A method for calculating the atomic chemical potential and a method for using the atomic chemical potential to predict reaction sites

Abstract

The present invention provides a method for calculating atomic chemical potential and a method for predicting reaction sites using atomic chemical potential, which relates to the technical field of data operation, and includes: obtaining molecular orbital information in a molecule; determining frontier molecular orbitals, HOMO and LUMO according to the molecular orbital information, and calculating the contributions of atoms in each frontier molecular orbital according to HOMO and LUMO; calculating the frontier occupied layer energy and frontier unoccupied layer energy of atoms according to the contributions of atoms in each frontier molecular orbital; calculating the atomic chemical potential based on the frontier occupied layer energy and frontier unoccupied layer energy of atoms. The present invention first proposes the description quantity of "atomic chemical potential" to describe the reactivity of reaction sites in various reactions, directly uses the atomic chemical potential to predict reaction sites, and the prediction process is fast and efficient, ensuring that the calculation time of reaction sites of molecules with a molecular weight < 1000 is < 1 s.

Description

Technical Field

[0001] The present invention relates to the technical field of data operation, and specifically relates to a method for calculating atomic chemical potential and a method for using atomic chemical potential to predict reaction sites. Background Art

[0002] Single-step reactions in organic chemistry are the basic components of organic synthesis routes. The reaction sites of a single step can directly affect chemists' judgments on multiple consecutive reactions, or even the entire route. For example: in the synthesis of polysubstituted aromatic hydrocarbons, it is necessary to determine which group substitutions follow electrophilic mechanisms and which follow nucleophilic mechanisms, which may lead to the rearrangement of the reaction sequence; in nucleophilic substitutions, multiple attack sites such as amines, amides, and hydrazines need to be considered, and possible competing sites need to be protected or weakened to avoid side reactions. Such problems are directly related to the success or failure of experiments. Therefore, chemists usually spend a great deal of effort on manual judgment and literature research to determine the best route design. Even experienced chemists find it difficult to consider all possible types of chemical reactions comprehensively.

[0003] The reactivity of a reaction site refers to the possibility of a chemical reaction occurring at each part (sub-structure, site) of a molecule under given reaction conditions. The meaningful value here is the relative value, that is, the relative activity order of each site. Theoretically, the reaction activity depends on the energy of the reaction transition state. A reliable calculation of the transition state energy often involves molecular dynamics (meta-dynamics) sampling of the molecular structure and energy calculation for each structure. Such operations usually require manual intervention and analysis, and the time consumption ranges from several hours to several months, far exceeding the acceptable feedback time for chemical synthesis personnel to analyze reactivity in industrial scenarios (usually, the design of a route takes dozens of minutes, including the judgment of reaction sites for dozens of molecules).

[0004] Currently, relatively fast methods for predicting reaction sites include:

[0005] 1. Nuclear magnetic resonance 13 The predicted value of the chemical shift of the 13C spectrum (hereinafter referred to as 13 13C spectrum). The electrophilic property of a molecule is characterized by the chemical shift, and then the most likely reaction sites in aromatic electrophilic substitution reactions (SEAr) are determined.

[0006] 2. Atomic partial charge. By using the atomic charge distribution calculated once, the positive and negative charges of each reaction site are compared to represent the activity order of the sites in a chemical reaction. The calculation methods of atomic charge include Mulliken, Hirshfeld, atom-in-molecule (AIM), etc.

[0007] 3. Fukui function. The difference in atomic charge distribution of a molecule under the conditions of +1 charge, neutral, and -1 charge is used to characterize the activity order of sites in aromatic electrophilic substitution and aromatic nucleophilic substitution (SNAr).

[0008] 4. Electrophilic superdelocalisability, nucleophilic superdelocalisability. The electron and empty orbital densities weighted by molecular orbital energy are used to characterize the activity order of sites in electrophilic substitution and nucleophilic substitution.

[0009] 5. Judging the activity order of aromatic electrophilic substitution sites by the energy of protonation reaction intermediates.

[0010] 6. Machine learning, for example, WLN uses Reaxys data to learn and judge the activity order of aromatic electrophilic substitution sites.

