A method for optimizing reaction transition states using genetic algorithms

Optimizing the reaction transition state through genetic algorithms, the problem of inaccurate initial guess structure and long calculation time is solved, and the rapid and accurate determination of reaction transition state is achieved, which reduces the calculation period and improves the accuracy of energy values. It is suitable for chemistry Industries guidance.

CN112270962BActive Publication Date: 2024-11-08JIANGNAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202011283854.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-11-17
Publication Date
2024-11-08
Estimated Expiration
2040-11-17

AI Technical Summary

Technical Problem

When the prior art is looking for transition states of chemical reactions, the initial guessing structure is inaccurate, time-consuming and the calculation results are incorrect, making the transition state difficult to determine, affecting the guidance of the chemical industry.

Method used

Genetic algorithms are used to optimize the reaction transition state, and the molecular formula of the input reactants is automatically converted into structural formulas. Genetic algorithms are used to search for possible transition states and reaction paths, reducing artificial regulation, and improving computing efficiency and accuracy.

Benefits of technology

It realizes the rapid and accurate determination of the reaction transition state, reduces the calculation cycle, avoids artificial input errors, and has a lower energy value for finding the transition state, which is in line with the real situation, saves the search cycle and supports path comparison analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112270962B_ABST
    Figure CN112270962B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for optimizing reaction transition state by using genetic algorithm, and belongs to the field of computational chemistry. The method of the present invention automatically converts the molecular formula of the input reactant into a structural formula, first sets the coordinates of the target atom as a variable, randomly generates multiple groups of initial values, first calculates the initial value that meets the conditions by a conventional algorithm, and then uses the genetic algorithm to automatically mutate and cross-combine the obtained initial value to obtain a new initial value, and repeatedly calculates to obtain the initial value that meets the conditions, until the calculation of the initial value that meets the conditions is completed, and then according to the reaction path corresponding to the initial value that meets the conditions, the desired transition state and reaction path are finally selected. The present invention does not need to adjust the molecular structure manually, the calculation process is fast and the result is more accurate.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a method for optimizing reaction transition states by utilizing a genetic algorithm, and belongs to the field of computational chemistry. Background Art

[0002] There is often a transition state in molecular conformational changes or chemical reactions. The transition state structure corresponds to the highest energy point in the reaction path on the potential energy surface, connecting the reactants and products through the minimum energy path. Determining the transition state helps to understand the reaction mechanism and calculate the energy barrier to infer the reaction rate.

[0003] However, in the process of chemical reaction, the transition state is unstable and has a very short existence time, so it cannot be obtained through experimental methods. Computational chemistry methods are currently the best way to find the transition state of the reaction, but there are still some difficulties in quantum chemical calculations. There are currently three main methods for finding the transition state of chemical reactions: TS, QST2 and QST3. Currently, the TS method is the commonly used method. When using the opt=ts method in Gauss to find the transition state, it is necessary to provide an initial guess structure. The initial guess structure has a great influence on the optimization results, and there are often the following problems: 1. The initial guess structure is wrong; 2. The initial guess structure is not the transition state of this reaction; 3. The energy of the initial guess structure is not the minimum energy of the transition state in this reaction path. The inaccuracy of the initial guess structure can easily lead to the failure of finding the transition state, and the incorrect initial guess often requires the initial guess structure to be changed again, which is time-consuming and may not necessarily obtain the optimal transition state.

