Multi-scale coarse-grained modeling method and molecular dynamics simulation method

Abstract

A multi-scale coarse-grained modeling method and a molecular dynamics simulation method, the modeling method includes dividing a biological macromolecule composed of multiple monomers bonded together into multiple structural regions according to structural features; performing coarse-grained modeling of the multiple structural regions with different precisions, and obtaining multiple coarse-grained models corresponding to the multiple structural regions, respectively, the multiple coarse-grained models include interaction models that characterize interaction sites with single or multiple spheres; and linking the multiple coarse-grained models corresponding to the multiple structural regions with spring bonds to jointly form a multi-scale coarse-grained model of the biological macromolecule. The present invention performs large-scale molecular dynamics simulation based on the multi-scale coarse-grained model, which can improve the simulation efficiency by several to hundreds of times compared with the molecular dynamics of all-atom molecular dynamics or monomer-level coarse-grained models, and can achieve simulation of large-scale statistical behaviors at larger time and space scales.

Description

Technical Field

[0001] The present invention relates to the field of molecular dynamics technology, and in particular to a multi-scale coarse-grained modeling method and a molecular dynamics simulation method. Technical Background

[0002] Cells are composed of a large number of biomacromolecules, such as proteins, nucleic acids, and carbohydrates, which form the cell structure and realize various functions within the cell. However, how a large number of biomacromolecules work together to realize structure and function, and how to organize complex biochemical reactions within cells are still important scientific issues and have received widespread attention. For example, in order to create compartments suitable for specific biochemical reactions, biomolecules will be regionalized to form membraneless organelles, such as nucleoli, P particles, Cajal bodies, stress granules, etc. Their formation depends on the liquid-liquid phase separation process of biomacromolecules; in addition, centriole growth, microscopic growth, chromatin regionalization and other processes play an important role in cells. These intracellular life processes are constantly gaining new insights, but how to understand the formation mechanism of these processes in essence, how to understand the effects of mutations and environmental factors on various processes and even on functions, still poses huge challenges in experiments.

[0003] Molecular dynamics simulation is an important means of theoretical research and has become a key step in studying the molecular mechanisms within cells. In the past decade, with the rapid development of computer technology, theoretical chemists have been able to perform all-atom dynamics simulations to study conformational transitions of proteins with molecular weights of millions, receptor recognition, and other processes. Therefore, molecular dynamics simulation provides a promising method for understanding the relationship from microscopic biomolecules to macroscopic intracellular phenomena.

[0004] However, for the above-mentioned statistical behaviors such as phase separation and aggregation, since tens of thousands of biological macromolecules are involved, if all-atom dynamics simulation is used, the computational cost is extremely high and difficult to achieve under current technology. Therefore, it is necessary to adopt a more coarse-grained modeling and simulation. The inventors found that the coarse-graining methods of biological macromolecules in the prior art generally only involve a single scale, which is mainly divided into two categories: one is to coarse-grain the entire protein or RNA into a sphere. This treatment cannot reflect more structural information within the molecule and reflect the interaction site, and it is difficult to achieve accurate characterization when studying mutations and other problems; the other is to coarse-grain each monomer in the macromolecule into a sphere, thereby forming a soft ball chain. This treatment method can reflect sequence information, but it is insufficient for the high-level structural characterization of a large number of locally ordered macromolecules, and faces the problem of still huge computational complexity. Therefore, in the present invention, by modeling the multi-scale coarse-grained biological macromolecules, large-scale and efficient dynamic simulation can be achieved while grasping the main structural characteristics of the macromolecules and the distribution of the action sites, which plays an important role in studying the mechanism of the statistical behavior of biological macromolecules. Summary of the invention

[0005] In view of this, the main purpose of the present invention is to provide a multi-scale coarse-grained modeling method and a molecular dynamics simulation method, in order to at least partially solve at least one of the above-mentioned technical problems.

