Ligand Optimization Method, Terminal Device, and Storage Medium

Abstract

The embodiments of the present application are applicable to the field of drug design technology, and provide a ligand optimization method, a terminal device, and a storage medium. The method includes: generating an original ligand group composed of multiple original ligands, where each original ligand is configured with information in multiple dimensions; performing differential processing on the information in each dimension of each original ligand to obtain differential information in each dimension; respectively generating a new ligand corresponding to each original ligand according to the differential information in all dimensions of each original ligand; respectively calculating the ligand energy of each original ligand and respectively calculating the ligand energy of the corresponding new ligand; updating the original ligand group according to the ligand energy of each new ligand and the ligand energy of each original ligand to obtain a target ligand group; the total ligand energy of the target ligand group is lower than the total ligand energy of the original ligand group. By using the above method, preferred ligands can be comprehensively searched out.

Description

Technical Field

[0001] This application belongs to the technical field of drug design, and particularly relates to a ligand optimization method, a terminal device, and a storage medium. Background Art

[0002] Molecular docking is one of the important modules in drug design. It mainly predicts the binding mode of a receptor and a ligand by analyzing the property characteristics and interactions of the receptor and the ligand, so as to predict whether the receptor and the ligand after binding have the potential to become candidate drugs. Generally, the lower the energy between the receptor and the ligand after binding, the more stable the binding. Therefore, before binding a ligand to a receptor, a stable ligand also needs to be determined.

[0003] Generally, a ligand is composed of multiple atoms, and the stability of the ligand is usually determined by multiple dimensions of information such as the relative positions of the atoms, the overall rotation of the ligand, and the rotation angles of the rotatable bonds (the connecting bonds between atoms) in the ligand. In the prior art, when designing a stable ligand, Monte Carlo simulated annealing or quasi-Newton methods are usually used for searching to search for ligands composed of different relative positions, the overall rotation of the ligand, and rotation angles, and calculate the energy of each ligand to obtain a preferred ligand with low energy.

[0004] However, one variable usually has a strong correlation with other variables, and there are strong interactions between variables. However, when using the above methods, each time only one dimension of information can be mutated for searching or when multiple dimensions of information are mutated, the multiple dimensions of information are independent and cannot be correlated with each other, so that it is impossible to comprehensively search for preferred ligands. Summary of the Invention

[0005] The embodiments of this application provide a ligand optimization method, device, terminal device, and storage medium, which can solve the problem of being unable to comprehensively search for preferred ligands.

[0006] In a first aspect, the embodiments of this application provide a ligand optimization method, which includes:

[0007] Generate an original ligand group; the original ligand group includes multiple original ligands, and each original ligand is configured with multiple dimensions of information;

[0008] Perform differential processing on the information of each dimension of each original ligand to obtain the differential information of each dimension of each original ligand;

[0009] Generate a new ligand corresponding to each original ligand respectively according to the differential information of all dimensions of each original ligand;

[0010] Calculate the ligand energy of each original ligand according to the information of multiple dimensions of each original ligand, and calculate the ligand energy of the corresponding new ligand according to the differential information of multiple dimensions of each original ligand respectively;

[0011] Update the original ligand group according to the ligand energy of each new ligand and the ligand energy of each original ligand to obtain a target ligand group; the total ligand energy of the target ligand group is lower than the total ligand energy of the original ligand group.

[0012] In a second aspect, an embodiment of the present application provides a ligand optimization device, which includes:

[0013] An original ligand group generation module, configured to generate an original ligand group; the original ligand group includes multiple original ligands, and each original ligand is configured with information of multiple dimensions;

[0014] A difference module, configured to perform difference processing on the information of each dimension of each original ligand to obtain the difference information of each dimension of each original ligand;

[0015] A new ligand generation module, configured to generate a new ligand corresponding to each original ligand respectively according to the difference information of all dimensions of each original ligand;

[0016] A ligand energy calculation module, configured to calculate the ligand energy of each original ligand according to the information of multiple dimensions of each original ligand, and calculate the ligand energy of the corresponding new ligand according to the difference information of multiple dimensions of each original ligand respectively;

[0017] An update module, configured to update the original ligand group according to the ligand energy of each new ligand and the ligand energy of each original ligand to obtain a target ligand group; the total ligand energy of the target ligand group is lower than the total ligand energy of the original ligand group.

[0018] In a third aspect, an embodiment of the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method described in the first aspect above is implemented.

[0019] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in the first aspect above is implemented.

[0020] In a fifth aspect, an embodiment of the present application provides a computer program product. When the computer program product runs on a terminal device, the terminal device is enabled to execute the method described in the first aspect above.

[0021] The beneficial effects of the embodiments of the present application compared with the prior art are as follows: After generating the original ligand composed of multiple dimension information, differential processing is respectively performed on each dimension information to obtain the differential information of each dimension, and a new ligand is formed based on the differential information. In this way, the dimension information between the generated new ligands can have multiple mutations simultaneously with the dimension information of the original ligand, enhancing the search ability for ligands. Moreover, the differential information of multiple dimensions can reflect the variable relationship between the information of multiple dimensions. Therefore, the new ligands obtained based on the differential information of multiple dimensions can maintain the variable relationship between the information of each dimension in the new ligands, enabling the terminal device to generate new ligands more comprehensively. Furthermore, the terminal device can further achieve a comprehensive search for ligands. Finally, the original ligand group is updated according to the ligand energy of each new ligand to obtain a target ligand group with a lower total ligand energy, such that each ligand in the target ligand group is an optimal ligand with better stability under comprehensive search. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0023] Figure 1 is a flowchart of the implementation of a ligand optimization method provided by an embodiment of the present application;

[0024] Figure 2 is a schematic structural diagram of the original ligand in an embodiment of the present application;

