Formula optimization method of alkaloid tobacco suckercide
Through systematic mathematical modeling and machine learning algorithm optimization, the problem of optimization of the alkaloid tobacco bud inhibitor formula is solved, and efficient and environmentally friendly bud inhibition effects are achieved, and the health and environmental pollution problems of chemical bud inhibitors are solved.
Patent Information
- Application Number
- CN202411404387.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-10
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art lacks the formulation optimization method for more than one alkaloid compounds as tobacco bud inhibitors, resulting in unsatisfactory bud inhibitors and the risk of health and environmental pollution of chemical bud inhibitors.
Using systematic mathematical modeling methods combined with machine learning algorithms, we establish a set of bud inhibitory effect equations including chemical equilibrium, metabolic process, bud inhibitory activity, duration, growth influence, residual degradation and synergistic effects, and train and genetic algorithms through machine learning algorithms. Optimize and screen out the best alkaloid tobacco bud inhibitor formula.
More accurate bud inhibitor performance prediction is achieved, R&D efficiency is improved, excellent bud inhibition effect is achieved, and the environmental friendliness and low toxicity of the formula is ensured.
Smart Images

Figure CN118942577A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of alkaloid tobacco sprout inhibitors, and in particular relates to a method for optimizing the formula of an alkaloid tobacco sprout inhibitor. Background Art
[0002] Tobacco is an important crop widely grown and used in the world, mainly used to make smoking products such as cigarettes and cigars. In the process of tobacco cultivation, inhibiting the growth of buds is a very critical technical issue. Uninhibited tobacco buds will lead to a decline in plant quality and affect storage and processing quality. At present, chemical bud inhibitors are an effective measure that is widely used. Common chemical bud inhibitors include dimethoate, butofol, etc. These chemical bud inhibitors inhibit the growth of tobacco buds by inhibiting the synthesis of plant growth hormones or interfering with cell division.
[0003] However, there are some problems with the use of existing chemical sprout inhibitors. First, these chemical sprout inhibitors may remain in tobacco, which may have adverse effects on the health of smokers. Second, these chemical sprout inhibitors may cause a certain degree of environmental pollution, such as residues in water and soil. Third, the existing chemical sprout inhibitors also have certain limitations in terms of effectiveness and duration of use, and need to be frequently reused. Therefore, the development of a new type of alkaloid sprout inhibitor has become a hot topic in current research.
[0004] Alkaloids are a class of nitrogen-containing organic compounds naturally present in plants, which are usually physiologically active. Some alkaloid compounds, such as nicotine and anthraquinones, have been shown to have good sprout inhibitory activity. Compared with chemical sprout inhibitors, alkaloid sprout inhibitors have the advantages of being environmentally friendly, biodegradable, and less toxic. However, a single alkaloid compound is usually difficult to achieve the ideal sprout inhibitory effect, and its sprout inhibitory performance needs to be optimized through reasonable formula design. The prior art lacks a method for optimizing the formula of more than one alkaloid compound as a tobacco sprout inhibitor. Summary of the invention
[0005] In view of this, the present invention provides a method for optimizing the formulation of an alkaloid tobacco sprout inhibitor, which can solve the problem in the prior art of lacking a method for optimizing the formulation of more than one alkaloid compound as a tobacco sprout inhibitor.
[0006] The present invention is achieved in that: The present invention provides a method for optimizing the formulation of an alkaloid tobacco sprout inhibitor, which comprises the following steps: S10, obtaining formula data of different formulas of the alkaloid tobacco sprout inhibitor; S20, establishing a group of equations for the bud inhibition effect taking into account the ratio data, the molecular structure of each component, and the metabolic process data of the action mechanism; S30, using a machine learning algorithm, using the acquired formula data of different formulas to train the established sprout inhibition effect equation group, and obtain a preliminary formula-effect relationship model; S40, generating a large number of recipe data sets, and calculating the large number of recipe data sets using the preliminary recipe-effect relationship model in a traversal manner to obtain a large number of candidate recipes and their theoretical effect data; S50, clustering the large number of candidate formulations according to the theoretical effect data to obtain a plurality of cluster centers; S60, taking the formula corresponding to each cluster center as the formula set to be tested, conducting a small-scale field test and collecting actual effect data; S70, using genetic algorithm, setting the crossover rate and mutation rate, and obtaining the optimal formula through multi-generation iterative optimization.
[0007] The formula data include the ratio, the molecular structure and action mechanism of each component, metabolic process data and effect data obtained from specific experiments.
[0008] Furthermore, the effect data specifically include bud inhibition rate, bud inhibition duration, degree of impact on tobacco growth and environmental residue.
[0009] Furthermore, the sprout inhibition effect equation group includes a chemical balance equation, a metabolic process equation, a sprout inhibition activity equation, a duration equation, a growth influence equation, a residual degradation equation and a synergistic effect equation.
[0010] Furthermore, the method for generating a large number of recipe data sets is specifically: according to a preset recipe component range and step length, a traversal method is used to obtain the large number of recipe data sets.
[0011] Furthermore, the method for clustering the large number of candidate recipes is k-means clustering.
[0012] Furthermore, the genetic algorithm uses multiple groups of recipes with the best actual effects in the recipe set to be tested as initial populations.
[0013] Furthermore, the genetic algorithm uses normalized effect data as a fitness function.
[0014] Furthermore, the step of using a machine learning algorithm to train the established sprout inhibition effect equation group using the acquired formula data of different formulas specifically involves using a machine learning algorithm to continuously iteratively optimize the parameters of the sprout inhibition effect equation group using the acquired formula data of different formulas.
[0015] Furthermore, the machine learning algorithm uses a random forest algorithm.
[0016] Further, the specific description of the bud inhibition effect equation group is as follows: 1. Chemical Balance Equation ; In the formula, is the equilibrium constant, dimensionless; is the standard Gibbs free energy change, unit: J / mol; is the gas constant, 8.314 J / (mol·K); is temperature, unit: K; , are the activities of products and reactants, dimensionless; , are the stoichiometric quantities of products and reactants, respectively, and are dimensionless.
[0017] activity Calculation: ; In the formula, is the activity coefficient, dimensionless; For components Concentration, unit: mol / L; The standard concentration is 1 mol / L.
[0018] Activity coefficient Calculation (introducing molecular descriptors): ; In the formula, is the Debye-Hückel constant, which is related to the solvent and temperature and can be obtained by looking up the table; is the ionic charge, dimensionless; is the ionic strength, unit: mol / L, calculation formula: ; Descriptor The weight coefficients of are obtained through multiple regression analysis; For molecules No. A descriptor, dimensionless.
[0019] Among them, the molecular descriptor The method for obtaining is: using computational chemistry software (such as Gaussian, MOPAC, etc.) to calculate. Specific steps: 1. Draw the molecular structure; 2. Optimize the molecular geometry; 3. Calculate the electronic structure; 4. Extract the required descriptors (such as Wiener index, molecular surface area, HOMO-LUMO energy gap, etc.).
[0020] 2. Metabolic process equation ; In the formula, For components Concentration, unit: mol / L; is time, unit: s; For components The generation rate constant, unit: mol / (L·s); For components The elimination rate constant, unit: 1 / s; For enzyme Maximum reaction rate, unit: mol / (L·s); For enzyme The Michaelis constant, unit: mol / L; For components Diffusion coefficient, unit: m² / s; is the Laplace operator.
[0021] Michaelis constant Calculation of (considering enzyme-substrate interactions): ; In the formula, is the basic Michaelis constant, unit: mol / L; For interaction The weight coefficient of , dimensionless; For enzyme Between the substrate The interaction energy is in kJ / mol.
[0022] The interaction energy The method for obtaining is: using molecular docking software (such as AutoDock, GOLD, etc.) to calculate. The steps are as follows: 1. Prepare the three-dimensional structure of the enzyme and substrate; 2. Define the docking region; 3. Set the docking parameters; 4. Run the docking simulation; 5. Analyze the docking results and extract the interaction energy.