[0011] The existing technical solutions mentioned above have problems in accuracy, application scope, or time cost.

[0012] Nuclear magnetic resonance 13 The prediction of 13C chemical shift only depends on the surrounding chemical environment, so it is the fastest. However, the literature is limited to the discussion of aromatic electrophilic substitution reactions, and this type of reaction is only one of many reaction types, resulting in a limited application scope of this method. In terms of accuracy, 13 13C spectrum has serious errors in some simple aromatic electrophilic substitution problems. For example, for the electrophilic substitution sites of nitrobenzene, it will give the wrong conclusion of ortho > meta > para (the correct one is meta > para > ortho); for the more complex fused ring situation, 13 13C spectrum often needs to be combined with 1 1H spectrum for comprehensive judgment, and the mixed judgment logic adds uncertainty to site prediction.

[0013] The time cost of atomic charge, Fukui function, electrophilic / nucleophilic superdelocalizability, and RegioSQM(20) is relatively high. Among them, the strict Fukui function requires multiple calculations. The calculation of the Fukui function with molecular orbital energy weights and electrophilic / nucleophilic superdelocalizability requires the extraction of converged wave functions for analysis, which further increases the time cost. The optimized RegioSQM(20) also takes nearly 1 minute for small molecules. In terms of accuracy, the Fukui function is relatively accurate for the judgment of aromatic electrophilic / nucleophilic reactions, RegioSQM(20) is relatively accurate for the judgment of aromatic electrophilic reactions, and electrophilic superdelocalizability is relatively accurate for the judgment of molecular electrophilic reactions; the nucleophilic superdelocalizability is unstable for the judgment of molecular nucleophilic reactions, especially aromatic nucleophilic substitution. This is because there is a contribution to the energy of all empty orbitals in its parameters, and conventional ground-state density functional theory (DFT) methods are difficult to accurately describe high-energy empty orbitals (more expensive methods do not have the possibility of obtaining results in a short time); atomic charge, regardless of which charge partitioning method is used, is limited to the contribution of the overall charge and ignores the more important contribution of the frontier orbitals, resulting in errors in the judgment of all aromatic sites. In terms of generality, atomic charge and the two superdelocalizabilities are generally applicable to all types in principle (but do not guarantee accuracy), while the focus of the Fukui function on the frontier orbitals makes it unable to judge the reactivity of non-aromatic sites (some theoretical Fukui functions with orbital energy weights can solve the generality problem, but there will be the same judgment accuracy problem as the nucleophilic superdelocalizability). RegioSQM(20) and the machine learning method WLN are limited to aromatic electrophilic substitution. Summary of the Invention

[0014] In view of this, the present invention provides a method for calculating atomic chemical potential and a method for using atomic chemical potential to predict reaction sites. For the first time, the descriptive quantity "atomic chemical potential" is proposed to describe the reactivity of reaction sites in various reactions. This calculation method mainly targets the frontier orbitals, but not the frontier orbitals of the entire molecule in the usual sense, but the orbitals with the highest atomic orbital composition at the reaction sites, so as to comprehensively consider all types of sites including aromatic sites (ensuring reactivity).

[0015] The technical object of the present invention is achieved as follows:

[0016] On the one hand, the present invention provides a method for calculating atomic chemical potential, and the method includes:

[0017] Obtain molecular orbital information in a molecule;

[0018] Determine the frontier molecular orbitals, HOMO, and LUMO according to the molecular orbital information, and calculate the contribution of atoms in each frontier molecular orbital according to HOMO and LUMO;

[0019] Calculate the frontier occupied layer energy and frontier unoccupied layer energy of an atom according to the contribution of the atom in each frontier molecular orbital;

[0020] Calculate the atomic chemical potential based on the frontier occupied layer energy and frontier unoccupied layer energy of the atom.

[0021] On the basis of the above scheme, when calculating the atomic chemical potential, introduce a reaction parameter to balance the frontier occupied layer energy and frontier unoccupied layer energy of the atom, and the reaction parameter is related to the type of chemical reaction.