[0006] To achieve the above object, the technical solution of the present invention is as follows:

[0007] According to one aspect of the present invention, a multi-scale coarse-grained modeling method for biomacromolecules is provided, comprising:

[0008] A biological macromolecule composed of multiple monomers bonded together is divided into multiple structural regions according to structural features; coarse-grained modeling of the multiple structural regions is performed with different accuracies to obtain multiple coarse-grained models corresponding to the multiple structural regions, respectively, and the multiple coarse-grained models include interaction models that characterize interaction sites with single or multiple spheres; the multiple coarse-grained models corresponding to the multiple structural regions are linked with spring bonds to jointly form a multi-scale coarse-grained model of the biological macromolecule.

[0009] According to another aspect of the present invention, a molecular dynamics simulation method based on multi-scale coarse-grained modeling is provided, comprising: establishing a multi-scale coarse-grained model using the multi-scale coarse-grained modeling method as described above; generating an initial configuration of the system required for dynamics simulation based on the multi-scale coarse-grained model and the size of the simulation system; evolving the system using the molecular dynamics motion equation based on the initial configuration of the system to obtain simulation results regarding the multi-scale coarse-grained model.

[0010] Based on the above technical solution, the multi-scale coarse-grained modeling method and molecular dynamics simulation method of the present invention have at least one or part of the following beneficial effects:

[0011] (1) This multi-scale coarse-grained model is used for large-scale molecular dynamics simulations, which divides biological macromolecules into different structural regions. Coarse-grained modeling with different accuracies is performed according to the characteristics of different structural regions. This can improve the simulation efficiency by several to hundreds of times compared with all-atom molecular dynamics or molecular dynamics of monomer-level coarse-grained models, and can achieve simulation of large-scale statistical behaviors at larger time and space scales.

[0012] (2) The present invention generates the initial configuration of the simulation system based on a multi-scale coarse-grained model, and simulates the evolution of the system by selecting a kinetic equation adapted to the model, thereby obtaining the formation mechanism of statistical formations such as phase transition and aggregation, as well as intermolecular statistical information that is difficult to obtain experimentally, and providing efficient guidance and solutions for corresponding biological system experiments through parameter dependence. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1It is a coarse-grained schematic diagram of the EB1 protein sequence structure information and each structural region in Example 1 of the present invention;

[0014] Figure 2 is a multi-scale coarse-grained model structure diagram of the EB1 protein in Example 1 of the present invention;

[0015] Figure 3 is a system state diagram of the equilibrium state of the multi-scale coarse-grained model based on EB1 protein in Example 1 of the present invention in molecular dynamics simulation;

[0016] Figure 4A It is a statistical diagram of droplet growth dynamics during molecular dynamics simulation of the multi-scale coarse-grained model based on EB1 protein in Example 1 of the present invention;

[0017] Figure 4B It is a statistical diagram of the proportion of intramolecular interactions in the molecular dynamics simulation process of the multi-scale coarse-grained model based on EB1 protein in Example 1 of the present invention;

[0018] Figure 5 is a coarse-grained schematic diagram of the G3BP1 protein sequence structure information and each structural region in Example 2 of the present invention;

[0019] Figure 6 is a multi-scale coarse-grained model structure diagram of the G3BP1 protein in Example 2 of the present invention;

[0020] Figure 7 is a system state diagram of the equilibrium state of the multi-scale coarse-grained model based on the G3BP1 protein in Example 2 of the present invention in molecular dynamics simulation;

[0021] Figure 8 It is the average valence statistics of the multi-scale coarse-grained model based on G3BP1 protein in Example 2 of the present invention and RNA at the equilibrium state of molecular dynamics simulation. DETAILED DESCRIPTION