[0023] 3. Bud inhibition activity equation ; In the formula, Is the sprout inhibition activity, unit: %; is the maximum sprout inhibition activity, unit: %; is the activity constant, unit: L / mol; For components The weight factor of , dimensionless; Based on molecular descriptors Function of , dimensionless; For components Concentration, unit: mol / L; For components Activity index, dimensionless; is the circadian rhythm influencing factor, dimensionless; is time, unit: h.
[0024] in, The function is constructed by training using machine learning methods (such as random forest, support vector machine, etc.). The steps are as follows: 1. Collect molecular descriptors and sprout inhibition activity data of a large number of compounds; 2. Data preprocessing (standardization, normalization, etc.); 3. Divide the training set and test set; 4. Select a suitable machine learning algorithm; 5. Train the model; 6. Verify the model; 7. Use the trained model to predict the activity of new compounds. value.
[0025] 4. Duration Equation ; In the formula, is the duration, unit: d; is the initial duration, unit: d; is time, unit: d; is the characteristic time, unit: d; For components The persistence factor of , dimensionless; Based on the physicochemical properties Function of , dimensionless; For components The concentration of the substance, unit: mol / L.
[0026] Characteristic time Calculation: ; In the formula, is the pre-exponential factor, unit: d; is the activation energy, unit: J / mol; is the gas constant, 8.314 J / (mol·K); is the temperature, unit: K.
[0027] in, The function is constructed based on physical and chemical properties (such as octanol-water partition coefficient, water solubility, etc.). The steps are as follows: 1. Collect the physical and chemical property data of the compound; 2. Perform correlation analysis and select key properties; 3. Construct a polynomial regression model: ; In the formula, , ,..., , ,..., is the regression coefficient, , ,..., for the selected physicochemical properties.
[0028] 5. Growth Influence Equation ; In the formula, is the tobacco growth rate, unit: cm / d; is the basic growth rate without sprout inhibitor, unit: cm / d; For components Growth inhibition coefficient, unit: L / mol; Based on compound Affinity for the site of action Function of , dimensionless; For components Concentration, unit: mol / L; is the temperature sensitivity coefficient, unit: K-1; - temperature, unit: K; - Humidity sensitivity coefficient, unit: %^(-1); Relative humidity, unit: %.
[0029] in, The function is constructed based on the affinity data obtained by molecular docking or quantum chemical calculation. The steps are as follows: 1. Perform molecular docking or quantum chemical calculation to obtain the binding free energy of the compound and the action site. ; 2. Convert binding free energy into affinity constant : 3. Construction Functions, for example: ,in is the reference affinity constant.
[0030] 6. Residue degradation equation ; In the formula, For components Residue, unit: mol / kgsoil; is time, unit: d; For components Thermal degradation rate constant, unit: 1 / d; For components The photodegradation rate constant, unit: m² / (J·d); For components The hydrolysis rate constant, unit: 1 / (%·d); , , are the activation energies of thermal degradation, photodegradation and hydrolysis, respectively, in J / mol; is the gas constant, 8.314 J / (mol·K); is temperature, unit: K; is the light intensity, unit: J / (m²·d); Relative humidity, unit: % , , Functions that consider molecular stability, photosensitivity, and hydrolytic sensitivity, respectively, dimensionless.
[0031] in, , , The function is constructed based on the molecular structure features (such as chemical bond type, functional group, etc.). The steps are as follows: 1. Analyze the molecular structure and identify key structural features; 2. Assign weights to each feature; 3. Construct the function, for example: ; In the formula, Features The weight of For molecules Medium Features the amount or intensity of.
[0032] 7. Synergy Equation ; In the formula, is the synergy intensity (plural), dimensionless; For components and The synergy coefficient between them is dimensionless; , For components and Concentration, unit: mol / L; For components and The synergy constant between them, unit: mol / L; For components and The phase angle between them, unit: rad; is an imaginary unit; For molecules and The structural similarity coefficient of is dimensionless; For molecules and The similarity coefficient of the mechanism of action is dimensionless.
[0033] Structural similarity coefficient Calculation: ; In the formula, For molecules The number of features in ; For molecules The number of features in ; For molecules and The number of features in common.
[0034] Among them, the structural similarity coefficient The calculation method is: using Tanimoto coefficient. The feature can be a molecular fingerprint or other structural descriptor. The steps are as follows: 1. Generate a molecular fingerprint (such as ECFP, MACCSkeys, etc.); 2. Calculate the number of common features; 3. Calculate the Tanimoto coefficient.
[0035] Similarity coefficient of mechanism of action Calculation: ; In the formula, and Respectively represent compounds and In the A binary value (0 or 1) on a mechanism of action.
[0036] Among them, the similarity coefficient of action mechanism The calculation method is: based on the information of target, signal pathway, etc. The steps are as follows: 1. Collect the target and signal pathway data of the compound; 2. Construct a binary vector to represent the mechanism of action of each compound; 3. Calculate the cosine similarity between vectors.
[0037] 8. Mechanism equation d[Target] dt = k on ⋅[Compound]⋅[Target t free ]- k off ⋅[Compound-Target]- k deg ⋅[Target] [Targe t free ]=[Targe t total ]-[Compound-Target] d[Compound-Target] dt = k on ⋅[Compound]⋅[Target t free ]- k off ⋅[Compound-Target] ; In the formula, [Target] is the total concentration of the target protein, unit: mol / L; [Targe t free ] is the concentration of unbound target protein, unit: mol / L; [Targe t total ] is the total concentration of the target protein, unit: mol / L; [Compound] is the concentration of the sprout inhibitor, unit: mol / L; [Compound-Target] is the concentration of the sprout inhibitor-target protein complex, unit: mol / L; is the binding rate constant, unit: L / (mol·s); is the dissociation rate constant, unit: 1 / s; is the degradation rate constant of the target protein, unit: 1 / s.
[0038] The parameter acquisition method is: 1. and :Use surface plasmon resonance (SPR) experiment to measure. The steps are as follows: a) fix the target protein on the sensor chip; b) pass different concentrations of inhibitor solution; c) record the response curve of the binding and dissociation process; d) use the kinetic model to fit the curve and obtain and .
[0039] 2. :Determined by protein half-life experiment. The steps are as follows: a) express the target protein in cells or in vitro systems; b) add protein synthesis inhibitors (such as cycloheximide); c) collect samples at different time points and determine the protein content by Western blot and other methods; d) fit the protein degradation curve and calculate the half-life ; e) Calculation .
[0040] 3. [Targe t total ] : Determined by quantitative protein analysis methods such as ELISA or mass spectrometry.
[0041] 4. [Compound] : Determined by high performance liquid chromatography (HPLC) or liquid chromatography-mass spectrometry (LC-MS).
[0042] This mechanism equation describes the interaction dynamics between the sprout inhibitor and its target protein, including binding, dissociation, and protein degradation processes. It can help understand the mechanism of action of the sprout inhibitor and predict its effect at different concentrations.
[0043] This set of equations takes into account factors such as molecular structure, physicochemical properties, and mechanism of action, and can more accurately describe and predict the effects of sprout inhibitors.
[0044] Specifically, step S10 includes: collecting formula information of various alkaloid tobacco bud inhibitors available on the market, including the ratio, molecular structure of each component, mechanism of action, metabolic process data, and effect data obtained from specific experiments. These effect data mainly include bud inhibition rate, bud inhibition duration, degree of influence on tobacco growth, and environmental residues. These data can be obtained by consulting relevant literature, patents, enterprise technical information, and other channels.