[0025] Figure 3 is a schematic diagram of an implementation manner for calculating differential information in a ligand optimization method provided by an embodiment of the present application;

[0026] Figure 4 is a schematic diagram of an implementation manner for generating a new ligand in a ligand optimization method provided by an embodiment of the present application;

[0027] Figure 5 is a schematic diagram of an implementation manner for generating a target ligand group in a ligand optimization method provided by an embodiment of the present application;

[0028] Figure 6 is an effect diagram of the docking of ligands with receptors respectively under different docking software in an embodiment of the present application;

[0029] Figure 7 is an effect diagram of the docking of ligands with receptors respectively under different docking software in another embodiment of the present application;

[0030] Figure 8 This is an effect diagram of the docking of ligands with receptors under different docking software in another embodiment of the present application;

[0031] Figure 9 This is a schematic structural diagram of a ligand optimization device provided by an embodiment of the present application;

[0032] Figure 10 This is a schematic structural diagram of a terminal device provided by an embodiment of the present application. Detailed implementation manners

[0033] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0034] It should be understood that when used in the specification of the present application and the appended claims, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0035] In addition, in the description of the specification of the present application and the appended claims, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0036] The ligand optimization method provided by the embodiments of the present application belongs to the field of drug design. In the field of drug design, molecular docking is usually used for design. Among them, the software for performing molecular docking includes, but is not limited to: GOLD (a molecular docking software program for calculating the binding mode of macromolecules and small molecules), rDock (a software for simulating molecular docking), AutodockVina (an open-source molecular simulation software), and other software. Among them, AutoDock Vina is the most widely used open-source framework. However, the problem with the Vina framework is that the accuracy of molecular docking is not high, and the search ability of the search algorithm is limited. Based on this, the embodiments of the present application have improved AutodockVina, providing a new molecular docking software koto and a new ligand optimization method.

[0037] Among them, the scoring function used in the koto software is the same as that in the Autodock Vina software. The specific difference lies in the ligand optimization method provided by this embodiment running in the koto software.

[0038] In this embodiment, the above ligand optimization method can be applied to terminal devices such as tablet computers, notebook computers, ultra-mobile personal computers (UMPCs), and netbooks installed with Koto software. The specific type of the terminal device is not limited in the embodiments of the present application.

[0039] Please refer to Figure 1 , Figure 1 which shows a flowchart of the implementation of a ligand optimization method provided by an embodiment of the present application. The method includes the following steps:

[0040] S101. The terminal device generates an original ligand group; the original ligand group includes a plurality of original ligands, and each original ligand is configured with information in multiple dimensions.

[0041] In one embodiment, the above original ligand group includes a plurality of original ligands. In this embodiment, the group size of the original ligand group is not limited.

[0042] In one embodiment, each original ligand is composed of a plurality of atoms. Among them, the information in multiple dimensions includes but is not limited to: the position of each atom in the original ligand, the relative position between atoms, the overall rotation of the ligand, and the rotation angle of the rotatable bond (the connection bond between atoms) in the ligand. There is no limitation on this. Among them, when generating the original ligand, the information in each dimension configured for the original ligand is randomly generated.

[0043] Among them, it should be particularly noted that for the information in the dimension of the overall rotation of the ligand, in the prior art, it is usually represented by a quaternion (a, b, c, d). However, these four numbers are not actually independent and need to satisfy the mathematical relationship of a 2 +b 2 +c 2 +d 2 = 1. It can be seen that they have only 3 degrees of freedom. If quaternions are used for group optimization search, the search performance will be reduced. Therefore, it is necessary to be able to separate appropriate variables to represent the overall rotation of the ligand.

[0044] Based on this, in this embodiment, the following specific method can be used to transform the above quaternion to improve the search performance of the terminal device:

[0045] Refer to Figure 2 , Figure 2 which is a schematic structural diagram of an original ligand in an embodiment of the present application; where x, y, and z respectively represent a pre-established space coordinate system. Then, three angles are defined to be mutually converted with the quaternion to represent the overall rotation of the ligand.

[0046] Specifically, the formula is as follows: The quaternion q is defined as:

[0047]

[0048] Meanwhile, it needs to satisfy:

[0049] a 2 +b 2 +c 2 +d 2 = 1; (2)

[0050] where a, b, c, and d are all real numbers, being the unit vectors of the three-dimensional coordinate axes.

[0051] Another form of its geometric definition is:

[0052]

[0053] Each of its components corresponds to a, b, c, and d in the original definition respectively. It represents that the overall rotation of the ligand is along the axis, rotating by an angle θ.

[0054] Therefore, the terminal device only needs to define the axis of rotation and the angle of rotation, and it can use other definitions to convert to quaternions, thereby separating the degrees of freedom.

[0055] Specifically, the above definition can be that θ represents the angle of rotation. And the axis of rotation can be represented by two angles α and β. For details, refer to Figure 2 .

[0056] Therefore, the conversion relationship can be as follows:

[0057] v x = sinα * cosβ; (4)

[0058] v y = sinα * sinβ; (5)

[0059] v z = cosα (6)

[0060] So far, the terminal device can use (θ, α, β) to convert with the quaternion (a, b, c, d) mutually. Among them, θ represents the angle of rotation, α represents the angle between the axis of rotation and the z-axis, and β represents the angle between the projection of the axis of rotation on the xy plane and the x-axis.

[0061] S102. The terminal device performs differential processing on the information of each dimension of each original ligand to obtain the differential information of each dimension of each original ligand.

[0062] In one embodiment, the above differential processing is used to adjust the information of the dimension, and generate more optimal dimension information as differential information.

[0063] In a specific embodiment, with reference to Figure 3 , the terminal device can perform differential processing on the information of each dimension through the following steps S301 - S304 to obtain the differential information of each dimension, which is described in detail as follows:

[0064] S301. The terminal device selects N target original ligands from the original ligand group according to the ligand energy of each original ligand, and determines the optimal original ligand from the target original ligands; the ligand energy of the N target original ligands is lower than that of any other original ligand in the original ligand group except the target original ligands.