[0022] On the basis of the above scheme, the molecular orbital information includes the electronic structure of the molecule, the molecular orbital coefficients, and the molecular orbital energy, where the electronic structure of the molecule includes the electron occupancy of each molecular orbital.

[0023] On the basis of the above scheme, the contribution of the atom in each frontier molecular orbital refers to the number of electrons or empty orbitals belonging to the atom in each frontier molecular orbital.

[0024] On the basis of the above scheme, the frontier occupied layer energy of the atom refers to the average energy of the occupied layer of the atom in the frontier molecular orbital, and the frontier unoccupied layer energy of the atom refers to the average energy of the unoccupied layer of the atom in the frontier molecular orbital.

[0025] On the basis of the above scheme, the calculating the frontier occupied layer energy and frontier unoccupied layer energy of the atom according to the contribution of the atom in each frontier molecular orbital includes:

[0026] Set a predetermined occupancy threshold, and preset the starting frontier molecular orbital as j and the ending frontier molecular orbital as k;

[0027] Extract the number of electrons belonging to a single atom from the starting frontier molecular orbital j to the HOMO, add them in order of the molecular orbital energy from near to the HOMO, until the predetermined occupancy threshold is reached, and perform a weighted average of the energies of the added electrons with respect to the number of electrons to obtain the frontier occupied layer energy of the atom;

[0028] Extract the number of empty orbitals belonging to a single atom from the LUMO to the ending frontier molecular orbital k, add them in order of the molecular orbital energy from near to the LUMO, until the predetermined occupancy threshold is reached, and perform a weighted average of the energies of the added empty orbitals with respect to the number of empty orbitals to obtain the frontier unoccupied layer energy of the atom.

[0029] On the basis of the above scheme, the starting frontier molecular orbital j = max{1, HOMO - 9}, and the ending frontier molecular orbital k = min{LUMO + 9, n}, where n is the total number of molecular orbitals.

[0030] On the other hand, the present invention also provides a method for predicting reaction sites using atomic chemical potential, the method comprising:

[0031] Obtain the SMILES information of the molecule and analyze it to obtain the three-dimensional molecular structure;

[0032] Perform single-point energy calculation on the three-dimensional molecular structure and extract the molecular orbital information;

[0033] Combined with the molecular orbital information, calculate the atomic chemical potential using the calculation method described in any of the above;

[0034] Sort according to the atomic chemical potential from large to small, and take the atom corresponding to the maximum value as the best reaction site.

[0035] Based on the above solution, the three-dimensional molecular structure is one or more.

[0036] Based on the above solution, when there are multiple three-dimensional molecular structures, the method further comprises:

[0037] Use the UFF force field for structure optimization and select the three-dimensional molecular structure with the lowest energy for single-point energy calculation.

[0038] The method of the present invention has the following beneficial effects compared with the prior art:

[0039] (1) The present invention first proposes the descriptive quantity of "atomic chemical potential" to describe the reactivity of reaction sites in various reactions. In the calculation method of atomic chemical potential provided, the energy localized to the atom is used as a parameter to calculate the atomic chemical potential, rather than the energy-weighted charge, avoiding the huge uncertainty brought by high orbital energy;

[0040] (2) Based on the characteristic of measuring the ease of gaining and losing electrons at the Fermi level, the present invention introduces the energy localized to the atom as a parameter, such as the energy of the atomic frontier occupied layer and the energy of the atomic frontier unoccupied layer, to calculate the atomic chemical potential, and uses the atomic chemical potential to reflect the tendency of electrophilic and nucleophilic reactions to occur at each reaction site in the molecule;

[0041] (3) The calculation method provided by the present invention is aimed at the frontier orbitals, but not the frontier orbitals of the whole molecule in the general sense, but the orbitals with the highest atomic orbital composition at the reaction site, taking into account all types of sites including aromatic sites (ensuring reactivity);

[0042] (4) For the method of predicting reaction sites provided by the present invention, the atomic chemical potential is directly used to predict reaction sites, and the prediction process is fast and efficient, ensuring that the calculation time of reaction sites for molecules with a molecular weight < 1000 is < 1 s (CPU 4 cores 3.0 GHz). Description of the Drawings

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0044] Figure 1 It is a flowchart of the calculation method according to the embodiment of the present invention;

[0045] Figure 2 It is a flowchart of the prediction method according to the embodiment of the present invention. Detailed implementation manners

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0047] Term explanations:

[0048] Transition state: The (structural) state with the highest energy in the reaction path of a single-step elementary reaction in a chemical reaction.