[0045] Wherein, the step S20 specifically includes: firstly, establishing a chemical equilibrium equation to describe the chemical equilibrium state of the bud inhibitor in the solution, which needs to consider factors such as the standard Gibbs free energy change, the activity of reactants and products, and the calculation of the activity also needs to introduce molecular descriptors. Secondly, establishing a metabolic process equation to describe the absorption, distribution, metabolism and excretion process of the bud inhibitor in the body, which needs to consider the generation rate constant, elimination rate constant, enzyme kinetic parameters and diffusion coefficient, etc. Establishing a bud inhibitor activity equation again to describe the bud inhibitor activity of the bud inhibitor, which needs to consider the maximum bud inhibitor activity, activity constant, weight factors of each component, activity function based on molecular descriptors and influencing factors of circadian rhythm.
[0046] Among them, the step S30 specifically includes: using a variety of machine learning algorithms, such as artificial neural networks, support vector machines, decision trees, etc., through fitting training of a large amount of experimental data, to obtain a preliminary model that can predict the relationship between the sprout inhibitor formula and the effect.
[0047] Wherein, the step S40 specifically includes: setting a reasonable range and variation step of the formula components according to the actual situation, and then using a computer to perform a large number of combination calculations to obtain a large number of theoretically possible formulas and their predicted effect data.
[0048] Wherein, the step S50 specifically includes: using K-means clustering algorithm or other clustering methods to cluster these candidate recipes according to theoretical effect data, so as to obtain multiple representative recipe clustering centers.
[0049] Wherein, the step S60 specifically includes: conducting small-scale field tests on these representative formulations in an actual planting environment to determine their actual sprout inhibition effects, impacts on tobacco growth, etc.
[0050] Wherein, the step S70 specifically includes: using a genetic algorithm for optimization, first using the optimal formula obtained from the experiment as the initial population, normalizing the effect data as the fitness function, performing multiple generations of iterative optimization through operations such as crossover mutation, and finally obtaining the formula with the best performance.
[0051] Optionally, the establishment of the chemical equilibrium equation in step S20 also needs to consider the calculation of the activity coefficient, and the calculation of the activity coefficient needs to introduce molecular descriptors, such as Wiener index, molecular surface area, etc., and obtain the corresponding weight coefficients through multivariate regression analysis; the establishment of the metabolic process equation in step S20 also needs to consider the enzyme-substrate interaction energy, which is calculated through molecular docking software; the construction of the sprout inhibition activity equation in step S20 adopts machine learning methods, such as random forests, support vector machines, etc., which are obtained through training of molecular descriptors and sprout inhibition activity data of a large number of compounds.
[0052] Compared with the prior art, the method for optimizing the formulation of an alkaloid tobacco bud inhibitor provided by the present invention has the following beneficial effects: 1. A systematic mathematical modeling method was adopted to comprehensively consider the ratio, molecular structure, mechanism of action, metabolic process and other factors of alkaloid components, and a set of sprout inhibition effect prediction models including chemical balance, metabolic process, sprout inhibition activity, duration, growth effect, residual degradation and synergistic effect were established. This comprehensive mathematical modeling method can more accurately describe and predict the performance of alkaloid sprout inhibitors, providing a solid theoretical basis for optimizing the formula.
[0053] 2. Use machine learning algorithms and existing experimental data to train the established mathematical model to obtain a preliminary formula-effect relationship model. Then, use the model to perform theoretical calculations and evaluations on a large number of possible formula schemes to screen out multiple potential optimal formulas. This method of combining mathematical modeling and machine learning can greatly improve the research and development efficiency of sprout inhibitor formulas.
[0054] 3. The optimal formula initially screened out was verified through small-scale field trials, and further optimized using genetic algorithms to ultimately obtain the best sprout inhibitor formula. This method of combining experimental verification with algorithm optimization can ensure that the final formula obtained has excellent actual sprout inhibitory effects.
[0055] In general, the alkaloid tobacco sprout inhibitor formulation optimization method proposed in the present invention can more accurately predict the performance of the sprout inhibitor, greatly improve the research and development efficiency, and obtain the optimal sprout inhibitor effect, solving the problem in the prior art of lack of optimization methods for the formulation of more than one alkaloid compound as a tobacco sprout inhibitor. BRIEF DESCRIPTION OF THE DRAWINGS
[0056] Figure 1 A flow chart of the method provided by the present invention. DETAILED DESCRIPTION
[0057] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.
[0058] like Figure 1 As shown, it is a flow chart of a method for optimizing the formulation of an alkaloid tobacco bud inhibitor provided by the present invention, and the method comprises the following steps: S10, obtaining formula data of different formulas of the alkaloid tobacco bud inhibitor, including the ratio, the molecular structure and action mechanism of each component, metabolic process data and effect data obtained from specific experiments, wherein the effect data specifically includes bud inhibition rate, bud inhibition duration, degree of influence on tobacco growth and environmental residue; S20, establishing a group of sprout inhibition effect equations that take into account the ratio data, the molecular structure of each component, and the metabolic process data of the action mechanism, including a chemical equilibrium equation, a metabolic process equation, a sprout inhibition activity equation, a duration equation, a growth influence equation, a residual degradation equation, and a synergistic effect equation; S30, using a machine learning algorithm, and using the obtained formula data of different formulas to train the established equation group to obtain a preliminary formula-effect relationship model; S40, according to the preset formula component range and step length, a large number of formula data sets are obtained by traversing, and the large number of formula data sets are calculated by the preliminary formula-effect relationship model by traversing to obtain a large number of candidate formulas and their theoretical effect data; S50, clustering the large number of candidate formulations according to the theoretical effect data to obtain a plurality of cluster centers; S60, taking the formula corresponding to each cluster center as the formula set to be tested, conducting a small-scale field test and collecting actual effect data; S70, taking multiple groups of recipes with the best actual effects in the recipe set to be tested as the initial population, taking the normalized effect data as the fitness function, setting the crossover rate and mutation rate, using a genetic algorithm, and obtaining the optimal recipe through multi-generation iterative optimization.
[0059] The specific implementation methods of the above steps are described in detail below: Step S10, obtaining the formula data of different formulas of the alkaloid tobacco bud inhibitor. First, it is necessary to collect the formula information of various alkaloid tobacco bud inhibitors available on the market, including the ratio, the molecular structure of each component, the mechanism of action, the metabolic process data and the effect data obtained by specific experiments. These effect data mainly include the bud inhibition rate, the duration of bud inhibition, the degree of influence on tobacco growth and the environmental residue, etc. These data can be obtained by consulting relevant literature, patents, enterprise technical information and other channels.
[0060] Step S20, establishes the sprout inhibition effect equation group considering the ratio data, the molecular structure of each component and the action mechanism metabolic process data. This equation group includes the following parts: 1. Chemical equilibrium equation. This equation describes the chemical equilibrium state of the sprout inhibitor in the solution, and needs to consider factors such as the standard Gibbs free energy change, the activity of reactants and products, etc. The calculation of activity also requires the introduction of molecular descriptors, such as the Wiener index, molecular surface area, etc., and the corresponding weight coefficients are obtained through multivariate regression analysis.
[0061] 2. Metabolic process equation. This equation describes the absorption, distribution, metabolism and excretion process of the sprout inhibitor in the body, and needs to consider the generation rate constant, elimination rate constant, enzyme kinetic parameters (maximum reaction rate, Michaelis constant) and diffusion coefficient, etc. Among them, the calculation of the Michaelis constant also needs to consider the enzyme-substrate interaction energy.
[0062] 3. Sprouting activity equation. This equation describes the sprouting activity of the sprout inhibitor, and needs to consider the maximum sprouting activity, activity constant, weight factor of each component, activity function based on molecular descriptors, and the influencing factor of circadian rhythm. The activity function can be constructed using machine learning methods, such as random forest, support vector machine, etc.
[0063] 4. Duration equation. This equation describes the duration of the sprout inhibition effect, which needs to consider the initial duration, characteristic time constant and the persistence factor of each component. The calculation of the characteristic time constant needs to consider the activation energy factor, and the construction of the persistence factor needs to be based on physical and chemical properties.
[0064] 5. Growth effect equation. This equation describes the effect of bud inhibitors on tobacco growth, and needs to consider the basic growth rate, the growth inhibition coefficient of each component, the function based on the affinity of the compound-action site, and the sensitivity factors of temperature and humidity.