[0065] In one embodiment, the ligand energy of the above original ligand is the energy calculated according to the information of multiple dimensions of the original ligand. Specifically, when calculating the ligand energy of the original ligand, the information of dimensions such as the position of the root atom (the first atom) in the original ligand, the relative position between atoms, the overall rotation of the original ligand, and the rotation angle of the rotatable bond in the ligand needs to be used. Among them, calculating the ligand energy according to the above dimension information is a prior art, and no detailed description will be given here. Among them, the position of each remaining atom can be determined according to the information of dimensions such as the position of the root atom, the relative position between atoms, the overall rotation of the original ligand, and the rotation angle of the rotatable bond in the ligand.

[0066] It should be noted that for each original ligand, a lower ligand energy of the original ligand indicates that the original ligand is more stable. Therefore, the terminal device can sort each original ligand from low to high according to the ligand energy of each original ligand. Then, the first N original ligands are determined as the target original ligands. That is, the ligand energy of the N target original ligands is lower than that of any other original ligand in the original ligand group except the target original ligands. Among them, the above N is a value set according to actual needs, and the specific value of N is not limited in this embodiment.

[0067] In another embodiment, the above N value can be specifically determined in the following manner: count the group size of the original ligand group, and then randomly generate a target probability. Multiply the group size by the target probability to determine the N value. Exemplarily, if the group size of the original ligand group is 50 and the target probability is 0.1, then the above N value is 50 * 0.1 = 5. Then, according to the level of each ligand energy, the 5 original ligands with lower ligand energy are determined as the target original ligands. That is, the number of target original ligands is not fixed in each iteration process.

[0068] It should be added that the optimal original ligand is any one of the N target original ligands, that is, the optimal original ligand is a target original ligand randomly selected from the N target original ligands.

[0069] S302. The terminal device randomly selects a target number of random original ligands from the original ligand group.

[0070] In one embodiment, the above target number can also be a value set by the terminal device according to actual needs, and this is not limited. Exemplarily, the above target number can specifically be 2.

[0071] After that, for any original ligand in the original ligand group, the terminal device can process it through the following steps:

[0072] S303. The terminal device obtains the difference factor of the original ligand.

[0073] In one embodiment, the above difference factor can specifically be randomly generated according to the probability distribution of the target parameter. Among them, the target parameter can be set by the terminal device according to actual needs, or the target parameter in the current iteration process can be updated according to the ligand energy of the new ligand and the ligand energy of the original ligand in the previous iteration process, and this is not limited.

[0074] Specifically, the terminal device obtains the number of iterations in which the original ligand group is updated; if the number of iterations is less than or equal to the first preset number of times, a set of initial parameters is randomly selected as the target parameter in the current iteration process; if the number of iterations is greater than the first preset number of times, the ligand energy of the new ligand and the ligand energy of the original ligand in the previous iteration process are obtained to update the target parameter in the previous iteration process, and the target parameter in the current iteration process is obtained.

[0075] In one embodiment, the above first preset number of times can be a number set by the terminal device according to actual needs. However, in this embodiment, in order to obtain a preferred difference factor, the above first preset number of times can be set to 1. That is, in the first iteration, the terminal device will pre-set multiple sets of initial parameters. After that, a set of initial parameters is randomly selected as the target parameter, and the difference factor of the original ligand is generated according to the probability distribution of the target parameter. After that, the target parameter in each iteration process is calculated according to the ligand energy of the new ligand and the ligand energy of the original ligand in the previous iteration process. And, the difference factor in each iteration process is generated according to the probability distribution of the target parameter in the current iteration process.

[0076] It can be understood that when searching for a ligand with better stability, it is usually necessary to perform multiple iterations on the original ligand group. Therefore, after each iteration, the terminal device can calculate and record the ligand energy of the new ligand and the ligand energy of the original ligand in the previous iteration process.

[0077] It should be noted that there may be multiple ligand energies of the new ligand corresponding to those of the original ligand in the previous iteration process. However, the terminal device does not calculate based on the ligand energy of each new ligand and that of the original ligand. Instead, when the ligand energy of the new ligand is lower than that of the original ligand, the ligand energy of the new ligand and that of the original ligand are involved in the calculation. Specifically, the calculation formulas for the above differential factor are as follows in formulas (7) and (8):

[0078]

[0079]

[0080] Among them, the above f(U i,g ) is the ligand energy of the new ligand; f(X i,g ) is the ligand energy of the original ligand; g is the current iteration number; i represents the currently calculated dimension; K starts from 0 at the beginning of the iteration and increments by 1 for each iteration. In the calculation of w k , |f(U i,g ) - f(X i,g )| represents how much lower (taking the absolute value) the energy of the new ligand U i,g with better differential evolution is than that of the original ligand in the g-th generation of the iteration. The denominator is the sum of the absolute values of how much lower the energies of all new ligands with better differential evolution are than that of the original ligand. S F stores the new ligands with successful differential evolution. Successful differential evolution means that the ligand energy of the new ligand is lower than that of the original ligand.

[0081] S304. The terminal device imports the differential factor, the information of each dimension of the optimal original ligand, and the information of each dimension of the random original ligand into the differential calculation formula to obtain the differential information of each dimension of each original ligand.

[0082] In one embodiment, the above differential calculation formula is as follows:

[0083]

[0084] Among them, i and g are described above and will not be elaborated here. V i,g represents the differential information of the i-th dimension of the original ligand in the g-th iteration; X i,g represents the information of the i-th dimension of the original ligand in the g-th iteration; F i represents the differential factor of the information of the i-th dimension; X pbest,, represents the information of the i-th dimension of the optimal original ligand in the g-th iteration. is the information of the i-th dimension of the first random original ligand; is the information of the i-th dimension of the second random original ligand.