[0022] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0023] The present invention aims at the problems of low simulation efficiency and difficulty in accurately characterizing biological macromolecules in current coarse-grained simulation of biological molecules. The present invention provides a multi-scale coarse-grained modeling method and a molecular dynamics simulation method based on multi-scale coarse-grained modeling. Specifically, with the coarse-grained simulation of biological macromolecules as the core, coarse-grained modeling of different precisions is performed on different structural regions of molecules or their polymers according to the structural characteristics of biological molecules, and then a model of interaction sites is established in the coarse-grained model corresponding to the structural region to provide the most efficient model for dynamic simulation. Furthermore, based on the generation of the multi-scale coarse-grained model, the initial configuration of the simulation system is determined, and the evolution of the system is simulated by selecting the kinetic equations adapted to the model, so as to obtain the formation mechanism of statistical behaviors such as phase change and aggregation and the intermolecular statistical information that is difficult to obtain experimentally, and provide efficient guidance and solutions for the corresponding biological system experiments through parameter dependence.

[0024] Specifically, according to an embodiment of the present invention, a multi-scale coarse-grained modeling method for biomacromolecules is provided, comprising steps A to C.

[0025] In step A, a biomacromolecule composed of a plurality of monomers bonded together is divided into a plurality of structural regions according to structural features.

[0026] In some embodiments, the plurality of structural regions are independently disordered soft-stranded regions or ordered structural regions containing secondary or higher order structures.

[0027] Take protein as an example. It is a biological macromolecule composed of bonded amino acid monomers. Some regions of the protein molecule cannot fold themselves into a clear and unique spatial structure, that is, the disordered soft chain region, and some regions are ordered structural regions containing secondary or higher structures, which can be folded into ordered structures, including secondary structures such as α-helix, β-fold, and β-turn, and tertiary structures such as specific spatial structures formed by further coiling and folding of polypeptide chains based on the secondary structure.

[0028] In step B, coarse-grained modeling with different precisions is performed on multiple structural regions to obtain multiple coarse-grained models corresponding to the multiple structural regions, respectively. The multiple coarse-grained models include interaction models that characterize interaction sites with single or multiple spheres.

[0029] In some embodiments, when the structural region is a disordered soft chain region, a single sphere is used to represent the monomer, such as the amino acid monomer of a protein, to form a spherical soft chain model composed of multiple spheres connected together; when the structural region is an ordered structural region, a single sphere or multiple spheres are used to form a rigid body model that is similar in morphology to the ordered structural region.

[0030] It should be noted that the bead-soft-chain model or the rigid body model is for molecular dynamics calculations, and can be distinguished based on the calculation of the interaction potential and the dynamics equation. Specifically, the interaction potential between the two spheres in the bead-soft-chain model is set as a spring, indicating that the two spheres can move relative to each other, thus forming a soft chain, while there is no interaction between the two spheres in the rigid body model or the resultant force is 0, and when solving by molecular dynamics, the position of each sphere in the rigid body model is updated according to its relative position relative to the center of mass of the rigid body model.

[0031] More specifically, when the structural region is an ordered structural region, if the ordered structural region can reflect the main physical properties when it is approximated as a single sphere, it is modeled as a single spherical rigid body model; otherwise, multiple spheres are used to outline the outline of the ordered structural region according to the shape characteristics of the ordered structural region to form a rigid body.

[0032] In some embodiments, an interaction model can be established based on the interactions that are important factors of interest during the molecular dynamics simulation, wherein a single or multiple spheres used to characterize the interaction sites are located within the bead soft chain model or on the surface of the rigid body model, wherein each sphere is used to characterize a single interaction site.

[0033] It should be noted that the single or multiple spheres used to characterize the interaction sites are located in the bead-soft chain model, which means that the single or multiple spheres used to characterize the monomers in the bead-soft chain model are also used to characterize the interaction sites. These small spheres used to characterize the interaction sites can be attached with charges or interaction potentials, etc., according to the interactions of interest, in order to perform molecular dynamics calculations.