[0065] 6. Residual degradation equation. This equation describes the degradation process of sprout inhibitors in the environment, which needs to consider thermal degradation, photodegradation and hydrolysis, and introduces a degradation function based on molecular structure characteristics.
[0066] 7. Synergistic effect equation. This equation describes the synergistic effect between the components of the sprout inhibitor, and needs to consider the synergistic coefficient, synergistic constant, phase angle, as well as the similarity factors of molecular structure and mechanism of action.
[0067] Step S30, using a machine learning algorithm, using the obtained formula data of different formulas to train the established equation group, and obtain a preliminary formula-effect relationship model. Here, a variety of machine learning algorithms can be used, such as artificial neural networks, support vector machines, decision trees, etc., and a preliminary model that can predict the relationship between the sprout inhibitor formula and the effect can be obtained through fitting training of a large amount of experimental data.
[0068] Step S40, according to the preset formula component range and step length, a large number of formula data sets are obtained by traversing, and the large number of formula data sets are calculated by traversing using the preliminary formula-effect relationship model to obtain a large number of candidate formulas and their theoretical effect data. Here, it is necessary to set the reasonable range and change step length of the formula components according to the actual situation, and then use a computer to perform a large number of combination calculations to obtain a large number of theoretically possible formulas and their predicted effect data.
[0069] Step S50: cluster the large number of candidate recipes according to the theoretical effect data to obtain multiple cluster centers. K-means clustering algorithm or other clustering methods can be used here to cluster the candidate recipes according to the theoretical effect data to obtain multiple representative recipe cluster centers.
[0070] Step S60, taking the formula corresponding to each cluster center as the formula set to be tested, a small-scale field test is conducted to collect actual effect data. This step is to verify the results obtained by the previous theoretical calculation. It is necessary to conduct a small-scale field test on these representative formulas in the actual planting environment to measure their actual bud inhibition effect and impact on tobacco growth.
[0071] Step S70, taking multiple groups of recipes with the best actual effects in the tested recipe set as the initial population, normalizing the effect data as the fitness function, setting the crossover rate and mutation rate, and using a genetic algorithm to obtain the final optimal recipe through multi-generation iterative optimization. Here, a genetic algorithm is used for optimization. First, the optimal recipe obtained by the experiment is used as the initial population, the effect data is normalized as the fitness function, and multi-generation iterative optimization is performed through operations such as crossover and mutation to finally obtain the recipe with the best performance.
[0072] The specific description of the bud inhibition effect equation group is as follows: 1. Chemical Balance Equation ; In the formula, is the equilibrium constant, dimensionless; is the standard Gibbs free energy change, unit: J / mol; is the gas constant, 8.314 J / (mol·K); is temperature, unit: K; , are the activities of products and reactants, dimensionless; , are the stoichiometric quantities of products and reactants, respectively, and are dimensionless.
[0073] activity Calculation: ; In the formula, is the activity coefficient, dimensionless; For components Concentration, unit: mol / L; The standard concentration is 1 mol / L.
[0074] Activity coefficient Calculation (introducing molecular descriptors): ; In the formula, is the Debye-Hückel constant, which is related to the solvent and temperature and can be obtained by looking up the table; is the ionic charge, dimensionless; is the ionic strength, unit: mol / L, calculation formula: ; Descriptor The weight coefficients of are obtained through multiple regression analysis; For molecules No. A descriptor, dimensionless.
[0075] Among them, the molecular descriptor The method for obtaining is: using computational chemistry software (such as Gaussian, MOPAC, etc.) to calculate. Specific steps: 1. Draw the molecular structure; 2. Optimize the molecular geometry; 3. Calculate the electronic structure; 4. Extract the required descriptors (such as Wiener index, molecular surface area, HOMO-LUMO energy gap, etc.).
[0076] 2. Metabolic process equation ; In the formula, For components Concentration, unit: mol / L; is time, unit: s; For components The generation rate constant, unit: mol / (L·s); For components The elimination rate constant, unit: 1 / s; For enzyme Maximum reaction rate, unit: mol / (L·s); For enzyme The Michaelis constant, unit: mol / L; For components Diffusion coefficient, unit: m² / s; is the Laplace operator.
[0077] Michaelis constant Calculation of (considering enzyme-substrate interactions): ; In the formula, is the basic Michaelis constant, unit: mol / L; For interaction The weight coefficient of , dimensionless; For enzyme Between the substrate The interaction energy is in kJ / mol.
[0078] The interaction energy The method for obtaining is: using molecular docking software (such as AutoDock, GOLD, etc.) to calculate. The steps are as follows: 1. Prepare the three-dimensional structure of the enzyme and substrate; 2. Define the docking region; 3. Set the docking parameters; 4. Run the docking simulation; 5. Analyze the docking results and extract the interaction energy.
[0079] 3. Bud inhibition activity equation ; In the formula, Is the sprout inhibition activity, unit: %; is the maximum sprout inhibition activity, unit: %; is the activity constant, unit: L / mol; For components The weight factor of , dimensionless; Based on molecular descriptors Function of , dimensionless; For components The concentration, unit: mol / L; For components Activity index, dimensionless; is the circadian rhythm influencing factor, dimensionless; is time, unit: h.
[0080] in, The function is constructed by training using machine learning methods (such as random forest, support vector machine, etc.). The steps are as follows: 1. Collect molecular descriptors and sprout inhibition activity data of a large number of compounds; 2. Data preprocessing (standardization, normalization, etc.); 3. Divide the training set and test set; 4. Select a suitable machine learning algorithm; 5. Train the model; 6. Verify the model; 7. Use the trained model to predict the activity of new compounds. value.
[0081] 4. Duration Equation ; In the formula, is the duration, unit: d; is the initial duration, unit: d; is time, unit: d; is the characteristic time, unit: d; For components The persistence factor of , dimensionless; Based on the physicochemical properties Function of , dimensionless; For components The concentration of the substance, unit: mol / L.
[0082] Characteristic time Calculation: ; In the formula, is the pre-exponential factor, unit: d; is the activation energy, unit: J / mol; is the gas constant, 8.314 J / (mol·K); is the temperature, unit: K.
[0083] in, The function is constructed based on physical and chemical properties (such as octanol-water partition coefficient, water solubility, etc.). The steps are as follows: 1. Collect the physical and chemical property data of the compound; 2. Perform correlation analysis and select key properties; 3. Construct a polynomial regression model: ; In the formula, , ,..., , ,..., is the regression coefficient, , ,..., for the selected physicochemical properties.
[0084] 5. Growth Influence Equation ; In the formula, is the tobacco growth rate, unit: cm / d; is the basic growth rate without sprout inhibitor, unit: cm / d; For components Growth inhibition coefficient, unit: L / mol; Based on compound Affinity for the site of action Function of , dimensionless; For components Concentration, unit: mol / L; is the temperature sensitivity coefficient, unit: K-1; - temperature, unit: K; - Humidity sensitivity coefficient, unit: %^(-1); Relative humidity, unit: %.
[0085] in, The function is constructed based on the affinity data obtained by molecular docking or quantum chemical calculation. The steps are as follows: 1. Perform molecular docking or quantum chemical calculation to obtain the binding free energy of the compound and the action site. ; 2. Convert binding free energy into affinity constant : 3. Construction Functions, for example: ,in is the reference affinity constant.
[0086] 6. Residue degradation equation ; In the formula, For components Residue, unit: mol / kgsoil; is time, unit: d; For components Thermal degradation rate constant, unit: 1 / d; For components The photodegradation rate constant, unit: m² / (J·d); For components The hydrolysis rate constant, unit: 1 / (%·d); , , are the activation energies of thermal degradation, photodegradation and hydrolysis, respectively, in J / mol; is the gas constant, 8.314 J / (mol·K); is temperature, unit: K; is the light intensity, unit: J / (m²·d); Relative humidity, unit: % , , Functions that consider molecular stability, photosensitivity, and hydrolytic sensitivity, respectively, dimensionless.