[0085] Exemplarily, the terminal device will first perform a differential operation on each original ligand in turn. Suppose the first original ligand X 1 is obtained; then, 2 random original ligands are randomly selected from the original ligand group and an optimal original ligand X is randomly selected from the first N target original ligands pbest . It should be noted that the selected individuals do not repeat. If they repeat, re-selection is required. Then, differential processing is performed according to the above differential calculation formula (9).

[0086] Differential processing means that each of the above ligands has information in multiple dimensions such as position parameters (x, y, z), rotation parameters (θ, α, β), and rotatable bond parameters (torsions). Then, the terminal device can calculate the difference for each parameter according to the differential calculation formula, that is, calculate the x parameter in X 1 and the x parameters in the other three ligands to obtain a new x parameter (i.e., the differential information in the x dimension). The remaining y, z and other parameters are also calculated according to the differential calculation formula to obtain the differential information in each dimension.

[0087] It should be noted that in this embodiment, the target number of random original ligands used is 2, that is, the above calculation formula can be used. It can be understood that in another feasible way, when the target number is greater than 2, the differential calculation formula can also be adjusted accordingly. That is, in the above formula sum, the formula of is added to participate in the calculation. Where n represents the n-th random original ligand, and there is no limit to this.

[0088] After that, after calculating the differential information in all dimensions of each of the above original ligands, the terminal device can process the differences through the following steps to generate a new ligand corresponding to each original ligand; that is, the above process of iterating the original ligand group.

[0089] S103. The terminal device generates a new ligand corresponding to each original ligand according to the differential information in all dimensions of each original ligand.

[0090] In one embodiment, as described above, the differential information also corresponds to multiple dimensions. Therefore, when generating a new ligand, it is necessary to further determine whether the information in each dimension of the new ligand is the information of the dimension of the selected original ligand or the differential information of the corresponding dimension.

[0091] Specifically, refer to Figure 4, the terminal device can generate a new ligand through the following steps S401 - S405, which are described in detail as follows:

[0092] S401. The terminal device obtains the crossover probability of the original ligand.

[0093] In one embodiment, the above-mentioned crossover probability is obtained in a similar way to the differential factor, both are generated according to the number of iterations and the first preset number. Specifically, referring to the above formulas (7) and (8), just replace F in the formulas with CR.

[0094] It should be noted that when the above-mentioned crossover probability and differential factor are randomly generated for the first time according to the probability distribution of the target parameters, their generation methods are different. Specifically, please refer to the following formulas (10) and (11):

[0095] F i = randc(M F,K , 0.05); (10)

[0096] CR i = randn(M CR,K , 0.1); (11)

[0097] Among them, F i the above represents the differential factor, and CR i represents the crossover probability. Among them, F i follows the Cauchy distribution (that is, the differential factor corresponding to each dimension follows the Cauchy distribution), and the standard deviation is 0.05; CR i follows the normal distribution (that is, the crossover probability corresponding to each dimension follows the normal distribution), and the standard deviation is 0.1.

[0098] S402. The terminal device generates the selection probability of the information of the target dimension of the new ligand; the target dimension is any one of multiple dimensions.

[0099] S403. The terminal device randomly generates the information of the target dimension of the new ligand. Or,

[0100] S404. If the selection probability is less than or equal to the crossover probability, the terminal device determines the differential information of the target dimension in the original ligand as the information of the target dimension of the new ligand; if the selection probability is greater than the crossover probability, the terminal device determines the information of the target dimension in the original ligand as the information of the target dimension of the new ligand.

[0101] S405. The terminal device generates a new ligand according to the information of the target dimension of the new ligand.

[0102] In one embodiment, for the information of any target dimension, the above-mentioned crossover probability is used to select from the differential information of the target dimension in the original ligand and the information of the target dimension in the original ligand, so as to adaptively generate the information of the target dimension of the new ligand.

[0103] It should be noted that in the initial iteration stage of the original ligand group, randomly generating the crossover probability according to the probability distribution to generate the information of multiple dimensions of the new ligand can enable the terminal device to perform sufficient exploratory actions in the global search space of the ligand. After that, using the adaptive selection probability and crossover probability to generate the information of multiple dimensions of the new ligand can make the value of the crossover probability decrease adaptively to alleviate the fitness degradation when the information rate of any dimension affects the fitness of the population.

[0104] In one embodiment, the above-mentioned selection probability is a randomly generated probability. Specifically, the terminal device can generate the information of the target dimension of the new ligand according to the following formula (12):

[0105]

[0106] where the above-mentioned rand(0,1) represents randomly generating a number within the range of 0 to 1 as the selection probability; i represents the target dimension, g represents the number of iterations; V j,i,g represents the differential information of the target dimension in the original ligand; X j,i,g represents the information of the target dimension in the original ligand; j rand represents randomly generating the information of a target dimension, so that each time differential processing is performed, the differential information of one dimension can be retained, which is also the key for the terminal device to perform local search when CR is 0 finally.

[0107] After that, the terminal device can generate a new ligand according to the information of each target dimension.

[0108] In a specific example, assuming that the first ligand for which crossover processing is currently being performed, the terminal device can find the original ligand X before differentiation 1 and the differential information V after differentiation 1 . Then, for each of their target dimensions, a selection probability is randomly generated between 0 and 1. If the selection probability is less than the generated crossover probability, the differential information of the corresponding target dimension of V 1 is taken, otherwise the information of the corresponding target dimension of X 1 is taken. Repeat the above operation for each dimension in the ligand to obtain the information of each target dimension and save it in U 1 , which is the generation of the new ligand.