[0004] Genetic algorithm is a computational model of biological evolution process that simulates the natural selection and genetic mechanism of Darwin's theory of biological evolution. It is a method to search for the optimal solution by simulating the natural evolution process. That is, "survival of the fittest, elimination of the unfit". The main process: starting from a population that represents the potential solution set of the problem, the population is composed of individuals, each of which is encoded by genes, so the individual is actually a chromosome with characteristics. Chromosomes are the carriers of material inheritance, and the collection of multiple genes determines the external performance. Therefore, at the beginning, it is necessary to map the required phenotype to the genotype controlled by the content, that is, encoding work. Since the gene-simulation coding is rather cumbersome, it is simplified to binary coding. After the first generation of population is generated, according to the principle of survival of the fittest, it evolves from generation to generation to produce better and better approximate solutions. In each generation, individuals are selected according to the fitness of individuals in the problem domain, and combined crossover and mutation are performed with the help of genetic operators of natural genetics to generate a population representing a new solution set. This process will lead to the population being more adapted to the environment than the previous generation, just like natural evolution. The best individuals in the last generation of population can be decoded and used as the approximate optimal solution to the problem. Therefore, finding a simpler and more accurate method for calculating the transition state of chemical reactions is of great significance for the determination of chemical reaction transition states and the guidance of subsequent chemical industries. Summary of the invention

[0005] In view of the problems that the initial guess structure in the process of existing chemical calculation of reaction transition state is difficult to determine, has strong randomness, and the process of finding the reaction transition state is time-consuming, the calculation results are often wrong or not ideal, and the calculation cycle is long, the present invention provides a method for optimizing the reaction transition state using a genetic algorithm. The method of the present invention automatically converts the molecular formula of the input reactant into a structural formula and then uses a genetic algorithm to find the possible transition state of the reactant, calculates the corresponding reaction path, and finally selects the desired transition state and reaction path. The present invention does not require manual adjustment of the molecular structure, the calculation process is fast and the results are more accurate.

[0006] Specifically, the technical solution of the present invention is a method for optimizing a reaction transition state using a genetic algorithm, the method comprising:

[0007] (1) Input the molecular formula of the reactant, convert the molecular formula into the structural formula through RDKit and output the corresponding .mol ​​file;

[0008] (2) Use Multiwfn to convert the .mol ​​file obtained in step (1) into a .gjf file;

[0009] (3) Set the coordinates of the target atoms in the .gjf file as variables, set the range of the coordinate variables, and randomly generate multiple sets of initial values ​​of the target atomic coordinates;

[0010] (4) Input the initial value of the target atomic coordinate variable, calculate using the basis set algorithm, calculate to obtain the .out file, and make a judgment: when the result ends abnormally, delete the result and substitute other initial values ​​for iterative calculation; when the result ends normally, but there is no imaginary frequency or there is more than one imaginary frequency in the frequency calculation result, delete the result and substitute other initial values ​​for iterative calculation; when the result ends normally, and there is only one imaginary frequency in the frequency calculation result, write the .out file through the file reading and writing method and calculate the reaction path corresponding to the corresponding transition state in the .out file, and then substitute other initial values ​​for iterative calculation;

[0011] (5) After the initial value iteration is completed, the initial values ​​that meet the conditions calculated in step (4) are sorted out, and new initial values ​​are obtained by automatic mutation and crossover combination using the genetic algorithm. Steps (4) and (5) are repeated until the initial values ​​that meet the conditions are calculated;

[0012] (6) After the iteration is completed, all calculated transition states and their corresponding reaction paths are sorted out, and the transition states that meet the conditions are judged and determined based on the reaction products.

[0013] In one embodiment of the present invention, in step (1), the molecular formula of the input reactant should be input according to the rules of RDKit, wherein the RDKit is an open source chemical informatics software.

[0014] In one embodiment of the present invention, Multiwfn is a quantum chemistry wave function analysis program.

[0015] In one embodiment of the present invention, in step (2), the selected basis set and precautions (such as solvation effects) need to be written into the .gjf file according to actual conditions, which is consistent with the input of the traditional calculation method.

[0016] In one embodiment of the present invention, the basis set algorithm in step (4) is: any algorithm that can be calculated on Gaussian software can be written into a file and call Gaussian software in this step; including any one of the semi-empirical method, ab initio algorithm, density functional method, and MPn method.

[0017] In one embodiment of the present invention, the range of the coordinate variable is preferably the maximum value or a larger range of the coordinate ranges of the remaining three atoms, that is, the maximum and minimum values ​​of the x, y, z coordinates of the remaining three atoms are respectively used as the maximum and minimum values ​​of the x, y, z coordinates of the coordinate variable, that is, the coordinate range of the target atom is obtained, or a coordinate range greater than the aforementioned coordinate range.

