Ionizable lipid molecule prediction model generation method, device and equipment

Abstract

The invention provides an ionizable lipid molecule prediction model generation method, device and equipment, and belongs to the field of machine learning, and the method comprises the steps: obtaining the initial characteristics of ionizable lipid molecules; determining a target feature according to the initial feature, wherein the target feature is a high-dimensional combination of the initial feature; constructing a prediction model according to the target characteristics, wherein the prediction model is used for predicting the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA; obtaining a training set; and training the constructed prediction model according to the training set to obtain a target prediction model. In this way, the transfection effect of the ionizable lipid molecules on construction of the lipid nanoparticles used for carrying mRNA can be predicted, the design screening efficiency of the ionizable lipid molecules is improved, and the research and development speed is increased.

Description

Technical Field

[0001] The present application belongs to the field of machine learning, and specifically relates to a method, device and equipment for generating a prediction model of ionizable lipid molecules. Background Art

[0002] In recent years, mRNA nanomedicines have become a research hotspot for emerging gene drugs due to their rapid development, flexible design and efficient therapeutic potential. Compared with traditional small molecule drugs or protein drugs, mRNA nanomedicines have programmability and wide adaptability, and can accurately treat a variety of diseases. However, the effective delivery of mRNA nanomedicines remains one of the key bottlenecks in their clinical application. As the main carrier for mRNA delivery, lipid nanoparticles (LNPs) have been widely used in the preparation of mRNA nanomedicines due to their excellent biocompatibility, stability and targeting. Ionizable lipid molecules play a core role in the construction of LNPs, and their physicochemical properties directly affect the encapsulation efficiency, stability and targeted delivery effect of mRNA. Therefore, optimizing the structure of ionizable lipid molecules to improve the delivery efficiency and therapeutic effect of mRNA nanomedicines has become an important direction of current research.

[0003] At present, the design and optimization of ionizable lipid molecules mainly rely on experimental screening, which has the disadvantages of long cycle and high cost, and it is difficult to efficiently develop carriers that meet the needs of specific mRNA nanomedicines. Therefore, how to quickly predict the transfection effect of mRNA-LNPs after the construction of ionizable lipid molecules has become an important problem that needs to be solved urgently. Summary of the invention

[0004] The present application provides a method, device and equipment for generating a prediction model of ionizable lipid molecules, which are used to predict the transfection effect of ionizable lipid molecules on lipid nanoparticles for carrying mRNA, improve the design and screening efficiency of ionizable lipid molecules, and accelerate the research and development speed.

[0005] The present application provides a method for generating a prediction model of ionizable lipid molecules, comprising: Obtaining initial characteristics of ionizable lipid molecules; Determine a target feature according to the initial feature, wherein the target feature is a high-dimensional combination of the initial features; Constructing a prediction model according to the target feature, wherein the prediction model is used to predict the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA; Get the training set; The constructed prediction model is trained according to the training set to obtain a target prediction model.

[0006] According to the method for generating a prediction model for ionizable lipid molecules provided in the present application, the initial features include multiple chemical structure features and multiple spatial features, and determining the target features based on the initial features includes: combining the initial features through mathematical operations to obtain multiple candidate features; recursively selecting features whose correlation with the transfection effect is higher than a preset value from the candidate features; combining the features whose correlation is higher than the preset value to obtain multiple feature combinations; and determining the candidate feature included in the feature combination with the best prediction effect among the multiple feature combinations as the target feature.

[0007] According to the method for generating a prediction model for ionizable lipid molecules provided in the present application, the constructed prediction model is trained according to the training set to obtain a target prediction model, including: randomly selecting multiple experimental data as sub-training sets from the training set to obtain multiple sub-training sets; training the constructed prediction model according to the multiple sub-training sets respectively to obtain multiple trained prediction models; obtaining a test set, the test set including multiple experimental data; obtaining the actual transfection effect of each experimental data in the multiple experimental data included in the test set; obtaining the predicted transfection effect of each experimental data predicted by the trained prediction model; and selecting a target prediction model from the multiple trained prediction models according to the actual transfection effect and the predicted conversion effect.