[0034] In some embodiments, when there is NMR data reflecting the interaction strength between monomers of biological macromolecules, the position and size of a single or multiple spheres are determined based on the crystal diffraction structure of the biological macromolecules and the NMR data; for example, taking protein molecules as an example, the amino acid monomers in the protein molecules that can produce key interactions can be determined based on the NMR data, and then these amino acid monomers can be combined with the crystal structure of the protein molecules to establish an interaction model as a single or multiple spheres.

[0035] Alternatively, in the absence of NMR data, the position and size of a single or multiple spheres are determined based on the crystal diffraction structure of the biomacromolecule and the simulation experimental parameters. For example, the combination of crystal structure and simulation experimental parameters such as pH and ion concentration can be used to determine the charge of amino acid monomers in the protein, and the position and size of the spheres representing the interaction site are determined based on the charge of these amino acid monomers.

[0036] On this basis, in the case of modeling a single sphere as a biological molecule, the model does not adequately reflect the biological molecule, can provide less statistical information, and the factors that can be compared with the experiment are more qualitative, and the predictive ability is poor, which is directly manifested in the large error between the simulation results and the experiment. The coarse-grained interaction sites determined based on NMR and crystal diffraction data are in good agreement with the actual biological experimental results, indicating that the multi-scale simulation process is highly accurate. It will help lay the foundation for future studies on the impact of gene mutations on the development mechanism of diseases, as well as for the treatment of diseases using this gene as a target.

[0037] In some embodiments, if the molecular dynamics simulation involves intracellular processes, if the biological macromolecules have formed oligomers before the intracellular processes occur, modeling can be performed as oligomers. Specifically, the coarse-grained models corresponding to the structural region where polymerization occurs in the multi-scale coarse-grained models of two or more of the biological macromolecules are combined into a single rigid body model.

[0038] In step C, multiple coarse-grained models corresponding to multiple structural regions are linked by spring bonds to form a multi-scale coarse-grained model of the biomacromolecule.

[0039] According to an embodiment of the present disclosure, a molecular dynamics simulation method based on multi-scale coarse-grained modeling is also provided, comprising steps a to c.

[0040] In step a, a multi-scale coarse-grained model is established using the multi-scale coarse-grained modeling method as described above.

[0041] In step b, the initial configuration of the system required for the dynamics simulation is generated based on the multi-scale coarse-grained model and the size of the simulation system.

[0042] In some embodiments, step b includes sub-steps b1-b2:

[0043] In sub-step b1, a simulation box with a preset periodic boundary is established according to the size of the simulation system, and the simulation box contains a multi-scale coarse-grained model of a preset number of biomacromolecules;

[0044] In sub-step b2, the initial coordinates of each sphere in the multi-scale coarse-grained model of each biomacromolecule are determined cyclically using the same number of cycles as the preset number of biomacromolecules.

[0045] Specifically, the initial coordinates of the first sphere in the multi-scale coarse-grained model are first determined, and then the initial coordinates of the remaining spheres in the multi-scale coarse-grained model are obtained by random walk based on the connection order of the coarse-grained models corresponding to each structural region.

[0046] In some embodiments, the initial coordinates of each sphere located on the rigid body model are determined based on the initial coordinates of the center of mass of the rigid body model and its relative position to the center of mass of the rigid body model.

[0047] In step c, the molecular dynamics equation of motion is used to evolve the system based on the initial configuration of the system to obtain simulation results of the multi-scale coarse-grained model.

[0048] In some embodiments, step c includes sub-steps c1 to c6:

[0049] In sub-step c1, a full neighbor list particle relationship matrix of all spheres in the simulation system is established based on the initial coordinates, initial velocity and interaction potential cutoff range of each sphere in the multi-scale coarse-grained model.