[0109] It can be seen that if the value of the crossover probability is 0, the information of all target dimensions will be in the original ligand X 1Obtained from. However, in actual application scenarios, it is desired that the newly generated ligands can at least produce variations. Therefore, in each crossover, the terminal device can choose to randomly generate information for a target dimension; afterwards, if the terminal device does not randomly generate information for a target dimension, then determine the information for the target dimension according to the selection probability and the crossover probability. In this way, the terminal device can, to a certain extent, ensure that the finally generated U 1 at least one piece of information for a target dimension in the information for the target dimension is obtained from the differential information V 1 Obtained from.

[0110] S104. The terminal device calculates the ligand energy of each original ligand respectively according to the information of multiple dimensions of each original ligand, and calculates the ligand energy of the corresponding new ligand respectively according to the differential information of multiple dimensions of each original ligand.

[0111] In one embodiment, it has been described in the above S301 that the ligand energy is related to the information of dimensions and belongs to the prior art, and no detailed description will be given here.

[0112] S105. The terminal device updates the original ligand group according to the ligand energy of each new ligand and the ligand energy of each original ligand to obtain a target ligand group; the total ligand energy of the target ligand group is lower than the total ligand energy of the original ligand group.

[0113] In one embodiment, after calculating the ligand energy of the new ligand and the ligand energy of the original ligand, in order to obtain ligands with better stability, the terminal device can select new ligands or original ligands with low ligand energy to form a target ligand group.

[0114] It should be noted that in the selection process, the ligand energy of the original ligand is compared with the ligand energy of the corresponding new ligand, and the ligand with low ligand energy is selected. Specifically, the way to select the ligand with low ligand energy can be as the following formula (13):

[0115]

[0116] Among them, f(U i,g ) represents the ligand energy of the a-th new ligand in the g-th iteration process; f(X i,g ) represents the ligand energy of the a-th original ligand in the g-th iteration process.

[0117] Therefore, according to the above formula, the specific way to obtain the target ligand group is: for any new ligand, if the ligand energy of the new ligand is less than the ligand energy of the corresponding original ligand, then determine the new ligand as the preferred ligand; if the ligand energy of the new ligand is greater than or equal to the ligand energy of the corresponding original ligand, then determine the original ligand as the preferred ligand; construct the target ligand group according to the preferred ligands.

[0118] Based on this, it can be determined that the total ligand energy of the target ligand group composed of preferred ligands with low ligand energy will be lower than that of the original ligand group. That is, each preferred ligand in the target ligand group has better stability.

[0119] In this embodiment, after generating the original ligands composed of multiple dimension information, differential processing is performed on each dimension information respectively to obtain the differential information of each dimension, and new ligands are formed based on the differential information. In this way, the dimension information between the generated new ligands can have multiple mutations simultaneously with the dimension information of the original ligands, enhancing the search ability for ligands. Moreover, the differential information of multiple dimensions can reflect the variable relationship between the information of multiple dimensions. Therefore, the new ligands obtained based on the differential information of multiple dimensions can maintain the variable relationship between the information of each dimension in the new ligands, enabling the terminal device to generate new ligands more comprehensively. Furthermore, the terminal device can further achieve a comprehensive search for ligands. Finally, the original ligand group is updated according to the ligand energy of each new ligand to obtain a target ligand group with a lower total ligand energy, such that each ligand in the target ligand group is a preferred ligand with better stability under comprehensive search.

[0120] In one embodiment, since each original ligand and new ligand has multiple dimensions of information, when performing a global search for ligands, the terminal device will face the problem of dimensionality disaster, resulting in poor optimization effect when optimizing the original ligand group. Therefore, in this embodiment, with reference to Figure 5 , the terminal device can specifically improve the optimization performance in the docking problem with multiple dimensions through the following steps S501 - S503.

[0121] Details are as follows:

[0122] S501. The terminal device determines the group size of the target ligand group according to the iteration times when the original ligand group is updated.

[0123] In one embodiment, the original ligand group usually contains a large number of original ligands, and this quantity is usually large; if the size of the original ligand group remains unchanged in each iteration, the terminal device will consume a large amount of time in each iteration process. Based on this, the terminal device can gradually reduce the group size of the target ligand group in each iteration according to the iteration times.

[0124] Specifically, the terminal device can calculate the group size through the following formula (14):

[0125]

[0126] where N final and N initrespectively represent that both the minimum scale and the maximum scale of the population are preset scales; g represents the current number of iterations; G represents the second preset number of times, which can be the total number of times the original ligand group needs to be iterated; round represents the rounding function.

[0127] S502. For any new ligand of the terminal device, if the new ligand is determined to be an optimal ligand, perform gradient descent on the ligand energy of the new ligand to obtain the ligand energy of the target new ligand.

[0128] In one embodiment, the above gradient descent is used to decrease the ligand energy of the new ligand, generate an adaptive learning rate for the ligand energy of each new ligand, and use the exponentially decaying average of the gradients in the previous iteration process to suppress oscillations, so as to accelerate the convergence rate of generating the target ligand group.

[0129] In a specific embodiment, the terminal device can specifically process the ligand energy of the new ligand through the Adaptive Moment Estimation (Adam) algorithm; the details are as follows:

[0130] Adam gradient descent can perform an exponentially weighted average on the original gradient, then perform a normalization process, and then update the gradient value:

[0131] First, the first-order momentum in the current iteration process can be processed using the following formula (15):

[0132] m t =β 1 m t-1 +(1 - β 1 )dw; (15)

[0133] Then calculate the second-order momentum in the previous iteration process through formula (10):

[0134] v t =β 2 v t-1 +(1 - β 2 )*(dw) 2 (16)

[0135] where m 0 =0, v 0 =0, t is the number of iterations, dw is the gradient calculated when calculating the ligand energy of the new ligand, this formula starts to calculate from t = 0, and t is incremented by one each time it is calculated; β 1 and β 2 are preset parameters.