[0087] in, , , The function is constructed based on the molecular structure features (such as chemical bond type, functional group, etc.). The steps are as follows: 1. Analyze the molecular structure and identify key structural features; 2. Assign weights to each feature; 3. Construct the function, for example: ; In the formula, Features The weight of For molecules Medium Features the amount or intensity of.
[0088] 7. Synergy Equation ; In the formula, is the synergy intensity (plural), dimensionless; For components and The synergy coefficient between them is dimensionless; , For components and The concentration, unit: mol / L; For components and The synergy constant between them, unit: mol / L; For components and The phase angle between them, unit: rad; is an imaginary unit; For molecules and The structural similarity coefficient of is dimensionless; For molecules and The similarity coefficient of the mechanism of action is dimensionless.
[0089] Structural similarity coefficient Calculation: ; In the formula, For molecules The number of features in ; For molecules The number of features in ; For molecules and The number of features in common.
[0090] Among them, the structural similarity coefficient The calculation method is: using Tanimoto coefficient. The feature can be a molecular fingerprint or other structural descriptor. The steps are as follows: 1. Generate a molecular fingerprint (such as ECFP, MACCSkeys, etc.); 2. Calculate the number of common features; 3. Calculate the Tanimoto coefficient.
[0091] Similarity coefficient of mechanism of action Calculation: ; In the formula, and Respectively represent compounds and In the A binary value (0 or 1) on a mechanism of action.
[0092] Among them, the similarity coefficient of action mechanism The calculation method is: based on the information of target, signal pathway, etc. The steps are as follows: 1. Collect the target and signal pathway data of the compound; 2. Construct a binary vector to represent the mechanism of action of each compound; 3. Calculate the cosine similarity between vectors.
[0093] 8. Mechanism equation d[Target] dt = k on ⋅[Compound]⋅[Target t free ]- k off ⋅[Compound-Target]- k deg ⋅[Target] [Targe t free ]=[Targe t total ]-[Compound-Target] d[Compound-Target] dt = k on ⋅[Compound]⋅[Target t free ]- k off ⋅[Compound-Target] ; In the formula, [Target] is the total concentration of the target protein, unit: mol / L; [Targe t free ] is the concentration of unbound target protein, unit: mol / L; [Targe t total ] is the total concentration of the target protein, unit: mol / L; [Compound] is the concentration of the sprout inhibitor, unit: mol / L; [Compound-Target] is the concentration of the sprout inhibitor-target protein complex, unit: mol / L; is the binding rate constant, unit: L / (mol·s); is the dissociation rate constant, unit: 1 / s; is the degradation rate constant of the target protein, unit: 1 / s.
[0094] The parameter acquisition method is: 1. and :Use surface plasmon resonance (SPR) experiment to measure. The steps are as follows: a) fix the target protein on the sensor chip; b) pass different concentrations of inhibitor solution; c) record the response curve of the binding and dissociation process; d) use the kinetic model to fit the curve and obtain and .
[0095] 2. :Determined by protein half-life experiment. The steps are as follows: a) express the target protein in cells or in vitro systems; b) add protein synthesis inhibitors (such as cycloheximide); c) collect samples at different time points and determine the protein content by Western blot and other methods; d) fit the protein degradation curve and calculate the half-life ; e) Calculation .
[0096] 3. [Targe t total ] : Determined by quantitative protein analysis methods such as ELISA or mass spectrometry.
[0097] 4. [Compound] : Determined by high performance liquid chromatography (HPLC) or liquid chromatography-mass spectrometry (LC-MS).
[0098] This mechanism equation describes the interaction dynamics between the sprout inhibitor and its target protein, including binding, dissociation, and protein degradation processes. It can help understand the mechanism of action of the sprout inhibitor and predict its effect at different concentrations.
[0099] This set of equations takes into account factors such as molecular structure, physicochemical properties, and mechanism of action, and can more accurately describe and predict the effects of sprout inhibitors.
[0100] The following is a specific implementation method for obtaining relevant parameters: Chemical equilibrium equation parameter acquisition: (Equilibrium constant): How to obtain: experimental determination or theoretical calculation. Experimental steps: Prepare solutions of known concentrations of reactants, allow the reaction to reach equilibrium, determine the concentrations of each component at equilibrium, and use the equilibrium constant formula to calculate .
[0101] (Standard Gibbs free energy change): How to obtain: thermodynamic database query or calorimetric determination. Calorimetric method steps: Use an adiabatic calorimeter to measure the heat of reaction, determine the temperature change before and after the reaction, and calculate the enthalpy change , measure or calculate entropy change , using the formula calculate.
[0102] (Gas Constant): Fixed value: 8.314 J / (mol·K) How to obtain temperature: Use a calibrated thermometer to measure.
[0103] , (activity): Acquisition method: experimental determination or theoretical calculation. Experimental steps (taking vapor pressure method as an example): Measure the vapor pressure of pure solute and solution, using the formula Calculate, where is the partial pressure of component i in the solution, is the saturated vapor pressure of pure component i.
[0104] , How to obtain (stoichiometric coefficient): Determine based on the balanced reaction equation.
[0105] (Activity coefficient): How to obtain: experimental determination or calculation using the activity coefficient model. Experimental method (taking isothermal distillation as an example): prepare a binary mixture of known composition, partially distill the mixture at a constant temperature, analyze the composition of the vapor and liquid, and use the formula Calculate, where and are the mole fractions of component i in vapor and liquid, respectively, is the total pressure.
[0106] Method for obtaining (concentration): experimental determination, such as titration, spectrophotometry, etc.
[0107] (Standard state concentration): fixed value: 1 mol / L; How to obtain (Debye-Hückel constant): Obtained by looking up the table, it is related to the solvent and temperature.
[0108] How to obtain (ion charge): Determined based on the chemical formula of the ion.
[0109] (Ionic strength) acquisition method: calculation or experimental determination. Calculation formula: (Descriptor weight coefficient): Acquisition method: Multiple regression analysis. Steps: Collect a large number of compound descriptor data and experimentally determined activity coefficients, use multiple linear regression or other machine learning methods to build a model, and obtain the weight coefficient of each descriptor .
[0110] (Molecular descriptors): Acquisition method: Calculation by computational chemistry software. Steps: Use software (such as Gaussian, MOPAC) to draw molecular structures, optimize molecular geometry, calculate electronic structures, and extract required descriptors (such as Wiener index, molecular surface area, HOMO-LUMO energy gap, etc.).
[0111] Obtaining parameters of metabolic process equations: (Generation rate constant): Method for obtaining: experimental determination or kinetic modeling. Experimental steps: monitor the generation of the compound in an in vitro or in vivo system, measure the concentration at different time points, fit the concentration-time curve, and obtain .
[0112] (elimination rate constant): Obtaining method: experimental determination or kinetic modeling. Experimental steps: administer known doses of compounds, monitor blood drug concentrations at different time points, fit the concentration-time curve, and obtain .
[0113] (maximum reaction rate): Method for obtaining: enzyme kinetics experiment. Steps: prepare substrate solutions of different concentrations, add a fixed amount of enzyme, measure the initial reaction rate under different substrate concentrations, draw a Lineweaver-Burk plot or use nonlinear regression to obtain .
[0114] (Michaelis constant) acquisition method: The same enzyme kinetics experiments were performed.
[0115] (Diffusion coefficient): Acquisition method: experimental determination or theoretical calculation. Experimental method (Taylor diffusion method as an example): Inject the tracer into the flowing carrier liquid, measure the concentration distribution of the tracer at different positions, and calculate the diffusion coefficient based on the variance of the concentration distribution.
[0116] (Basic Michaelis constant): How to obtain: Determine under conditions without interaction effects .
[0117] (Interaction weight coefficient): How to obtain: Multiple regression analysis, similar to Method of obtaining .