[0018] In one embodiment of the present invention, the initial value of the coordinate variable is a coordinate value of a target atom randomly generated within the range of the coordinate variable.

[0019] In one embodiment of the present invention, the multiple groups of initial values ​​described in step (3) can be selected according to actual conditions, such as 100 groups, 200 groups, 300 groups, etc.

[0020] In one embodiment of the present invention, the genetic algorithm in step (5) automatically mutates and crosses over to obtain new initial values, that is, by setting parameter values ​​of the crossover rate and the mutation rate, and then using the genetic algorithm to mutate and cross over the initial values ​​that meet the conditions to obtain new initial values.

[0021] In one embodiment of the present invention, the crossover rate is 0.6-1, preferably 0.8.

[0022] In one embodiment of the present invention, the mutation rate is less than 0.1, preferably 0.003.

[0023] In one embodiment of the present invention, in step (4), when the result is a normal end and there is only one virtual frequency, the energy value in the .out file obtained at this time is compared with the energy values ​​of all previously obtained .out files that meet the conditions. When the difference between the energy value and the previous energy value is within the range of ±0.00002Hatree (where 1Hartree = 2625.5 kJ mol -1 ), the energy value obtained this time will be deleted, and there is no need to calculate the corresponding reaction path at this time; the reason for deletion is that the difference between the two energy values ​​is too small, and the corresponding reaction paths are consistent.

[0024] In one embodiment of the present invention, the reaction pathway is calculated in a conventional manner.

[0025] In one implementation of the present invention, the number of iterations in the iteration process can be set according to actual needs.

[0026] The present invention also provides the application of the method for optimizing reaction transition states using genetic algorithms in computational chemistry.

[0027] Beneficial effects:

[0028] (1) The present invention directly inputs the molecular formula of the reactant according to the reaction path, without adjusting the pre-guessed structure, directly calls Gaussian software and directly judges whether the result meets the requirements of the transition state result according to the output file (there is only one virtual frequency), and directly calls Gaussian software according to the output result that meets the requirements to calculate the corresponding reaction path, and finally finds the transition state corresponding to the required reaction path from all the calculation results. This process does not require manual attention to the calculation results at all times, greatly saves the search cycle, and can compare and analyze the transition states of different paths.

[0029] (2) The method of the present invention can avoid the inaccuracy of manual input and the error in fine-tuning the structure during the optimization process, which leads to a lot of time spent or even no results in finding the optimized transition state; in addition, the transition state found by the method of the present invention may have a lower energy value than the transition state found by manual adjustment of the structure, and the result is more in line with the actual situation. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The results satisfying the conditions are obtained during the calculation process of Example 2, wherein (a) is the result of iterations 1 to 118; (b) is the result of iterations 119 to 237; (c) is the result of iterations 238 to 304; (d) is the result of iterations 305 to 337; (e) is the result of iterations 338 to 394; (f) is the result of iterations 395 to 410; (g) is the result of iterations 411 to 518; and (h) is the result of iterations 519 to 540.

[0031] Figure 2 This is the reaction pathway of the reaction between HCOOH and H2O2 finally confirmed in Example 2.

[0032] Figure 3 This is the reaction pathway of the reaction between HCOOH and H2O2 finally confirmed in Example 3. DETAILED DESCRIPTION

[0033] The present invention is further described below in conjunction with examples, but the embodiments of the present invention are not limited thereto.