[0136] Subsequently, perform bias correction and gradient update processing according to the following formulas (17), (18), and (19) respectively:

[0137]

[0138]

[0139]

[0140] where w t is the gradient at the t-th iteration; α and ∈ are preset parameters respectively.

[0141] In this embodiment, the above parameters are specifically: β 1 = 0.5, β 2 = 0.999, α = 0.01, ∈ = 10 -8 .

[0142] It should be added that the information of the overall rotation dimension of the ligand described in S101 can be represented by quaternions and three angles respectively. In this embodiment, the above quaternion is specifically used in the calculation of ligand energy and the process of gradient descent, while the three angles are used in the process of difference.

[0143] In one embodiment, after obtaining the gradient in the current iteration process, the ligand energy of the new ligand is updated by using the gradient to obtain the ligand energy of the target new ligand, so as to accelerate the convergence speed of generating the target ligand group.

[0144] S503. The terminal device selects preferred ligands with the same group size and low ligand energy to construct a target ligand group.

[0145] In one embodiment, after obtaining the ligand energy of the target new ligand, the ligand energy of the target new ligand is compared with the ligand energy of the corresponding original ligand, and the ligand with low energy is selected as the preferred ligand. Then, according to the ligand energy magnitude of each preferred ligand, preferred ligands with the same group size and low ligand energy are selected to construct a target ligand group. In this way, the terminal device can avoid spending a lot of time in each iteration process, and further accelerate the convergence speed of generating the target ligand group.

[0146] In one embodiment, when iteratively updating the original ligand group, the number of iterations is usually not 1. Therefore, the terminal device needs to preset the number of iterations for updating the original ligand group to jump out of the loop. Based on this, when obtaining the target ligand group, the terminal device can also count whether the number of iterations for updating the original ligand group is less than the second preset number. If it is less than the second preset number, the target ligand group is determined as the new original ligand group, and the new original ligand group is continuously updated. If it is equal to the second preset number, the target ligand group is directly output.

[0147] It is understandable that during the iteration process, the original ligand group in the current iteration is the target ligand group in the previous iteration process.

[0148] In one embodiment, the above-mentioned target ligand group is the group obtained by the terminal device executing the above-mentioned ligand optimization method using the koto molecular docking software. The purpose of obtaining the target ligand group is to be able to dock with the receptor to form a drug small molecule with stable effects. Therefore, the actual effect of executing the above-mentioned ligand optimization method on the above-mentioned koto molecular docking software can be characterized according to the binding degree of docking with the receptor. Specifically, the experimental results are compared as shown in Table 1 below:

[0149] Programs Top-score poses Best poses Glide 47.54％ 68.31％ GOLD 65.49％ 85.56％ rDock 57.89％ 76.84％ LeDock 70.53％ 82.11％ DOCK 60.53％ 74.06％ AutoDock 54.58％ 70.07％ AutoDock-Vina 64.56％ 74.74％ AutoDock-Koto 71.58％ 90.53％

[0150] Among them, Programs represents the type of molecular docking software, and the above types are all commonly used molecular docking software, which will not be elaborated here. Among them, AutoDock-Koto is the docking software used by the terminal device in the embodiments of the present application; Top-score poses represents the docking success rate of the ligand with the lowest energy in the target ligand group with the receptor, and Best poses represents the success rate of all ligands in the target ligand group docking with the receptor respectively.

[0151] Among them, the pictures generated by the above experiments can specifically refer to Figure 6 、 Figure 7 and Figure 8 ; among them, Figure 6 、 Figure 7 and Figure 8 are respectively the effect diagrams of different ligands (1o0h, 2wn9, 3bv9) docking with the receptor under different docking software in an embodiment of the present application. It can be seen from the experimental comparison that when using the AutoDock-Koto molecular docking software in the lower right corner of each figure and the above-mentioned ligand optimization method, the ligand and the receptor have the best docking accuracy (coincidence degree).

[0152] Please refer to Figure 9 , Figure 9 which is the structural block diagram of a ligand optimization device provided by an embodiment of the present application. In this embodiment, each module included in the ligand optimization device is used to execute Figure 1 、 Figures 3 to 5 the corresponding steps in the corresponding embodiments. Specifically, please refer to Figure 1 、 Figures 3 to 5 and Figure 1 、 Figures 3 to 5 for the relevant descriptions in the corresponding embodiments. For the sake of convenience, only the parts related to this embodiment are shown. See Figure 9, the ligand optimization device 900 may include: an original ligand group generation module 910, a difference module 920, a new ligand generation module 930, a ligand energy calculation module 940, and an update module 950, where:

[0153] The original ligand group generation module 910 is configured to generate an original ligand group; the original ligand group includes a plurality of original ligands, and each original ligand is configured with information in multiple dimensions.

[0154] The difference module 920 is configured to perform difference processing on the information in each dimension of each original ligand to obtain difference information in each dimension of each original ligand.

[0155] The new ligand generation module 930 is configured to generate a new ligand corresponding to each original ligand respectively according to the difference information in all dimensions of each original ligand.

[0156] The ligand energy calculation module 940 is configured to calculate the ligand energy of each original ligand respectively according to the information in multiple dimensions of each original ligand, and calculate the ligand energy of the corresponding new ligand respectively according to the difference information in multiple dimensions of each original ligand.

[0157] The update module 950 is configured to update the original ligand group according to the ligand energy of each new ligand and the ligand energy of each original ligand to obtain a target ligand group; the total ligand energy of the target ligand group is lower than the total ligand energy of the original ligand group.

