大分子材料分子量的计算方法，其特征在于，所述步骤S3建立AI算法评估Rouse模型具体方法为：
    a)设置一个合理的误差阈值ε th，当Rouse模型的预测误差ε小于ε th，则认为Rouse模型精度已足够高，可真实描述实验高分子体系的情况；
    b)设置一个合理的初始学习速率α 0，控制算法的迭代速度；
    c)采用梯度下降法对模型参数进行不断修正，直到预测误差小于0.01。
    优选地，所述步骤S3中AI模型优化算法包括梯度下降法(Gradient Descent)，共轭梯度法(Conjugate Descent)，Adam，AdamGrad，RMSProp。
  8. 如权利要求1-7任一项所述的基于AI算法的生物大分子材料分子量的计算方法，其特征在于，所述步骤S3中误差阈值ε th的选取范围在0.001-0.25，学习速率α 0的选取范围在0.0001-0.1。
  9. 如权利要求1-8任一项所述的基于AI算法的生物大分子材料分子量的计算方法，其特征在于，所述步骤S4中建立Rouse模型拟合生物高分子体系步骤为：
    a)假定生物高分子体系的结构满足Rouse模型，且体系的分子量分布满足正态分布；对实验体系进行建模，初始化分子量分布的平均值
    Figure PCTCN2022094288-appb-100001
    与标准差ΔM 0；
    b)使用
    Figure PCTCN2022094288-appb-100002
    和ΔM 0构建分子量的正态分布
    Figure PCTCN2022094288-appb-100003
    依据此分布进行采样，建立初始的模拟物理体系；
    Figure PCTCN2022094288-appb-100004
    c)根据模拟物理体系的分子量分布，按照公式(3)计算模式p下的弛豫时间τ ip；
    d)考虑所有振动模式p对储能模量G′和耗散模量G″的贡献，按照公式(1)和(2)计算模拟物理体系的G′和G″。优选地，计算中所用到的高分子在溶液中的密度ρ、溶液的零切黏度η 0和溶剂的粘度η s与实验数据一致；
    e)按照Rouse理论模型得到的模拟数据，可以计算出模拟物理体系的log G′～logω和log G″～logω关系。Rouse模型的预测值log G′、log G″与步骤S2测量实验值的差值ε，可通过均方根公式求解得到；
    f)计算的log G′与log G″数据的差值ε，描述了Rouse模型与实验结果的符合程度。优选地，拟合过程要求Rouse模型的预测结果趋近于实验测量结果；因此，差值ε须尽可能小，则Rouse模型的拟合演化为一个优化问题，优化目标为
    Figure PCTCN2022094288-appb-100005
    差值ε随着高分子的分布情况而变化，利用步骤S3所述的AI算法，对公式(5)的目标函数进行优化；比较实际ε与AI算法的误差阈值ε th，如ε≥ε th，则通过AI算法对正态分布参数
    Figure PCTCN2022094288-appb-100006
    和ΔM进行优化更新，更新后返回过程b)，重新计算Rouse模型；如ε＜ε th，则说明此时的Rouse模型已经能够准确描述实验结果。
  10. 如权利要求1所述的基于AI算法的生物大分子材料分子量的计算方法，其特征在于，所述步骤S5中，所述分子量数据为重均分子量、数均分子量和分子量分布。
    优选地，所述步骤S5中使用AI算法优化的Rouse模型计算分子量方法为：用步骤S1-S4中得到的优化Rouse模型，使用模拟物理体系的分子量分布
    Figure PCTCN2022094288-appb-100007
    完成重均分子量M w、数均分子量M n和分子量分布,即M w/M n的计算。
PCT/CN2022/094288 2022-05-20 2022-05-20 基于ai算法的生物大分子材料分子量的计算方法 Ceased WO2023024607A1 (zh)

Priority Applications (5)