[0118] (Interaction energy): Method for obtaining: molecular docking or quantum chemical calculation. Steps: prepare the three-dimensional structure of enzyme and substrate, use molecular docking software (such as AutoDock, GOLD) for simulation, analyze the docking results, and extract the interaction energy.
[0119] Parameters of the sprout inhibition activity equation are obtained: (maximum bud inhibition activity): acquisition method: experimental determination. Steps: prepare a series of bud inhibition agent solutions, treat tobacco samples, measure the bud inhibition effect at the highest concentration, and obtain .
[0120] (Activity constant) acquisition method: nonlinear regression fitting of experimental data.
[0121] (Weight factor) acquisition method: Multiple regression analysis, similar to Method of obtaining .
[0122] (Function based on molecular descriptors): Acquisition method: machine learning method (such as random forest, support vector machine). Steps: Collect molecular descriptors and sprout inhibition activity data of a large number of compounds, preprocess data (standardization, normalization), divide training set and test set, select appropriate machine learning algorithm, train model, verify model, use trained model to build function.
[0123] Method for obtaining (activity index): experimental determination or structure-activity relationship (SAR) analysis.
[0124] (Circadian rhythm influencing factor): Acquisition method: Experimental determination. Steps: Measure the inhibitory activity at different time points, fit the 24-hour period sine function, and obtain value.
[0125] Duration equation parameter acquisition: (Initial duration): Method of obtaining: Experimental determination. Steps: Apply the sprout inhibitor, monitor the sprout inhibitor effect until the effect disappears, and record the duration as .
[0126] Method for obtaining (characteristic time): experimental determination or calculation using the Arrhenius equation.
[0127] Method for obtaining (persistent factors): multiple regression analysis, similar to Method of obtaining .
[0128] (Based on the function of physicochemical properties): Acquisition method: Polynomial regression. Steps: Collect the physicochemical property data of the compound, perform correlation analysis, select key properties, and build a polynomial regression model.
[0129] Method for obtaining (pre-exponential factor): obtained through the linearized form of the Arrhenius equation.
[0130] (Activation energy) acquisition method: experimental determination or theoretical calculation. Experimental method (taking Arrhenius plot method as an example): measure the reaction rate at different temperatures and draw right The graph of .
[0131] Growth influence equation parameter acquisition: Method for obtaining (basic growth rate): experimental determination. Steps: measure the growth rate of tobacco without bud inhibitors and calculate the average growth per unit time.
[0132] Method for obtaining (growth inhibition coefficient): experimental determination. Steps: Apply different concentrations of sprout inhibitors, measure the corresponding growth inhibition effect, fit the concentration-inhibition curve, and obtain .
[0133] (Affinity-based function): Method for obtaining: Based on molecular docking or quantum chemical calculation. Steps: Perform molecular docking or quantum chemical calculation to obtain binding free energy , calculate the affinity constant: , construct functions such as .
[0134] (Temperature sensitivity coefficient): Method of obtaining: experimental determination. Steps: Measure the growth rate at different temperatures, calculate the partial derivative of the growth rate with respect to temperature, and normalize to obtain .
[0135] (Humidity sensitivity coefficient): Acquisition method: Experimental determination, similar to Method of obtaining .
[0136] Residual degradation equation parameter acquisition: , , (Degradation rate constant): Acquisition method: Experimental determination. Steps: Treat the sample under controlled conditions (constant temperature, light or humidity), measure the residual amount at different time points, fit the first-order kinetic equation, and obtain the corresponding rate constant.
[0137] , , (Activation Energy): Method of obtaining: Arrhenius plot method, similar to the one described above Get method.
[0138] , , (Molecular stability, photosensitivity and hydrolysis sensitivity function): Acquisition method: Based on molecular structure features. Steps: Analyze the molecular structure, identify key structural features, assign weights to each feature, and construct functions, such as .
[0139] Synergistic equation parameter acquisition: (Synergy coefficient): Determine by in vitro or in vivo experiments. The steps are as follows: Design components with different concentration ratios and The sprout inhibition effect of the mixture was determined and compared with the effect of each component used alone. The experimental data were analyzed using the isobolite method or isoquant method, and the synergistic coefficient was calculated.
[0140] , (Concentration of components): High performance liquid chromatography (HPLC) or liquid chromatography-mass spectrometry (LC-MS) is used for determination. The steps are as follows: prepare a standard curve, extract and treat samples, perform chromatographic analysis, and calculate the concentration based on peak area or peak height.
[0141] (Synergy constant): The experimental data were obtained by nonlinear regression analysis. The steps are as follows: design experiments with different concentration combinations, determine the synergistic effect strength, and use nonlinear regression software (such as GraphPadPrism) to fit the data to obtain value.
[0142] (Phase Angle): Determined by time series experiment. The steps are as follows: measure the components at different time points and The synergistic effect strength is calculated by Fourier transforming the time series data. and The phase difference between them.
[0143] (Structural similarity coefficient): Calculated using the Tanimoto coefficient: ,in , The molecules and The number of features in is the number of shared features. The steps are as follows: generate molecular fingerprints using cheminformatics software (such as RDKit), calculate the number of shared features, and apply the Tanimoto formula to calculate the similarity coefficient.
[0144] (Similarity coefficient of mechanism of action): Based on cosine similarity calculation: ,in and It is a compound and In the The steps are as follows: collect the target and signal pathway data of the compound, construct a binary vector to represent the mechanism of action of each compound, and calculate the cosine similarity between the vectors.
[0145] 8. Obtaining parameters of the action mechanism equation: (binding rate constant) and (Dissociation rate constant): The surface plasmon resonance (SPR) experiment was used for determination. The steps are as follows: the target protein is fixed on the sensor chip, different concentrations of inhibitor solution are introduced, the response curve of the binding and dissociation process is recorded, and the curve is fitted using a kinetic model to obtain and .
[0146] (Target protein degradation rate constant): Determined by protein half-life experiment. The steps are as follows: a) Express the target protein in cells or in vitro systems; b) adding protein synthesis inhibitors (such as cycloheximide); c) Collect samples at different time points and determine the protein content by Western blot or other methods; d) Fitting protein degradation curve and calculating half-life ; e) Calculation [Targe t total ] (Total concentration of target protein): Determined by quantitative protein analysis methods, such as enzyme-linked immunosorbent assay (ELISA) or mass spectrometry. The steps are as follows: a) Preparation of samples and standards; b) performing ELISA or mass spectrometry analysis; c) Calculate protein concentration based on the standard curve; [Compound] (Concentration of sprout inhibitor): The concentration of the components was determined by high performance liquid chromatography (HPLC) or liquid chromatography-mass spectrometry (LC-MS). The steps were the same as those mentioned above.
[0147] [Compound-Target] (Concentration of sprout inhibitor-target protein complex): It can be determined by equilibrium dialysis or ultrafiltration experiments combined with mass spectrometry. The steps are as follows: a) mixing the sprout inhibitor with the target protein to achieve equilibrium; b) separating the free sprout inhibitor and the complex by equilibrium dialysis or ultrafiltration; c) determining the concentration of the complex by mass spectrometry; [Targe t free ] (Unbound target protein concentration): Calculated according to the principle of conservation of mass: [Targe t free ]=[Targe t total ]-[Compound-Target] ; The methods for obtaining these parameters cover a variety of experimental techniques and data analysis methods, which can help to deeply understand the synergistic effects and mechanisms of action of sprout inhibitors.
[0148] Specifically, the principle of the present invention is: First, in step S10, the formula data of the existing alkaloid sprout inhibitor is obtained, including the ratio, molecular structure, mechanism of action, metabolic process and experimental effect data. These data lay the foundation for the subsequent mathematical modeling.