[0049] Molecular dynamics: A method for computer simulation of the motion of molecular atomic systems.

[0050] Metadynamics: A method for enhanced sampling of molecular conformations in molecular dynamics.

[0051] Chemical shift: The phenomenon that atoms with nuclear spin produce absorption peaks at different positions in the nuclear magnetic resonance spectrum under different chemical environments.

[0052] Nucleophilic reaction: A reaction in which an electron-deficient molecule is attacked by an electron-rich molecule.

[0053] Electrophilic reaction: A reaction in which an electron-rich molecule attacks an electron-deficient molecule.

[0054] Aromatic electrophilic substitution reaction (SEAr): A substitution reaction in which an electron-rich aromatic site of a molecule is attacked. Aromatic nucleophilic substitution reaction (SNAr): A substitution reaction in which an electron-deficient aromatic site of a molecule is attacked. Protonation: The process by which a compound molecule absorbs a hydrogen cation.

[0055] Atomic charge: The projected charge distribution on atoms in a chemical system.

[0056] Molecular orbital: A function formed by the combination of atomic orbitals in a molecule.

[0057] Wave function: A function in quantum mechanics that describes a system (e.g., electrons in chemistry).

[0058] Density functional theory: A quantum chemical method for describing many-electron systems.

[0059] Ground state: The lowest energy state of a microscopic particle in quantum mechanics.

[0060] Frontier molecular orbital: The molecular orbital that has the greatest influence on chemical properties, usually including the highest occupied molecular orbital and the lowest unoccupied molecular orbital.

[0061] Highest occupied molecular orbital (HOMO): The orbital with the highest energy level that has electrons in a molecule. Lowest unoccupied molecular orbital (LUMO): The orbital with the lowest energy level that does not have electrons in a molecule.

[0062] SMILES (Simplified Molecular Input Line Entry Specification): A simplified molecular linear input specification, which is a specification for clearly describing the molecular structure using ASCII strings.

[0063] RDKit: An open-source cheminformatics software.

[0064] UFF (Universal Force Field): A general molecular force field model.

[0065] Overlap integral: The integral of two atomic orbital functions in three-dimensional space.

[0066] xTB (Extended Tight Binding): A semi-empirical method of quantum chemistry based on the tight binding theory.

[0067] GFN1-xTB: A method for implementing xTB calculations.

[0068] As Figure 1 shown, the present invention provides a method for calculating the atomic chemical potential, and the method includes:

[0069] Obtain the molecular orbital information in a molecule;

[0070] Determine the frontier molecular orbitals, HOMO, and LUMO according to the molecular orbital information, and calculate the contribution of atoms in each frontier molecular orbital according to HOMO and LUMO;

[0071] Calculate the frontier occupied layer energy and frontier unoccupied layer energy of atoms according to the contribution of atoms in each frontier molecular orbital;

[0072] The atomic chemical potential is calculated based on the energy of the frontier occupied layer and the energy of the frontier unoccupied layer of the atom.

[0073] Specifically, when calculating the atomic chemical potential, reaction parameters are introduced to balance the energy of the frontier occupied layer and the energy of the frontier unoccupied layer of the atom, and the reaction parameters are related to the type of chemical reaction. For example, the reaction parameters for nucleophilic reactions and electrophilic reactions will be different.

[0074] Specifically, a molecular data model can be established to calculate and extract molecular orbital information. The molecular orbital information includes the electronic structure of the molecule, molecular orbital coefficients, and molecular orbital energies. Among them, the electronic structure of the molecule includes the electron occupation numbers of each molecular orbital.