Application Number Priority Date Filing Date Title
EP22859955.1A EP4528573A4 (en) 2022-05-20 2022-05-20 Method for calculating the molecular weight of biomacromolecular material based on an AI algorithm
CN202280097001.5A CN119365865B (zh) 2022-05-20 2022-05-20 基于ai算法的生物大分子材料分子量的计算方法
US18/875,765 US20250226062A1 (en) 2022-05-20 2022-05-20 Artificial Intelligence Algorithm-Based Method for Calculating the Molecular Weight of Biomacromolecular Materials
PCT/CN2022/094288 WO2023024607A1 (zh) 2022-05-20 2022-05-20 基于ai算法的生物大分子材料分子量的计算方法
KR1020247041818A KR20250023989A (ko) 2022-05-20 2022-05-20 Ai 알고리즘에 기반한 생체 거대분자 물질의 분자량 계산 방법

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2022/094288 WO2023024607A1 (zh) 2022-05-20 2022-05-20 基于ai算法的生物大分子材料分子量的计算方法

Publications (1)

Publication Number Publication Date
WO2023024607A1 true WO2023024607A1 (zh) 2023-03-02

Family

ID=85322369

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/094288 Ceased WO2023024607A1 (zh) 2022-05-20 2022-05-20 基于ai算法的生物大分子材料分子量的计算方法

Country Status (5)

Country Link
US (1) US20250226062A1 (zh)
EP (1) EP4528573A4 (zh)
KR (1) KR20250023989A (zh)
CN (1) CN119365865B (zh)
WO (1) WO2023024607A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118924948A (zh) * 2024-05-31 2024-11-12 复向丝泰医疗科技(苏州)有限公司 一种注射用丝蛋白复合羟基磷灰石微球填充剂及其制备方法
WO2025212665A1 (en) * 2024-04-03 2025-10-09 Dow Global Techologies LLC Rheology-based modeling to identify the molecular weight distribution of linear polymers

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103234868A (zh) * 2013-04-25 2013-08-07 常州大学 一种测量线性聚合物重均分子量的方法
US20150212101A1 (en) * 2011-10-06 2015-07-30 Laboratorios Sanifit, S.L. Method for the Direct Detection and/or Quantification of at Least One Compound with a Molecular Weight of at Least 200
JP2016118501A (ja) * 2014-12-22 2016-06-30 住友ゴム工業株式会社 高分子材料のシミュレーション方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3440543A4 (en) * 2016-04-05 2019-08-28 Deakin University SYSTEMS AND METHOD FOR MANUFACTURING A PRODUCT
CN111310348A (zh) * 2020-02-25 2020-06-19 江西理工大学 一种基于pso-lssvm的材料本构模型预测方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150212101A1 (en) * 2011-10-06 2015-07-30 Laboratorios Sanifit, S.L. Method for the Direct Detection and/or Quantification of at Least One Compound with a Molecular Weight of at Least 200
CN103234868A (zh) * 2013-04-25 2013-08-07 常州大学 一种测量线性聚合物重均分子量的方法
JP2016118501A (ja) * 2014-12-22 2016-06-30 住友ゴム工業株式会社 高分子材料のシミュレーション方法

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
DU LIJUN; SHEN JIFANG; WU JUNYI: "Application of Rheological Method in Molecular Weight and Distribution Test of Fluoroplastics", YOUJIFU GONGYE [ORGANO - FLUORINE INDUSTRY], YOUJIFU GONGYE BIANJIBU, CHINA, no. 01, 15 March 2014 (2014-03-15), CHINA , pages 12 - 16, XP009543782, ISSN: 1671-671X *
JIANG, TIQIAN; WANG, ZHIYU: "Rouse Model of Polydisperse Polymer Dilute Solution System", CIESC JOURNAL, HUAXUE GONGYE CHUBANSHE, CN, no. 4, 25 August 1989 (1989-08-25), CN , pages 445 - 452, XP009543768, ISSN: 0438-1157 *
See also references of EP4528573A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2025212665A1 (en) * 2024-04-03 2025-10-09 Dow Global Techologies LLC Rheology-based modeling to identify the molecular weight distribution of linear polymers
CN118924948A (zh) * 2024-05-31 2024-11-12 复向丝泰医疗科技(苏州)有限公司 一种注射用丝蛋白复合羟基磷灰石微球填充剂及其制备方法