[0034] Example 1

[0035] A method for optimizing a reaction transition state using a genetic algorithm, the method comprising:

[0036] (1) Input the molecular formula of the reactant, convert the molecular formula into the structural formula through RDKit and output the corresponding .mol ​​file;

[0037] (2) Use Multiwfn to convert the .mol ​​file obtained in step (1) into a .gjf file;

[0038] (3) Set the coordinates of the target atoms in the .gjf file as variables, set the range of the coordinate variables, and randomly generate multiple sets of initial values ​​of the target atomic coordinates;

[0039] (4) Input the initial value of the target atomic coordinate variable, calculate using the basis set algorithm, calculate to obtain the .out file, and make a judgment: when the result ends abnormally, delete the result and substitute other initial values ​​for iterative calculation; when the result ends normally, but there is no imaginary frequency or there is more than one imaginary frequency in the frequency calculation result, delete the result and substitute other initial values ​​for iterative calculation; when the result ends normally, and there is only one imaginary frequency in the frequency calculation result, write the .out file through the file reading and writing method and calculate the reaction path corresponding to the corresponding transition state in the .out file, and then substitute other initial values ​​for iterative calculation;

[0040] (5) After the initial value iteration is completed, the initial values ​​that meet the conditions calculated in step (4) are sorted out, and new initial values ​​are obtained by automatic mutation and crossover combination using the genetic algorithm. Steps (4) and (5) are repeated until the initial values ​​that meet the conditions are calculated;

[0041] (6) After the iteration is completed, all calculated transition states and their corresponding reaction paths are sorted out, and the transition states that meet the conditions are judged and determined based on the reaction products.

[0042] Furthermore, in step (2), the selected basis set and precautions (such as solvation effects) need to be written into the .gjf file according to the actual situation, which is consistent with the input of the traditional calculation method.

[0043] Furthermore, the basis set algorithm in step (4) includes any one of a semi-empirical method, an ab initio algorithm, a density functional method, and an MPn method.

[0044] Furthermore, the range of the coordinate variable is preferably the maximum value or a larger range of the coordinate ranges of the remaining three atoms, that is, the maximum and minimum values ​​of the x, y, z coordinates of the remaining three atoms are respectively used as the maximum and minimum values ​​of the x, y, z coordinates of the coordinate variable, that is, the coordinate range of the target atom is obtained, or a coordinate range larger than the aforementioned coordinate range.

[0045] Furthermore, the genetic algorithm in step (5) automatically mutates and crosses over to obtain new initial values, that is, by setting the parameter values ​​of the crossover rate and the mutation rate, and then using the genetic algorithm to mutate and crossover the initial values ​​that meet the conditions to obtain new initial values; the crossover rate is 0.6 to 1, preferably 0.8; the mutation rate is less than 0.1, preferably 0.003.

[0046] Furthermore, in step (4), when the result is a normal end and there is only one virtual frequency, the energy value in the .out file obtained at this time is compared with the energy values ​​of all previously obtained .out files that meet the conditions. When the difference between the energy value and the previous energy value is within the range of ±0.00002 Hartree (where 1 Hartree = 2625.5 kJ mol -1 ), the energy value obtained this time is deleted, and there is no need to calculate the corresponding reaction path at this time. Specifically, the reason for deletion is that the difference between the energy values ​​of the two is too small, and the corresponding reaction paths are consistent.

[0047] Example 2

[0048] Taking the reaction of HCOOH and H2O2 in vacuum at room temperature as an example, the genetic algorithm of the present invention is used to optimize the transition state in the reaction process of the two.

[0049] (1) Input the molecular formula of the reactant O[CH+](=O)[OH+]O, RDKit converts the molecular formula into a structural formula and outputs the corresponding .mol ​​file;

[0050] (2) Use Multiwfn to convert the .mol ​​file obtained in step (1) into a .gjf file, convert the coordinates of each atom in the .gjf file into Cartesian coordinates, select the calculation method and basis set as b3lyp / 6-31g(d), and write it into the .gjf file;

[0051] (3) Set the coordinates of the target atom in the .gjf file as variables. At the same time, according to the coordinates of the other three atoms C[0.46740000, 0.12730000, 0.36070000], O[0.08460000, -0.31430000, 1.47030000], O[-0.58080000, 0.21880000, -0.51860000], set the range of the target atom H coordinate variable to x[-0.6, 0.6], y[-0.4, 0.3], z[-0.6, 1.5], and randomly generate 100 sets of initial values;