[0050] In sub-step c2, the forces on the spheres that interact with each other in the multi-scale coarse-grained model are calculated in sequence according to the full neighbor list particle relationship matrix, so as to cumulatively calculate the resultant force on each sphere;

[0051] In sub-step c3, the resultant force of all spheres in the rigid body model corresponding to the ordered structure region is subjected to quaternion operation to obtain the resultant force on the rigid body model;

[0052] In sub-step c4, the random force and random torque of the Langevin heat bath acting on each sphere in the rigid body model and the sphere soft chain model are calculated according to the resultant force acting on the rigid body model and the resultant force acting on each sphere in the sphere soft chain model;

[0053] In sub-step c5, the resultant force, random force and random torque acting on all spheres in the rigid body model and the ball-bead soft chain model are substituted into the Langevin equation to solve the displacement and rotation speed of all spheres in the multi-scale coarse-grained model;

[0054] In sub-step c6, the Verlet velocity integral algorithm is used to calculate and update the velocity and position information of all spheres in the multi-scale coarse-grained model, and the position of each sphere in the rigid body model is updated according to its relative position to the center of mass of the rigid body model.

[0055] In the above sub-steps c1 to c6, the establishment of the full neighbor list ion relationship matrix, the calculation of the random force and random torque under the Langevin heat bath, and the solution process of the Langevin equation are already well known to those skilled in the art, so they are not described here in detail. The difference between the present invention and the system evolution in conventional dynamics simulation is that the system evolution process based on the multi-scale coarse-grained model includes both the rigid body model and the ball-bead soft chain model for the calculation of the resultant force and position coordinates of the sphere.

[0056] The following are multiple specific embodiments to illustrate the technical solution of the present invention in detail. It should be noted that the specific embodiments below are only used for illustration and are not intended to limit the present invention. The software used for simulation in the specific embodiments is a self-programmed large-scale coarse-grained parallel platform. All crystal structures are derived from Protein Data Bank (PDB).

[0057] Embodiment 1:

[0058] This example performs multi-scale modeling and molecular dynamics simulation on the phase separation behavior of EB1 protein and its mutants, mainly including the following steps:

[0059] (1) Obtain the crystal structure of EB1 and perform coarse-grained modeling based on the structural characteristics. The EB1 protein mainly contains four structural regions from the C-terminus to the N-terminus, such as Figure 1 As shown in A: CH domain, disordered linker region, coiled coil (C-coil) region and disordered c-tail region. The CH domain can be approximated as a spherical rigid body model, as shown in Figure 1 As shown in C; the C-coil structure area is outlined with about 10 spheres according to its rod-like structural characteristics, as shown in Figure 1 As shown in D, these spheres together constitute a rigid body model in molecular dynamics calculations; in order to reflect the intrinsic disorder characteristics of the linker disordered region (70 amino acids in total), a spherical bead soft chain model with a single amino acid as the accuracy is adopted, as shown in Figure 1 As shown in B, the c-tail disordered region is modeled in a similar way to the linker disordered region, using a bead soft chain model with single amino acid accuracy. Since EB1 usually dimerizes first when the phase separation process of interest occurs, the EB1 dimer is further used as the modeling unit of the coarse-grained simulation, and the c-coil parts of the two monomers of the dimer are combined into a single rigid body to reflect its dimer characteristics.

[0060] (2) Determine the interactions that are important factors in the scientific issues of interest and establish a model of the interaction sites. Taking the phase separation process of EB1 in vivo as an example, its intermolecular interaction sites are classified as follows: the C-terminal domain and the disordered region in the middle. The dominant interaction is electrostatic interaction. On the spherical rigid body model of the CH structure region, the relative positions of the charged amino acids on the surface are referenced (a total of four positive charge sites and three negative charge sites). According to the relative positions of these sites in the crystal structure, the relative positions of the spheres of the coarse-grained interaction sites on the surface of the spherical rigid body model are determined. For charged amino acids in the disordered structure region, the coarse-grained spheres corresponding to the charged amino acids in the model are given corresponding charges. The C-coil structure region is mainly involved in hydrophobic interactions. Each sphere on the rod-shaped rigid body model of the C-coil structure region is used as an interaction site and the hydrophobic interaction potential between the two EB1s is set. Thus, the coarse-grained models corresponding to each structural region are linked with spring bonds to form a multi-scale coarse-grained model, such as Figure 2 shown.