[0075] Specifically, the HOMO and LUMO can be determined according to the molecular orbital energies. HOMO refers to the highest occupied molecular orbital in the molecule. HOMO is the molecular orbital with the highest electron density in the molecule and is usually filled with occupied electrons. LUMO refers to the lowest unoccupied molecular orbital in the molecule. LUMO is the lowest energy level molecular orbital that is unoccupied in the molecule and is usually an empty orbital. And a set of molecular orbitals between HOMO and LUMO are the frontier molecular orbitals.

[0076] Specifically, after obtaining the frontier molecular orbitals, for each atom, calculate the contribution of a single atom to each frontier molecular orbital. This contribution refers to the number of electrons or empty orbitals belonging to the atom in each frontier molecular orbital. When it represents the number of electrons, its value is positive, and when it represents the number of empty orbitals, its value is negative.

[0077] Specifically, calculating the energy of the frontier occupied layer and the energy of the frontier unoccupied layer of the atom according to the contribution of the atom in each frontier molecular orbital includes:

[0078] Set a predetermined occupation number threshold, and preset the starting frontier molecular orbital as j and the ending frontier molecular orbital as k;

[0079] Extract the number of electrons belonging to a single atom from the starting frontier molecular orbital j to HOMO, and add them in order of the molecular orbital energy from near to far from HOMO until the predetermined occupation number threshold is reached. Weight-average the energy of the added electrons by the number of electrons to obtain the energy of the frontier occupied layer of the atom;

[0080] Extract the number of empty orbitals belonging to a single atom from LUMO to the ending frontier molecular orbital k, and add them in order of the molecular orbital energy from near to far from LUMO until the predetermined occupation number threshold is reached. Weight-average the energy of the added empty orbitals by the number of empty orbitals to obtain the energy of the frontier unoccupied layer of the atom.

[0081] Therefore, the energy of the frontier occupied shell of an atom can also be regarded as the average energy of the occupied shell of the atom in the frontier molecular orbital, while the energy of the frontier unoccupied shell of the atom is regarded as the average energy of the unoccupied shell of the atom in the frontier molecular orbital.

[0082] Specifically, when calculating the energy of the frontier occupied shell of an atom and the energy of the frontier unoccupied shell of the atom, the calculation process is constrained by j and k. This constraint is reflected in that the frontier molecular orbitals involved in the calculation are constrained by [j, k]. Since both j and k are near the frontier molecular orbitals, the present invention sets ranges for j and k. The starting frontier molecular orbital j = max{1, HOMO - 9}, and the ending frontier molecular orbital k = min{LUMO + 9, n}, where n is the total number of molecular orbitals.

[0083] Specifically, the present invention provides a specific embodiment to represent the process of the above-described calculation method. However, the formulas, parameters, etc. involved in this embodiment do not limit the protection scope of the present invention.

[0084] For a certain atom A in a molecule:

[0085] Define ε A，FOS and ε A，FUS as the energy of the frontier occupied shell (Frontier Occupied Shell) and the energy of the frontier unoccupied shell (Frontier Unoccupied Shell) of atom A, respectively. The calculation formula for the atomic chemical potential is as follows:

[0086] η A = γε A，FOS +(1 - γ)ε A，FUS (1);

[0087] In the formula, η A is the atomic chemical potential of atom A, and γ represents the reaction parameter. This reaction parameter is an empirical parameter related to the reaction type. In particular, in aromatic electrophilic substitution, γ = 0.8 - 1.0.

[0088] Formula (1) defines the atomic chemical potential as the energy between the frontier occupied layer and the frontier unoccupied layer of electrons in the atom.

[0089] And the calculation formulas for ε A，FOS and ε A，FUS are as follows:

[0090]

[0091] In the formula, ε i is the energy of the i-th frontier molecular orbital, and ρ A，iRefers to the contribution of atom A to the i-th frontier molecular orbital, x refers to the preset occupancy number threshold. In this embodiment, the effective range of x is set to 0 - 2, the starting frontier molecular orbital is i, and the ending frontier molecular orbital is k.