Also Published As

Publication number Publication date
CN119365865A (zh) 2025-01-24
CN119365865B (zh) 2025-07-04
US20250226062A1 (en) 2025-07-10
KR20250023989A (ko) 2025-02-18
EP4528573A4 (en) 2025-06-18
EP4528573A1 (en) 2025-03-26

Similar Documents

Publication Publication Date Title
Yin et al. A silk fibroin hydrogel with reversible sol–gel transition
Yang et al. Nonlinear behavior of gelatin networks reveals a hierarchical structure
Raeburn et al. The importance of the self-assembly process to control mechanical properties of low molecular weight hydrogels
Gosal et al. Globular protein gelation
Caliari et al. A practical guide to hydrogels for cell culture
Xiong et al. Topology evolution and gelation mechanism of agarose gel
Clark Structural and mechanical properties of biopolymer gels
Goudoulas et al. Phase transition kinetics and rheology of gelatin-alginate mixtures
Darling et al. Controlling the kinetics of thiol-maleimide Michael-type addition gelation kinetics for the generation of homogenous poly (ethylene glycol) hydrogels
Dahesh et al. Spontaneous gelation of wheat gluten proteins in a food grade solvent
Goudoulas et al. Nonlinear rheological behavior of gelatin gels: In situ gels and individual layers
WO2023024607A1 (zh) 基于ai算法的生物大分子材料分子量的计算方法
Su et al. From micropores to mechanical strength: Fabrication and characterization of edible corn starch-sodium alginate double network hydrogels with Ca2+ cross-linking
Ohnsorg et al. Nonlinear elastic bottlebrush polymer hydrogels modulate actomyosin mediated protrusion formation in mesenchymal stromal cells
Susa et al. Identifying the role of primary and secondary interactions on the mechanical properties and healing of densely branched polyimides
Aufderhorst-Roberts et al. Microrheology and microstructure of Fmoc-derivative hydrogels
Sun et al. Large amplitude oscillatory shear studies on the strain-stiffening behavior of gelatin gels
Schuster et al. Investigating the relationship between network mechanics and single-chain extension using biomimetic polysaccharide gels
Oztoprak et al. Reversibility of strain stiffening in silk fibroin gels
Rao et al. Dynamic rheological measurement of structure development in high‐methoxyl pectin/fructose gels
Legrand et al. Rheological properties of acid-induced carboxymethylcellulose hydrogels
Mhaske et al. Phase volume quantification of agarose-ghee gels using 3D confocal laser scanning microscopy and blending law analysis: A comparison
Burger et al. High-Pressure Effects on Gelatin Sol–Gel Transition
López-Santiago et al. Competition among physical, chemical, and hybrid gelation mechanisms in biopolymers
Eckert et al. An alginate-cellulose based bioink mimics the viscoelastic features of the melanoma microenvironment and its influence on cell cycle and invasion

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22859955

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 202280097001.5

Country of ref document: CN

WWE Wipo information: entry into national phase

Ref document number: 18875765

Country of ref document: US

WWE Wipo information: entry into national phase

Ref document number: 2022859955

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2022859955

Country of ref document: EP

Effective date: 20241218

WWP Wipo information: published in national office

Ref document number: 202280097001.5

Country of ref document: CN

WWP Wipo information: published in national office

Ref document number: 1020247041818

Country of ref document: KR

WWG Wipo information: grant in national office

Ref document number: 202280097001.5

Country of ref document: CN

WWP Wipo information: published in national office

Ref document number: 18875765

Country of ref document: US