[0061] (3) Determine the size of the simulation system and generate the initial configuration required for the dynamic simulation. Taking the simulation of 600 EB1 proteins with a concentration of 1.0 Mm as an example, a simulation box with a periodic boundary of 100 nm is required, and the simulation box contains a multi-scale coarse-grained model of 300 EB1 protein dimers. Based on the software programming "for" loop, all the coordinates of a dimer molecule are determined in each loop: first determine the coordinates of the first sphere during the loop, and then according to the multi-scale coarse-grained model established by the modeling method in steps (1) and (2), the coordinates of all spheres in the remaining coarse-grained structure are obtained by random walk according to the connection order of each structural region. The initial coordinates of the sphere located on the rigid body model are determined based on the rigid body's center of mass and relative position.

[0062] (4) Perform molecular dynamics simulation on the system, adopt Brownian dynamics for the coarse-grained biomacromolecules, and use the Langevin equation to evolve the system until the system evolves to an equilibrium state or the required time. Specifically, first, according to the initial coordinates, initial velocity and interaction potential cutoff range of each sphere in the multi-scale coarse-grained model, a full neighbor list particle relationship matrix of all spheres in the simulation system is established; then, according to the full neighbor list particle relationship matrix, the forces on the spheres that have interaction relationships in the multi-scale coarse-grained model are calculated in turn, so as to cumulatively calculate the resultant force on each sphere; then, the resultant force of all spheres in the rigid body model corresponding to the ordered structure region is subjected to quaternion operation to obtain the resultant force on the rigid body model; then, according to the resultant force on the rigid body model and the ball-bead soft chain model, the resultant force on the rigid body model is obtained. The resultant force on each sphere in the model is calculated, and the random force and random torque of the Langevin heat bath on each sphere in the rigid body model and the ball-ball soft chain model are calculated; then, the resultant force, random force and random torque on all spheres in the rigid body model and the ball-ball soft chain model are substituted into the Langevin equation to solve the displacement and rotation speed of all spheres in the multi-scale coarse-grained model; finally, the Verlet velocity integration algorithm is used to calculate and update the velocity and position information of all spheres in the multi-scale coarse-grained model, and the position of each sphere in the rigid body model is updated according to its relative position to the center of mass of the rigid body model.

[0063] (5) Based on step (4), the motion trajectory of each sphere in the coarse-grained model of the simulation process can be obtained over time. The screenshot of the system reaching the equilibrium state is as follows: Figure 3 As shown. Therefore, the simulation results can be analyzed and the droplet growth dynamics of EB1 phase separation can be statistically analyzed (such as Figure 4A ), the difference in the proportion of intramolecular interactions between the wild-type EB1 and each mutant (as shown in Figure 4B to explain the relevant mechanism.

[0064] Embodiment 2:

[0065] The present invention performs multi-scale modeling and molecular dynamics simulation on the G3BP1 protein, and the steps are as follows:

[0066] (1) Obtain the crystal structure of G3BP1 and perform coarse-grained modeling based on the structural characteristics. The G3BP1 protein mainly contains a domain: the RNA recognition motif (RRM), which can be approximated as a spherical rigid body model, such as Figure 5 As shown in A; a disordered structure region (RGG) is linked behind RRM, and the disordered structure region adopts a ball bead soft chain model with a single amino acid as the accuracy, as shown in Figure 5 As shown in B.

[0067] (2) Determine the interactions that are important factors in the scientific issues of concern and establish a model of the interaction sites. Taking the process of G3BP1 binding to RNA to form stress granules as an example, its dominant interaction with RNA is electrostatic interaction. Based on the experimental NMR data, it can be determined that there are 6 interaction sites on the RRM where G3BP1 binds to RNA and two interaction sites in the disordered region. Then, on the spherical rigid body model of RRM, the relative positions of the charged amino acids on the surface are referred to (a total of 6 positive charge sites), and the relative positions of these sites in the crystal structure are used to determine the relative positions of the spheres of the coarse-grained interaction sites on the surface of the spherical rigid body model. For charged amino acids in the disordered structural region, the coarse-grained spheres of the amino acids involved in the interaction determined by the NMR data are given corresponding charges. Thus, the coarse-grained models corresponding to each structural region are linked with spring bonds to form a multi-scale coarse-grained model, such as Figure 6 shown.