[0092] According to x, extract the number of electrons belonging to atom A in the starting frontier molecular orbital j to HOMO, and add them in the order of the molecular orbital energy from near to HOMO until the x value is reached. Then, perform a weighted average of the energies of the added electrons with respect to the number of electrons to obtain the frontier occupied layer energy of the atom. Extract the number of empty orbitals belonging to atom A in the LUMO to the ending frontier molecular orbital k, and add them in the order of the molecular orbital energy from near to LUMO until the x value is reached. Then, perform a weighted average of the energies of the added empty orbitals with respect to the number of empty orbitals to obtain the frontier unoccupied layer energy of the atom.

[0093] ρ A，i The calculation formula is as follows:

[0094]

[0095] In the formula, f i is the electron occupancy of the i-th frontier molecular orbital, c is the molecular orbital coefficient, S is the overlap integral matrix in the atomic orbital basis, and the subscripts μ, v represent atomic orbitals. ρ A，i defines the number of electrons or empty orbitals belonging to atom A in the frontier molecular orbital i.

[0096] Specifically, in this embodiment, in order to improve the calculation speed during calculation, matrix simplification operations can be used.

[0097] Specifically, ε A，FOS and ε A，FUS The calculation is constrained by i ∈ [j, k], so the calculation of ρ A，i and even the range of the input matrix can be constrained by i. Since both j and k are near the frontier molecular orbitals, in this patent, j = max{1, HOMO - 9} and k = min{LUMO + 9, n} (n is the total number of molecular orbitals), that is, at most 20 molecular orbitals are considered.

[0098] The present invention proposes a description quantity of atomic chemical potential and gives a specific calculation method. This atomic chemical potential is a potential (energy) defined on the atom, and this potential is based on the proportion of the atom in the frontier molecular orbitals. This definition ensures that the frontier orbitals most likely to undergo reactions are considered and reduces the dependence on the accuracy of molecular data operations as much as possible.

[0099] The atomic chemical potential can be directly used for predicting the reaction sites of the same type of chemical reactions. In the case of electrophilic reactions, the greater this value, the stronger the reactivity; in the case of nucleophilic reactions, the smaller this value, the stronger the reactivity. For example Figure 2As shown, based on the characteristics of atomic chemical potential, the present invention also provides a method for predicting reaction sites using atomic chemical potential, the method comprising:

[0100] Obtain the SMILES information of the molecule and analyze it to obtain the three-dimensional molecular structure;

[0101] Perform single-point energy calculation on the three-dimensional molecular structure and extract the molecular orbital information;

[0102] Combined with the molecular orbital information, calculate the atomic chemical potential using the calculation method described in any one of the above;

[0103] Sort according to the atomic chemical potential from large to small, and take the atom corresponding to the maximum value as the best reaction site.

[0104] Wherein, the three-dimensional molecular structure is one or more. When there are multiple three-dimensional molecular structures, the method further comprises:

[0105] Use the UFF force field for structure optimization and select the three-dimensional molecular structure with the lowest energy for single-point energy calculation.

[0106] Specifically, the present invention first obtains the SMILES information of the molecule, uses RDKit to generate one or more three-dimensional molecular structures from the SMILES, and uses the UFF force field to optimize the structure of the three-dimensional molecular structure, and selects the three-dimensional molecular structure with the lowest energy among them. Then perform single-point energy calculation on the selected three-dimensional molecular structure at the semi-empirical method GFN1-xTB level, and save the converged wave function or the corresponding numerical integral, and extract the molecular orbital coefficients, overlap integrals, molecular orbital energies, and molecular orbital electron occupancies therefrom. According to the extracted molecular orbital information, calculate the atomic chemical potential of each atom according to the above-described atomic chemical potential calculation method. To speed up the calculation, matrix simplification can be performed. Finally, use the atomic chemical potential as the basis for judging the reaction activity of each reaction site atom, arrange the atoms in descending order of their atomic chemical potential, and the atom with the maximum atomic chemical potential is the best reaction site.

[0107] Specifically, the present invention provides a specific embodiment to illustrate the above prediction method:

[0108]

[0109] The above formula is the structural formula of an aromatic compound, and the marked numbers in the formula are the numbers of each reaction site atom. For this compound, its potential aromatic electrophilic sites are: 4, 5, 7, 8, 10, 15, 16, 17, 18, 19.