[0052] (4) Input the initial values ​​of the coordinate variables and use the basis set algorithm to calculate, and obtain the .out file for judgment; when the result is abnormal, delete the result and substitute other initial values ​​for iterative calculation; when the result is normal, but there is no imaginary frequency or more than one imaginary frequency in the frequency calculation result, delete the result and substitute other initial values ​​for iterative calculation; when the result is normal, and there is only one imaginary frequency in the frequency calculation result, write the .out file through the file reading and writing method and calculate the reaction path (irc reaction path) corresponding to the corresponding transition state in the .out file, and then substitute other initial values ​​for iterative calculation;

[0053] (5) After the initial value iteration of step (4) is completed, the initial values ​​that meet the conditions calculated in step (4) are sorted, and the new initial values ​​are obtained by automatic mutation and crossover combination using the genetic algorithm. The crossover rate is 0.8, and the mutation rate of each site is 0.003. Repeat steps (4) and (5). When the result is a normal end and there is only one virtual frequency, the energy value in the .out file obtained at this time is compared with the energy values ​​of all the previously obtained .out files that meet the conditions. When the difference between it and the previous energy value is within the range of ±0.00002Hatree (where 1Hartree = 2625.5kJ mol -1 ), the energy value obtained this time is deleted, and there is no need to calculate the corresponding reaction path at this time.

[0054] (6) After the iteration is completed, all calculated transition states and their corresponding reaction paths are sorted out, and the transition states that meet the conditions are judged and determined based on the reaction products.

[0055] Finally, a total of 31 initial values ​​that meet the conditions were calculated, and the results are shown in Figure 1 (ah) and the energy values ​​are shown in Table 1.

[0056] Table 1 Example 2 Calculated initial values ​​that meet the conditions and the energy values ​​of the corresponding transition states

[0057]

[0058]

[0059] It can be seen that a total of 540 iterations were performed here, 31 possible solutions were screened out, and it took a total of 3 days (this calculation was run on a virtual machine on an ordinary computer. If it was run on a cluster computer, the calculation time would be greatly shortened). Combined with the screening of the reaction path of the subsequent transition state, it can be seen that the result obtained in the 237th iteration is the final optimized transition state, and the corresponding initial value coordinates are x:0.33630679, y:-0.12623329, z:-0.27125001, the Gibbs free energy is -341.200042Hartree, and the corresponding reaction path is shown in Figure 2 That is, one hydrogen on hydrogen peroxide is connected to the oxygen on the carbonyl group in formic acid, and the oxygen in hydrogen peroxide that loses hydrogen is connected to the carbon on formic acid.

[0060] The corresponding optimized transition state information is shown in Table 2.

[0061] Table 2 Four energy values ​​corresponding to the optimal transition state calculated in Example 2

[0062] energy Energy value (Hartree) The sum of electron energy and zero-point energy -314.170799 The sum of electron energy and thermal energy -341.164919 The sum of electronic enthalpy and thermal enthalpy -341.163975 Gibbs free energy -341.200042

[0063] Comparative Example 1

[0064] When the traditional calculation method is used to calculate the transition state of the reaction between HCOOH and H2O2, the method is as follows:

[0065] First, input the structural formula of HCOOH and H2O2 in Gaussian software, adjust the relative position of the reactants (especially the relative position of the reaction site), save it as a .gjf file, and then call Gaussian software for calculation. Check the output results and the following situations will appear: 1. Failure to converge normally; 2. Convergence but not a virtual frequency, not the required transition state; 3. Convergence and only one virtual frequency. If it does not converge or there is more than one virtual frequency, manually adjust the bond length, bond angle and dihedral angle of the reactive group and then call Gaussian for calculation. If it converges and there is only one virtual frequency, the irc reaction path calculation can be performed, and then determine whether the transition state found is the transition state connecting the reactant and the product.

[0066] After several adjustments to the structure, which took two weeks, the transition state structure was finally obtained. The reaction path corresponding to the transition state is shown in Figure 3 The Gibbs free energy corresponding to the transition state is -341.199807 Hartree. The results are shown in Table 3.