[0149] In step S20, corresponding mathematical equations are established for each of the above aspects to form a complete model for predicting the sprout inhibition effect. Among them, the chemical equilibrium equation describes the chemical equilibrium state of alkaloids in the solution; the metabolic process equation describes the absorption, distribution, metabolism and excretion process of alkaloids in the body; the sprout inhibition activity equation describes the sprout inhibition activity of alkaloids; the duration equation describes the duration of the sprout inhibition effect; the growth influence equation describes the effect of alkaloids on tobacco growth; the residual degradation equation describes the degradation process of alkaloids in the environment; and the synergistic effect equation describes the synergistic effect between alkaloid components. These equations fully consider multiple key factors such as molecular structure, physicochemical properties, and mechanism of action, and can more accurately simulate and predict the performance of alkaloid sprout inhibitors.
[0150] In step S30, the established mathematical model is trained using a machine learning algorithm to obtain a preliminary formula-effect relationship model. This method of integrating mathematical modeling and machine learning can make full use of existing experimental data and improve the prediction accuracy of the model.
[0151] In steps S40-S50, the established formula-effect relationship model is used to perform theoretical calculations and cluster analysis on a large number of possible formula solutions to screen out multiple potential optimal formulas. This method of numerical calculation and cluster analysis can greatly improve the research and development efficiency of sprout inhibitor formulas.
[0152] Finally, in steps S60-S70, the optimal formula initially screened is experimentally verified and further optimized using a genetic algorithm to ultimately obtain the best sprout inhibitor formula. This method of combining experimental verification with algorithm optimization can ensure that the final formula obtained has an excellent sprout inhibitor effect in practical applications.
[0153] In general, the present invention makes full use of advanced technical means such as mathematical modeling, machine learning, numerical calculation and optimization algorithm to systematically study the formula-effect relationship of alkaloid sprout inhibitors, and finally obtains the sprout inhibitor formula with the best performance.
[0154] In order to better understand and implement the present invention, a specific embodiment 1 of the present invention is provided below. The steps included in this embodiment 1 are as follows: First, in step S10, different formula data of alkaloid tobacco sprout inhibitors need to be obtained, including ratio, molecular structure, mechanism of action, metabolic process and experimental effect data. Among them, the experimental effect data includes the sprout inhibition rate , Bud suppression duration , Impact on tobacco growth and environmental residues wait.
[0155] In step S20, it is necessary to establish a group of equations for the sprout inhibition effect taking into account the above factors. Among them, the chemical equilibrium equation can be expressed as: ; in, is the equilibrium constant, is the standard Gibbs free energy change, is the gas constant, is temperature, and are the activities of the products and reactants, respectively, and are their stoichiometric numbers respectively. The calculation can be expressed as: ; in, is the activity coefficient, For components The concentration of is the standard concentration. Activity coefficient The calculation of can be done by the following formula, introducing the molecular descriptor : ; in, is the Debye-Hückel constant, is the ionic charge, is the ionic strength, Descriptor Weight coefficient of . Molecular descriptor It can be calculated by computational chemistry software, including Wiener index, molecular surface area, HOMO-LUMO energy gap, etc.
[0156] The metabolic process equation can be expressed as: ; in, For components The concentration of For time, and The components The generation and elimination rate constants, and Enzymes The maximum reaction rate and Michaelis constant, For components The diffusion coefficient of The calculation can be done using: ; in, is the basic Michaelis constant, For interaction The weight coefficient, For enzyme Between the substrate The interaction energy can be calculated by molecular docking software.
[0157] The germination inhibition activity equation can be expressed as: in, The maximum inhibitory activity is the activity constant, For components The weight factor, Based on molecular descriptors The function of For components The activity index, Factors affecting circadian rhythm. Functions can be constructed using machine learning methods, such as random forests, support vector machines, etc.
[0158] The duration equation can be expressed as: ; in, is the initial duration, is the characteristic time constant, For components The persistence factor, Based on the physicochemical properties Function of. Characteristic time constant The calculation can be done using: ; in, is the pre-exponential factor, is the activation energy. Functions can be constructed through polynomial regression, such as: ; The growth effect equation can be expressed as: ; in, is the basic growth rate without sprout inhibitor, For components The growth inhibition coefficient, Based on compound Affinity for the site of action The function of and are the sensitivity coefficients of temperature and humidity, respectively. Functions can be constructed based on affinity data obtained from molecular docking or quantum chemical calculations, such as: ; in, is the affinity constant, is the reference affinity constant.
[0159] The residual degradation equation can be expressed as: ; in, For components The residual amount, , and are the rate constants of thermal degradation, photodegradation and hydrolysis, respectively. , and are the corresponding activation energies, is the light intensity, is the relative humidity. , and To consider the function of molecular stability, photosensitivity and hydrolysis sensitivity, it can be constructed based on molecular structural characteristics, such as: ; in, Features The weight of For molecules Medium Features the amount or intensity of.
[0160] The synergistic effect equation can be expressed as: ; in, For components and The synergy coefficient between is the synergy constant, is the phase angle, For molecules and The structural similarity coefficient of is the similarity coefficient of mechanism of action. The Tanimoto coefficient can be used for calculation: ; in, and The molecules and The number of features in is the number of common features. Similarity coefficient of mechanism of action Cosine similarity can be calculated based on target and signal pathway information: ; in, and Respectively represent compounds and In the A binary value (0 or 1) on a mechanism of action.
[0161] The action mechanism equation can be expressed as: d[Target] dt = k on ⋅[Compound]⋅[Target t free ]- k off ⋅[Compound-Target]- k deg ⋅[Target] ; [Targe t free ]=[Targe t total ]-[Compound-Target] d[Compound-Target] dt = k on ⋅[Compound]⋅[Target t free ]- k off ⋅[Compound-Target] ; in, [Target] is the total concentration of the target protein, [Targe t free ] is the concentration of unbound target protein, [Targe t total ] is the total concentration of the target protein, [Compound] is the concentration of the sprout inhibitor, [Compound-Target] is the concentration of the sprout inhibitor-target protein complex, and are the association and dissociation rate constants, is the degradation rate constant of the target protein. These parameters can be determined by surface plasmon resonance (SPR) experiments and protein half-life experiments.
[0162] In order to further better understand and implement the present invention, the following provides Example 2 of a specific application scenario of the present invention: In recent years, a tobacco research unit has been studying alkaloid tobacco bud inhibitors, hoping to develop a new type of bud inhibitor that can effectively inhibit the growth of tobacco buds and is environmentally friendly and non-toxic. In the previous research, the researchers screened out several alkaloid compounds with good bud inhibition activity, including nicotine, anthraquinone compounds and some alkaloid derivatives. However, a single alkaloid compound is often difficult to achieve the ideal bud inhibition effect, so it was decided to use the formula optimization method proposed in the present invention to conduct a systematic formula study on these alkaloid compounds.
[0163] First, in step S10, detailed information of these alkaloid compounds is collected, including molecular structure, physical and chemical properties, mechanism of action, metabolic process and existing bud inhibition test data. Taking nicotine as an example, its molecular structure is , belongs to the pyridine alkaloids, widely found in tobacco, Solanaceae plants, etc. The mechanism of action of nicotine is mainly to inhibit the synthesis and transport of auxin, thereby inhibiting the development of buds. The metabolic process of nicotine in the body includes absorption, distribution, metabolism and excretion, involving a series of enzymatic reactions. According to the existing bud inhibition test data, pure nicotine at a concentration of 300 mg / L can inhibit the budding rate of tobacco buds by up to 80%, but the duration of bud inhibition is only about 10 days. Another anthraquinone compound, juglone, has a molecular structure of , which belongs to the terpenoid alkaloids, also shows certain sprout inhibition activity. The mechanism of action of juglone is to inhibit cell respiration, thereby inhibiting the growth of buds. At a concentration of 150mg / L, the sprout inhibition rate of juglone can reach 70%, and the duration is about 15 days.
[0164] Based on the above data, the researchers entered step S20 and established a complete model for predicting the sprout inhibition effect. First, in terms of chemical equilibrium, the researchers considered the dissociation equilibrium of alkaloids in aqueous solution: ; in, is the dissociation equilibrium constant, is the standard Gibbs free energy change, is the gas constant, is temperature, and are the activities of the protonated and neutral forms, respectively. The calculation of activity needs to take into account the ionic strength of the solution, using the modified Debye-Hückel equation: in, is the activity coefficient, is the ionic charge, is the ionic strength, is the Debye-Hückel constant.