[0110] The reaction site prediction process for this compound is as follows:

[0111] 1. First, input its SMILES: [CH3:1][O:2][C:3]1=[CH:4][CH:5]=[C:6]([CH:7]=[CH:8]1)[C:9]1=[CH:10][C:11](=[N:12][O:13]1)[C:14]1=[CH:15][CH:16]=[CH:17][CH:18]=[CH:19]1.

[0112] 2. Generate and optimize its three-dimensional molecular structure. Perform semi-empirical single-point energy calculations and extract the wave function for the constraints i ∈ [j, k] to obtain f i , c, S, ε i information: (c and S are matrices of 20×102 and 102×102 respectively, only partially shown)

[0113] f i =[22222222220000000000];

[0114] c = [[1.0000e+00 0.0000e+00 0.0000e+00...0.0000e+00 0.0000e+000.0000e+00]

[0115] [0.0000e+001.0000e+000.0000e+00...0.0000e+000.0000e+000.0000e+00]

[0116] [0.0000e+000.0000e+001.0000e+00...0.0000e+000.0000e+000.0000e+00] ...

[0118] [0.0000e+000.0000e+000.0000e+00...1.0000e+00-9.2614e-022.8119e-02]

[0119] [0.0000e+000.0000e+000.0000e+00...-9.2614e-021.0000e+008.5040e-10]

[0120] [0.0000e+00 0.0000e+00 0.0000e+00...2.8119e-02 8.5040e-101.0000e+00]];

[0121] S = [[-2.2911e-08, -1.7617e-03, -1.0107e-02,..., 2.6000e-08, -1.3862e-06, -1.0550e-07]]

[0122] [-1.6535e-02, 9.3436e-02, -1.7022e-02,..., 1.0670e-02, -2.2597e-01, -1.3703e-02]

[0123] [-1.9260e-02, 1.1450e-01, -2.0880e-02,..., -1.2099e-02, 1.7289e-01, 1.3317e-02] ...

[0125] [3.9265e-07, -1.6070e-03, -9.2193e-03,..., 1.1771e-08, 8.3252e-08, -4.4874e-08]

[0126] [6.0615e-03, -1.3256e-02, 2.7534e-03,..., -2.7194e-02, 1.4303e-02, -6.3465e-02]

[0127] [4.6337e-08, -5.7747e-04, -3.3132e-03,..., 1.9536e-07, -1.4155e-07, 1.5112e-07]];

[0128]

[0129] 3. Substitute the above data into Formula (3) and Formula (4) to calculate ε A，FOS as:

[0130] [-0.462, -0.428, -0.431, -0.438, -0.437, -0.427, -0.438, -0.437, -0.429, -0.42, -0.436, -0.409, -0.451, -0.426, -0.432, -0.435, -0.424, -0.436, -0.431, -0.446, -0.446, -0.471, -0.471, -0.471, -0.471, -0.471, -0.453, -0.468, -0.47, -0.469, -0.468, -0.469];

[0131] ε A，FUS as:

[0132] [-0.152, -0.225, -0.288, -0.275, -0.28, -0.285, -0.284, -0.273, -0.269, -0.276, -0.254, -0.293, -0.268, -0.282, -0.275, -0.267, -0.283, -0.268, -0.274, -0.154, -0.154, -0.152, -0.151, -0.126, -0.125, -0.141, -0.125, -0.125, -0.125, -0.125, -0.125, -0.125]。

[0133] 4. Select 4, 5, 7, 8, 10, 15, 16, 17, 18, 19, and calculate η according to formula (1). A (Take k = 0.9): [-0.421, -0.421, -0.423, -0.421, -0.405, -0.416, -0.418, -0.41, -0.419, -0.415].

[0134] 5. Sort η A in descending order, and find that the maximum value is -0.405, corresponding to reaction site 10, that is, reaction site 10 is the optimal reaction site.