[0068] (3) Determine the size of the simulation system and generate the initial configuration required for dynamic simulation. Taking the simulation of 600 G3BP1 proteins with a concentration of 1.0 Mm as an example, a simulation box with a periodic boundary of 100 nm is required. Based on the software programming "for" loop, all the coordinates of a molecule are determined in each loop: the loop first determines the coordinates of the first sphere, and then according to the multi-scale coarse-grained model established by the modeling method in steps (1) and (2), the coordinates of all spheres in the remaining coarse-grained structure are obtained by random walk according to the connection order of each structural region. The initial coordinates of the sphere located on the rigid body model are determined based on the rigid body's center of mass and relative position.

[0069] (4) Perform molecular dynamics simulation on the system, adopt Brownian dynamics for the coarse-grained biomacromolecules, and use the Langevin equation to evolve the system until the system evolves to an equilibrium state or the required time. Specifically, first, according to the initial coordinates, initial velocity and interaction potential cutoff range of each sphere in the multi-scale coarse-grained model, a full neighbor list particle relationship matrix of all spheres in the simulation system is established; then, according to the full neighbor list particle relationship matrix, the forces on the spheres that have interaction relationships in the multi-scale coarse-grained model are calculated in turn, so as to cumulatively calculate the resultant force on each sphere; then, the resultant force of all spheres in the rigid body model corresponding to the ordered structure region is subjected to quaternion operation to obtain the resultant force on the rigid body model; then, according to the resultant force on the rigid body model and the ball-bead soft chain model, the resultant force on the rigid body model is obtained. The resultant force on each sphere in the model is calculated, and the random force and random torque of the Langevin heat bath on each sphere in the rigid body model and the ball-ball soft chain model are calculated; then, the resultant force, random force and random torque on all spheres in the rigid body model and the ball-ball soft chain model are substituted into the Langevin equation to solve the displacement and rotation speed of all spheres in the multi-scale coarse-grained model; finally, the Verlet velocity integration algorithm is used to calculate and update the velocity and position information of all spheres in the multi-scale coarse-grained model, and the position of each sphere in the rigid body model is updated according to its relative position to the center of mass of the rigid body model.

[0070] (5) Based on step (4), the motion trajectory of each sphere in the coarse-grained model of the simulation process can be obtained over time. The simulation results are analyzed and the screenshot of the system reaching the equilibrium state is shown in the following figure. Figure 7 As shown. Therefore, the mechanism analysis can be carried out based on the simulation results, such as statistically analyzing its average valence state with RNA, as shown in Figure 8 shown.

[0071] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A multi-scale coarse-grained modeling method for biomacromolecules, comprising: Dividing a biomacromolecule formed by bonding multiple monomers into multiple structural regions according to structural characteristics, wherein the structural regions are independently disordered soft chain regions or ordered structural regions containing secondary or higher structures; The multiple structural regions are subjected to coarse-grained modeling with different precisions to obtain multiple coarse-grained models corresponding to the multiple structural regions, respectively. The multiple coarse-grained models include interaction models that characterize interaction sites with a single or multiple spheres, including: when the structural region is a disordered soft chain region, a single sphere is used to characterize the monomer to form a bead-soft chain model formed by connecting multiple spheres; and when the structural region is an ordered structural region, a single sphere or multiple spheres are used to form a rigid body model that is similar in morphology to the ordered structural region; Linking the plurality of coarse-grained models corresponding to the plurality of structural regions with spring bonds to jointly form a multi-scale coarse-grained model of the biomacromolecule; The interaction potential between the two spheres in the ball-bead soft chain model is set as a spring, and the two spheres can move relative to each other to form a soft chain. In the rigid body model, there is no interaction between the two spheres or the resultant force is 0. When solving the problem using molecular dynamics, the position of each sphere in the rigid body model is updated according to its relative position relative to the center of mass of the rigid body model.