[0165] For the metabolic process, researchers have established a kinetic equation that includes absorption, distribution, metabolism and excretion: in, is the alkaloid concentration, and are the generation and elimination rate constants, and is the kinetic parameter of the enzyme involved in metabolism, is the diffusion coefficient. Here, the enzyme-substrate interaction also needs to be considered, and the molecular docking method is used to calculate .
[0166] For the sprout inhibition activity, the researchers constructed the following equation: ; in, The maximum inhibitory activity is the activity constant, For components The weight factor, Based on molecular descriptors The function of For components The activity index, Factors affecting circadian rhythm. The function is trained by the random forest algorithm.
[0167] The duration equation is as follows: ; in, is the initial duration, is the characteristic time constant, For components The persistence factor, Based on the physicochemical properties Function of. Characteristic time constant The Arrhenius equation was used to calculate, It is constructed through polynomial regression.
[0168] The growth effect equation is: ; in, is the basic growth rate without sprout inhibitor, For components The growth inhibition coefficient, Compound-site affinity The function of and are the sensitivity coefficients of temperature and humidity respectively. The function was obtained through molecular docking calculation.
[0169] The residual degradation equation takes into account three main degradation mechanisms: thermal degradation, photodegradation and hydrolysis: ; in, For components The residual amount, , and are the rate constants of thermal degradation, photodegradation and hydrolysis, respectively. , and is the corresponding activation energy, is the light intensity, is the relative humidity. , and In order to consider the degradation function of molecular structure characteristics, it is constructed by analyzing factors such as chemical bond type and functional group.
[0170] Finally, the synergistic effect between the alkaloid components is described by the synergistic equation: ; in, For components and The synergy coefficient, is the synergy constant, is the phase angle, is the structural similarity coefficient, is the similarity coefficient of the mechanism of action. The structural similarity was calculated using the Tanimoto coefficient, and the mechanism of action similarity was calculated based on the target and signaling pathway information.
[0171] Through the above mathematical modeling, the researchers established a relatively complete sprout inhibition effect prediction model. In step S30, they used the random forest algorithm to train and optimize these model parameters to obtain a preliminary formula-effect relationship model.
[0172] Next, in steps S40 and S50, the researchers generated a large number of possible formula solutions based on the preset formula component range using computer traversal, and used the trained model to perform theoretical calculations to obtain the predicted effect data of each formula. Then, they performed cluster analysis on these candidate formulas and selected 5 representative formulas for small-scale field trial verification, as shown in Table 1.
[0173] Table 1 Specific composition of representative formulas and their predicted effects formula Nicotine (mg / L) Juglone(mg / L) Other derivatives (mg / L) Prediction of germination inhibition rate Prediction duration 1 200 100 - 82% 18 days 2 150 150 - 78% 22 days 3 150 50 Nicotine derivative A100 85% 20 days 4 100 100 Juglone derivative B50 83% 24 days 5 200 - Nicotine derivative C150 78% 16 days In step S60, the R&D personnel conducted a small-scale field test of the five formulas in a tobacco plantation. The experimental results showed that Formula 1 and Formula 3 had the best actual sprout inhibition effect, with a sprout inhibition rate of more than 80% and a duration of about 18-20 days, which was basically consistent with the theoretical prediction. Formula 2 and Formula 4 were second, with a slightly lower sprout inhibition rate but a longer duration. The actual effect of Formula 5 was not ideal, with a sprout inhibition rate of only about 75% and a shorter duration.
[0174] Based on these experimental results, the researchers selected Formula 1 and Formula 3 as the initial population and further optimized them using a genetic algorithm (step S70). After 200 generations of iterative calculations, the optimal formula was finally obtained: 195 mg / L nicotine and 105 mg / L juglone. This optimized formula showed a higher bud inhibition rate (87%) and duration (22 days) in a small-scale test, with less impact on tobacco growth and lower environmental residues.
[0175] In order to verify the performance of this optimal formula in actual production, a large-scale field test was conducted in tobacco plantations in two different regions. The test results showed that the formula can maintain excellent sprout inhibition effect under different environmental conditions, with an average sprout inhibition rate of 85% and a duration of about 3 weeks. It has little impact on tobacco growth and environmental residues are also controlled at a low level.
[0176] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A method for optimizing the formulation of an alkaloid tobacco sprout inhibitor, characterized in that: The following steps are involved: S10, obtaining formula data of different formulas of the alkaloid tobacco sprout inhibitor; S20, establishing a group of equations for the bud inhibition effect taking into account the ratio data, the molecular structure of each component, and the metabolic process data of the action mechanism; S30, using a machine learning algorithm, using the acquired formula data of different formulas to train the established sprout inhibition effect equation group, and obtain a preliminary formula-effect relationship model; S40, generating a large number of recipe data sets, and calculating the large number of recipe data sets using the preliminary recipe-effect relationship model in a traversal manner to obtain a large number of candidate recipes and their theoretical effect data; S50, clustering the large number of candidate formulations according to the theoretical effect data to obtain a plurality of cluster centers; S60, taking the formula corresponding to each cluster center as the formula set to be tested, conducting a small-scale field test and collecting actual effect data; S70, using genetic algorithm, setting the crossover rate and mutation rate, and obtaining the optimal formula through multi-generation iterative optimization.
2. The method for optimizing the formulation of an alkaloid tobacco sprout inhibitor according to claim 1, characterized in that: The formula data includes the ratio, the molecular structure and action mechanism of each component, metabolic process data and effect data obtained from specific experiments.
3. A method for optimizing the formulation of an alkaloid tobacco sprout inhibitor according to claim 2, characterized in that: The effect data specifically include bud inhibition rate, bud inhibition duration, degree of impact on tobacco growth and environmental residue.
4. The method for optimizing the formulation of an alkaloid tobacco sprout inhibitor according to claim 3, characterized in that: The sprout inhibition effect equation group includes a chemical balance equation, a metabolic process equation, a sprout inhibition activity equation, a duration equation, a growth influence equation, a residual degradation equation and a synergistic effect equation.
5. The method for optimizing the formulation of an alkaloid tobacco sprout inhibitor according to claim 4, characterized in that: The method for generating a large number of recipe data sets is specifically: according to a preset recipe component range and step length, a traversal method is used to obtain the large number of recipe data sets.
6. The method for optimizing the formulation of an alkaloid tobacco sprout inhibitor according to claim 5, characterized in that: The method for clustering the large number of candidate recipes is k-means clustering.
7. The method for optimizing the formulation of an alkaloid tobacco sprout inhibitor according to claim 6, characterized in that: The genetic algorithm uses multiple groups of recipes with the best actual effects in the recipe set to be tested as the initial population.
8. The method for optimizing the formulation of an alkaloid tobacco sprout inhibitor according to claim 7, characterized in that: The genetic algorithm uses normalized effect data as a fitness function.
9. The method for optimizing the formulation of an alkaloid tobacco sprout inhibitor according to claim 8, characterized in that: The step of using a machine learning algorithm to train the established sprout inhibition effect equation group using the obtained formula data of different formulas specifically involves using a machine learning algorithm to continuously iterate and optimize the parameters of the sprout inhibition effect equation group using the obtained formula data of different formulas.
10. The method for optimizing the formulation of an alkaloid tobacco sprout inhibitor according to claim 9, characterized in that: The machine learning algorithm uses the random forest algorithm.
Citation Information
Patent Citations
Artificial intelligence-based basic culture medium formula development method and system
CN113450882A
Multi-target Bayesian optimization formula recommendation method and system, medium and equipment
CN118349587A
Tobacco suckercide formula optimization method
CN118734716A
Cited By
Formula optimization method of plant polyphenol tobacco suckercide
CN119226673A