[0067] Table 3 Four energy values ​​corresponding to the optimal transition state calculated in Example 3

[0068]

[0069]

[0070] from Figure 2 and Figure 3 It can be seen that the reaction paths of the two are the same, that is, a hydrogen on hydrogen peroxide is attracted by the oxygen on the carbonyl group in formic acid, and the oxygen on hydrogen peroxide that loses hydrogen ions is connected to the carbon on the carbonyl group. However, the lowest Gibbs free energy value corresponding to this path automatically found by the genetic algorithm is -341.200042 Hartree, while the lowest Gibbs free energy value calculated by multiple manual adjustments is -341.199807 Hartree. The energy value of the transition state calculated by the method of the present invention is lower than that of comparative example 1, which is more in line with the actual situation. In addition, the method of the present invention does not require manual adjustment of the molecular structure, is time-consuming, and has obvious advantages.

[0071] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Anyone familiar with this technology can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be based on the definition of the claims.

Claims

1. A method for optimizing a reaction transition state using a genetic algorithm, characterized in that: The method comprises: (1) Input the molecular formula of the reactant, convert the molecular formula into the structural formula through RDKit and output the corresponding .mol ​​file; (2) Use Multiwfn to convert the .mol ​​file obtained in step (1) into a .gjf file; (3) Set the coordinates of the target atoms in the .gjf file as variables, set the range of the coordinate variables, and randomly generate multiple sets of initial values ​​of the target atomic coordinates; (4) Input the initial value of the target atomic coordinate variable, calculate using the basis set algorithm, calculate to obtain the .out file, and make a judgment: when the result ends abnormally, delete the result and substitute other initial values ​​for iterative calculation; when the result ends normally, but there is no imaginary frequency or more than one imaginary frequency in the frequency calculation result, delete the result and substitute other initial values ​​for iterative calculation; when the result ends normally and there is only one imaginary frequency in the frequency calculation result, write the .out file through the file reading and writing method and calculate the reaction path corresponding to the corresponding transition state in the .out file, and then substitute other initial values ​​for iterative calculation; (5) After the initial value iteration is completed, the initial values ​​that meet the conditions calculated in step (4) are sorted out, and new initial values ​​are obtained by automatic mutation and crossover combination using the genetic algorithm. Steps (4) and (5) are repeated until the initial values ​​that meet the conditions are calculated; (6) After the iteration is completed, all calculated transition states and their corresponding reaction paths are sorted out, and the transition states that meet the conditions are judged and determined based on the reaction products.

2. A method for optimizing reaction transition states using genetic algorithms according to claim 1, characterized in that: The basis set algorithm in step (4) includes any one of a semi-empirical method, an ab initio algorithm, a density functional method, and an MPn method.

3. A method for optimizing reaction transition states using genetic algorithms according to claim 1 or 2, characterized in that: The basis set algorithm described in step (4) is a density functional method.

4. The method for optimizing reaction transition state using genetic algorithm according to claim 1, characterized in that: The genetic algorithm in step (5) automatically mutates and crosses over to obtain new initial values, that is, by setting the parameter values ​​of the crossover rate and the mutation rate, and then using the genetic algorithm to mutate and cross over the initial values ​​that meet the conditions, a new initial value is obtained.

5. A method for optimizing reaction transition states using genetic algorithms according to claim 4, characterized in that: The crossover rate is 0.6-1.

6. According to the method for optimizing reaction transition states using genetic algorithm in claim 5, the crossover rate is 0.

8.

7. The method for optimizing reaction transition state using genetic algorithm according to claim 4, characterized in that: The mutation rate is less than 0.

1.

8. The method for optimizing reaction transition state using genetic algorithm according to claim 7, characterized in that: The mutation rate is 0.

003.

9. The method for optimizing reaction transition state using genetic algorithm according to claim 1, characterized in that: In step (4), when the conditions are met, the energy value in the .out file is compared with the energy value in the previous .out file. When the difference between the energy value and the previous energy value is within the range of ±0.00002Hatree, the energy value obtained this time is deleted.

10. Application of the method for optimizing reaction transition states using genetic algorithm as claimed in any one of claims 1 to 9 in the field of computational chemistry.