[0135] The calculation method and prediction method provided by the present invention are equivalent or superior to known methods in terms of versatility. And its accuracy is superior to atomic charge and nuclear magnetic resonance 13 C spectrum, superior to nucleophilic hyperconjugation in nucleophilic reactions; approximate or superior to the Fukui function in judging aromatic sites, approximate to electrophilic hyperconjugation in electrophilic reactions; approximate to RegioSQM20 in aromatic electrophilic substitution. And its calculation speed and prediction speed are superior to atomic charge, nuclear magnetic resonance 13 C spectrum, Fukui function, RegioSQM20 and machine learning methods, and the calculation time of reaction sites of molecules with molecular weight < 1000 is < 1 s.

[0136] The above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for calculating atomic chemical potential, characterized in that: The method comprises: Get molecular orbital information in molecules; Determine the frontier molecular orbital and HOMO and LUMO based on the molecular orbital information, and calculate the contribution of atoms in each frontier molecular orbital based on HOMO and LUMO; Calculate the frontier occupied layer energy and frontier unoccupied layer energy of the atom according to the contribution of the atom in each frontier molecular orbital; The atomic chemical potential is calculated based on the frontier occupied layer energy and frontier unoccupied layer energy of the atom; The frontier occupied layer energy of an atom refers to the average energy of the layer occupied by the atom in the frontier molecular orbital, and the frontier unoccupied layer energy of an atom refers to the average energy of the layer unoccupied by the atom in the frontier molecular orbital. The step of calculating the frontier occupied layer energy and the frontier unoccupied layer energy of the atom according to the contribution of the atom in each frontier molecular orbital comprises: Set a predetermined occupation number threshold, and preset the starting frontier molecular orbital as j and the ending frontier molecular orbital as k; Extract the number of electrons belonging to a single atom from the starting frontier molecular orbital j to HOMO, add them from near to far according to the molecular orbital energy from HOMO until the predetermined occupation number threshold is reached, and perform weighted average of the energy of the added electrons on the number of electrons to obtain the frontier occupation layer energy of the atom; Extract the number of empty orbitals belonging to a single atom from LUMO to the end frontier molecular orbital k, add them from near to far according to the molecular orbital energy from LUMO until the predetermined occupation number threshold is reached, and perform weighted average of the energy of the added empty orbitals and the number of empty orbitals to obtain the frontier unoccupied layer energy of the atom; The starting frontier molecular orbital j=max{1,HOMO-9}, and the ending frontier molecular orbital k=min{LUMO+9,n}, where n is the total number of molecular orbitals.

2. A method for calculating atomic chemical potential according to claim 1, characterized in that: When calculating the atomic chemical potential, reaction parameters are introduced to balance the frontier occupied layer energy and the frontier unoccupied layer energy of the atoms, and the reaction parameters are related to the chemical reaction type.

3. A method for calculating atomic chemical potential according to claim 1, characterized in that: The molecular orbital information includes the electronic structure, molecular orbital coefficient and molecular orbital energy of the molecule, wherein the electronic structure of the molecule includes the number of electrons occupied in each molecular orbital.

4. A method for calculating atomic chemical potential according to claim 3, characterized in that: The contribution of the atom in each frontier molecular orbital refers to the number of electrons or empty orbitals belonging to the atom in each frontier molecular orbital.

5. A method for predicting reaction sites using atomic chemical potential, characterized in that: The method comprises: Obtain the SMILES information of the molecule and analyze it to obtain the three-dimensional molecular structure; Perform single-point energy calculations on three-dimensional molecular structures and extract molecular orbital information; In combination with the molecular orbital information, the atomic chemical potential is calculated using the calculation method described in any one of claims 1 to 4; Atoms are sorted from large to small according to their chemical potential, and the atom corresponding to the maximum value is taken as the best reaction site.

6. The method for predicting reaction sites using atomic chemical potential according to claim 5, characterized in that: The three-dimensional molecular structure is one or more.

7. The method for predicting reaction sites using atomic chemical potential according to claim 6, characterized in that: When the three-dimensional molecular structure is multiple, the method further comprises: The UFF force field was used for structural optimization, and the three-dimensional molecular structure with the lowest energy was selected for single-point energy calculation.

Images