[0158] In one embodiment, the difference module 920 is further configured to:

[0159] Select N target original ligands from the original ligand group according to the ligand energy of each original ligand, and determine the optimal original ligand from the target original ligands; the ligand energy of the N target original ligands is lower than that of any other original ligand in the original ligand group except the target original ligands; randomly select a target number of random original ligands from the original ligand group; for any original ligand in the original ligand group, perform difference processing on the information in each dimension of the original ligand by using the information in each dimension of the optimal original ligand and the information in each dimension of the random original ligand to obtain difference information in each dimension of each original ligand.

[0160] In one embodiment, the target number is 2; the difference module 920 is further configured to:

[0161] Obtain the difference factor of the original ligand; import the difference factor, the information in each dimension of the optimal original ligand, and the information in each dimension of the random original ligand into a difference calculation formula to obtain difference information in each dimension of each original ligand; the difference calculation formula is as follows:

[0162]

[0163] wherein, i represents the i-th dimension; g represents the number of iterations; V i,g represents the differential information of the i-th dimension of the original ligand in the g-th iteration; X i,g represents the information of the i-th dimension of the original ligand in the g-th iteration; F i represents the differential factor of the information of the i-th dimension; X pbest,i,g represents the information of the i-th dimension of the optimal original ligand in the g-th iteration; is the information of the i-th dimension of the first random original ligand; is the information of the i-th dimension of the second random original ligand.

[0164] In one embodiment, the new ligand generation module 930 is further configured to:

[0165] Obtain the crossover probability of the original ligand; generate the selection probability of the information of the target dimension of the new ligand; the target dimension is any one of multiple dimensions; randomly generate the information of the target dimension of the new ligand; or, if the selection probability is less than or equal to the crossover probability, determine the differential information of the target dimension in the original ligand as the information of the target dimension of the new ligand; if the selection probability is greater than the crossover probability, determine the information of the target dimension in the original ligand as the information of the target dimension of the new ligand; generate a new ligand according to the information of the target dimension of the new ligand.

[0166] In one embodiment, the ligand optimization device 900 further includes:

[0167] An iteration number acquisition module, configured to acquire the number of iterations for which the original ligand group is updated.

[0168] A parameter generation module, configured to, if the number of iterations is less than or equal to a first preset number, randomly generate a differential factor and a crossover probability of the original ligand according to a probability distribution;

[0169] A parameter update module, configured to, if the number of iterations is greater than the first preset number, update the differential factor and the crossover probability in the current iteration process according to the ligand energy of the new ligand and the ligand energy of the original ligand in the previous iteration process.

[0170] In one embodiment, the update module 950 is further configured to:

[0171] For any new ligand, if the ligand energy of the new ligand is less than the ligand energy of the corresponding original ligand, determine the new ligand as the preferred ligand; if the ligand energy of the new ligand is greater than or equal to the ligand energy of the corresponding original ligand, determine the original ligand as the preferred ligand; construct a target ligand group according to the preferred ligand.

[0172] In one embodiment, the update module 950 is further configured to:

[0173] Determine the group size of the target ligand group according to the number of iterations for which the original ligand group is updated; for any new ligand, if the new ligand is determined to be a preferred ligand, perform gradient descent on the ligand energy of the new ligand to obtain the ligand energy of the target new ligand; select preferred ligands with a ligand energy lower than the group size and construct the target ligand group.

[0174] In one embodiment, the update module 950 is further configured to:

[0175] If the number of iterations for which the original ligand group is updated is less than the second preset number, determine the target ligand group as the new original ligand group; if the number of iterations for which the original ligand group is updated is equal to the second preset number, update the original ligand group to obtain the target ligand group.

[0176] It should be understood that Figure 9 In the structural block diagram of the ligand optimization device shown, each module is used to execute Figure 1 、 Figures 3 to 5 the respective steps in the corresponding embodiments, and for Figure 1 、 Figures 3 to 5 the respective steps in the corresponding embodiments have been explained in detail in the above embodiments. For details, please refer to Figure 1 、 Figures 3 to 5 and Figure 1 、 Figures 3 to 5 the relevant descriptions in the corresponding embodiments, which will not be elaborated here.

[0177] Figure 10 is the structural block diagram of a terminal device provided in an embodiment of the present application. As Figure 10 shown, the terminal device 1000 in this embodiment includes: a processor 1010, a memory 1020, and a computer program 1030 stored in the memory 1020 and executable on the processor 1010, such as a program for the ligand optimization method. When the processor 1010 executes the computer program 1030, it implements the steps in each of the above ligand optimization method embodiments, such as Figure 1 S101 to S105 shown. Alternatively, when the processor 1010 executes the computer program 1030, it implements the functions of each module in the corresponding embodiment above. For example, Figure 9 the functions of modules 910 to 950 shown. For details, please refer to Figure 9 the relevant descriptions in the corresponding embodiment. Figure 9

[0178] ​Exemplarily, the computer program 1030 can be divided into one or more modules. One or more modules are stored in the memory 1020 and executed by the processor 1010 to implement the ligand optimization method provided by the embodiments of the present application. One or more modules can be a series of computer program instruction segments capable of completing specific functions, and these instruction segments are used to describe the execution process of the computer program 1030 in the terminal device 1000. For example, the computer program 1030 can implement the ligand optimization method provided by the embodiments of the present application.

[0179] The terminal device 1000 may include, but is not limited to, a processor 1010 and a memory 1020. Those skilled in the art can understand that Figure 10 merely examples of the terminal device 1000, which do not constitute a limitation on the terminal device 1000. It may include more or fewer components than shown in the figure, or combine certain components, or different components. For example, the terminal device may also include input / output devices, network access devices, buses, etc.

[0180] The so-called processor 1010 may be a central processing unit, or may also be other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0181] The memory 1020 may be an internal storage unit of the terminal device 1000, such as the hard disk or memory of the terminal device 1000. The memory 1020 may also be an external storage device of the terminal device 1000, such as a plug-in hard disk, a smart memory card, a flash memory card, etc. equipped on the terminal device 1000. Further, the memory 1020 may also include both the internal storage unit and the external storage device of the terminal device 1000.