2. The multi-scale coarse-grained modeling method according to claim 1, characterized in that: The single or multiple spheres used to characterize the interaction sites are located within the bead soft chain model or on the surface of the rigid body model, wherein each sphere is used to characterize a single interaction site.

3. The multi-scale coarse-grained modeling method according to claim 2, characterized in that: The position and size of the single or multiple spheres are determined based on the crystal diffraction structure and nuclear magnetic resonance data of the biomacromolecule; or The position and size of the single or multiple spheres are determined based on the crystal diffraction structure of the biomacromolecule and simulation experiment parameters.

4. The multi-scale coarse-grained modeling method according to claim 1, characterized in that: In the case where two or more of the biomacromolecules form an oligomerization, the multi-scale coarse-grained modeling method further comprises: Taking the oligomer of two or more biomacromolecules as a unit, the coarse-grained models corresponding to the structural region where polymerization occurs in the multi-scale coarse-grained models of the two or more biomacromolecules are combined into a single rigid body model.

5. The multi-scale coarse-grained modeling method according to claim 1, characterized in that: The biomacromolecules are nucleic acids, proteins or sugars.

6. A molecular dynamics simulation method based on multi-scale coarse-grained modeling, comprising: Establishing a multi-scale coarse-grained model by using the multi-scale coarse-grained modeling method according to any one of claims 1 to 5; Generate an initial configuration of the system required for dynamic simulation based on the multi-scale coarse-grained model and the size of the simulation system; Based on the initial configuration of the system, the molecular dynamics equation of motion is used to evolve the system, and the simulation results of the multi-scale coarse-grained model are obtained.

7. The molecular dynamics simulation method according to claim 6, characterized in that: The generating of the initial system configuration required for the dynamics simulation based on the multi-scale coarse-grained model and the size of the simulation system comprises: A simulation box with a preset periodic boundary is established according to the size of the simulation system, wherein the simulation box contains a multi-scale coarse-grained model of a preset number of biomacromolecules; The initial coordinates of each sphere in the multi-scale coarse-grained model of each biomacromolecule are determined cyclically using the same number of cycles as the preset number of the biomacromolecules.

8. The molecular dynamics simulation method according to claim 7, characterized in that: The system evolution is performed based on the initial configuration of the system using the molecular dynamics equation of motion to obtain simulation results about the multi-scale coarse-grained model, including: According to the initial coordinates, initial velocity and interaction potential cutoff range of each sphere in the multi-scale coarse-grained model, a full neighbor list particle relationship matrix of all spheres in the simulation system is established; According to the full neighbor list particle relationship matrix, the forces on the spheres that have an interacting relationship in the multi-scale coarse-grained model are calculated in sequence, so as to cumulatively calculate the resultant force on each sphere; Performing quaternion operation on the resultant force of all spheres in the rigid body model corresponding to the ordered structure region to obtain the resultant force on the rigid body model; According to the resultant force on the rigid body model and the resultant force on each sphere in the ball-and-soft chain model, the random force and random torque of the Langevin heat bath on each sphere in the rigid body model and the ball-and-soft chain model are calculated; Substituting the resultant force, random force and random torque on all spheres in the rigid body model and the ball-bead soft chain model into the Langevin equation, the displacement and rotation speed of all spheres in the multi-scale coarse-grained model are obtained; The Verlet velocity integral algorithm is used to calculate and update the velocity and position information of all spheres in the multi-scale coarse-grained model, and the position of each sphere in the rigid body model is updated according to its relative position to the center of mass of the rigid body model.

Images