[0182] The embodiments of the present application provide a computer-readable storage medium, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the ligand optimization method in the above-mentioned various embodiments.

[0183] The embodiments of the present application provide a computer program product. When the computer program product runs on a terminal device, it causes the terminal device to execute the ligand optimization method in the above-mentioned various embodiments.

[0184] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A ligand optimization method, characterized in that, the method includes: generating an original ligand group; the original ligand group includes a plurality of original ligands, and each of the original ligands is configured with information in multiple dimensions; performing differential processing on the information in each dimension of each of the original ligands to obtain differential information in each dimension of each of the original ligands; generating a new ligand corresponding to each of the original ligands respectively according to the differential information in all dimensions of each of the original ligands; calculating the ligand energy of each of the original ligands respectively according to the information in multiple dimensions of each of the original ligands, and calculating the ligand energy of the corresponding new ligand respectively according to the differential information in multiple dimensions of each of the original ligands; updating the original ligand group according to the ligand energies of the respective new ligands and the ligand energies of the respective original ligands to obtain a target ligand group; the total ligand energy of the target ligand group is lower than the total ligand energy of the original ligand group; the performing differential processing on the information in each dimension of each of the original ligands to obtain differential information in each dimension of each of the original ligands includes: selecting N target original ligands from the original ligand group according to the ligand energies of the respective original ligands, and determining an optimal original ligand from the target original ligands; the ligand energies of the N target original ligands are all lower than any other original ligand in the original ligand group except the target original ligands; randomly selecting a target number of random original ligands from the original ligand group; the target number is 2; for any original ligand in the original ligand group, performing differential processing on the information in each dimension of the original ligand by using the information in each dimension of the optimal original ligand and the information in each dimension of the random original ligand to obtain differential information in each dimension of each of the original ligands; the performing differential processing on the information in each dimension of the original ligand by using the information in each dimension of the optimal original ligand and the information in each dimension of the random original ligand to obtain differential information in each dimension of each of the original ligands includes: obtaining a differential factor of the original ligand; importing the differential factor, the information in each dimension of the optimal original ligand, and the information in each dimension of the random original ligand into a differential calculation formula to obtain differential information in each dimension of each of the original ligands; the differential calculation formula is as follows: where, i represents the i-th dimension; g represents the number of iterations; V i,g represents the differential information of the i-th dimension of the original ligand in the g-th iteration; X i,g represents the information of the i-th dimension of the original ligand in the g-th iteration; F i represents the differential factor of the information of the i-th dimension; X pbest,i,g represents the information of the i-th dimension of the optimal original ligand in the g-th iteration; is the information of the i-th dimension of the first random original ligand; is the information of the i-th dimension of the second random original ligand; the generating a new ligand corresponding to each of the original ligands respectively according to the differential information in all dimensions of each of the original ligands includes: obtaining a crossover probability of the original ligand; generating a selection probability for the information in the target dimension of the new ligand; the target dimension is any one of multiple dimensions; randomly generating the information in the target dimension of the new ligand; alternatively, if the selection probability is less than or equal to the crossover probability, determining the differential information in the target dimension of the original ligand as the information in the target dimension of the new ligand; if the selection probability is greater than the crossover probability, determining the information in the target dimension of the original ligand as the information in the target dimension of the new ligand; Generate the new ligand according to the information of the target dimension of the new ligand; Updating the original ligand group according to the ligand energy of each new ligand and the ligand energy of the original ligand respectively to obtain a target ligand group, including: For any one of the new ligands, if the ligand energy of the new ligand is less than the ligand energy of the corresponding original ligand, then determine the new ligand as the preferred ligand; If the ligand energy of the new ligand is greater than or equal to the ligand energy of the corresponding original ligand, then determine the original ligand as the preferred ligand; Construct the target ligand group according to the preferred ligands.

2. The method according to claim 1, wherein, Before importing the difference factor, the information of multiple dimensions of the optimal original ligand, and the information of multiple dimensions of the random original ligand into the difference calculation formula to obtain the difference information of each dimension of each original ligand, it further includes: Obtain the number of iterations for which the original ligand group is updated; If the number of iterations is less than or equal to the first preset number, randomly select a set of initial parameters as the target parameters in the current iteration process; and if the number of iterations is greater than the first preset number, obtain the ligand energy of the new ligand and the ligand energy of the original ligand in the previous iteration process to update the target parameters in the previous iteration process to obtain the target parameters in the current iteration process; Generate the difference factor and crossover probability of the original ligand according to the probability distribution of the target parameters in the current iteration process.

3. The method according to claim 1, wherein, Updating the original ligand group according to the ligand energy of each new ligand and the ligand energy of the original ligand respectively to obtain a target ligand group, including: Determine the group size of the target ligand group according to the number of iterations for which the original ligand group is updated; For any one of the new ligands, if the new ligand is determined as the preferred ligand, perform gradient descent on the ligand energy of the new ligand to obtain the ligand energy of the target new ligand; Select the preferred ligands with ligand energies lower and equal to the group size to construct the target ligand group.

4. The method according to any one of claims 1-3, wherein, Updating the original ligand group according to the ligand energy of each new ligand and the ligand energy of the original ligand respectively to obtain a target ligand group, including: If the number of iterations for which the original ligand group is updated is less than the second preset number, determine the target ligand group as the new original ligand group; If the number of iterations for which the original ligand group is updated is equal to the second preset number, update the original ligand group to obtain the target ligand group.

5. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein, When the processor executes the computer program, the method according to any one of claims 1 to 4 is implemented.

6. A computer-readable storage medium, the computer-readable storage medium stores a computer program, wherein, When the computer program is executed by a processor, it implements the method according to any one of claims 1 to 4.

Images