[0008] According to the method for generating a prediction model for ionizable lipid molecules provided in the present application, the target prediction model is selected from the multiple trained prediction models according to the actual transfection effect and the predicted conversion effect, including: determining the prediction type of each trained prediction model for each experimental data according to the actual transfection effect and the predicted transfection effect; determining the accuracy, precision and recall rate of each trained prediction model according to the prediction type; scoring each trained prediction model according to the accuracy and the recall rate to obtain a target score; and selecting a target prediction model from the multiple trained prediction models according to the accuracy, the precision, the recall rate and the target score.

[0009] According to the method for generating a prediction model for ionizable lipid molecules provided in the present application, the prediction types include four types, wherein the first prediction type is used to indicate that the actual transfection effect is the first effect, and the predicted transfection effect is the first effect, the second prediction type is used to indicate that the actual transfection effect is the first effect, and the predicted transfection effect is the second effect, the third prediction type is used to indicate that the actual transfection effect is the second effect, and the predicted transfection effect is the first effect, the fourth prediction type is used to indicate that the actual transfection effect is the second effect, and the predicted transfection effect is the second effect, and the first effect is better than the second effect; the accuracy, precision and other factors of each trained prediction model are determined according to the prediction type of each experimental data. The accuracy and recall rate include: performing the following operations for each trained prediction model: determining the accuracy rate according to the number of the third prediction type and the fourth prediction type corresponding to the currently trained prediction model, and the total number of prediction types corresponding to the currently trained prediction model; determining the accuracy rate according to the number of the first prediction type corresponding to the currently trained prediction model, and the total number of the first prediction type and the third prediction type corresponding to the currently trained prediction model; determining the recall rate according to the number of the first prediction type corresponding to the currently trained prediction model, and the total number of the first prediction type and the second prediction type corresponding to the currently trained prediction model.

[0010] According to the method for generating a prediction model for ionizable lipid molecules provided in the present application, the initial features include spatial features, and the step of obtaining the initial features of the ionizable lipid molecules includes: obtaining multiple molecular conformations of the ionizable lipid molecules; aligning the molecular structures of the multiple molecular conformations respectively; density projecting the aligned multiple molecular structures to obtain a density map; and extracting the spatial features of the ionizable lipid molecules based on the density map.

[0011] According to the method for generating a prediction model of ionizable lipid molecules provided in the present application, the molecular structures of the multiple molecular conformations are aligned respectively, including: performing the following operations on the molecular structure of each molecular conformation in the multiple molecular conformations: obtaining a target central atom and two reference atoms from the head group of the current molecular structure; fixing the target central atom at the target coordinates; obtaining a first vector based on the target central atom and the molecular center of mass of the current molecular structure; obtaining a second vector based on the two reference atoms; rotating the current molecular structure so that the first vector is parallel to the first coordinate axis, and the second vector is parallel to the plane corresponding to the first coordinate axis and the second coordinate axis.

[0012] According to the method for generating a prediction model of ionizable lipid molecules provided in the present application, extracting the spatial characteristics of the ionizable lipid molecules based on the density map includes: obtaining a first boundary and a second boundary based on the density map, the density value corresponding to the first boundary belongs to a first threshold, and the density value corresponding to the second boundary belongs to a second threshold; extracting the spatial characteristics of the ionizable lipid molecules based on the first boundary and the second boundary, the spatial characteristics including at least one of the following: molecular length, molecular width, aspect ratio, cone angle cotangent value of the molecular head group, and symmetry of the molecular head group.

[0013] According to the method for generating a prediction model for ionizable lipid molecules provided in the present application, after the constructed prediction model is trained according to the training set to obtain a target prediction model, the method further includes: obtaining target spatial characteristics and target chemical structure characteristics of the target ionizable lipid molecules to be predicted; inputting the target spatial characteristics and the target chemical structure characteristics into the target prediction model to obtain a prediction result, and the prediction result is used to indicate the transfection effect of lipid nanoparticles for carrying mRNA constructed based on the target ionizable lipid molecules.

[0014] The present application also provides a device for generating a prediction model of an ionizable lipid molecule, comprising: a first acquisition unit, used to acquire initial features of ionizable lipid molecules; a determination unit, used to determine target features based on the initial features, wherein the target features are high-dimensional combinations of the initial features; a construction unit, used to construct a prediction model based on the target features, wherein the prediction model is used to predict the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA; a second acquisition unit, used to acquire a training set; and a training unit, used to train the constructed prediction model based on the training set to obtain a target prediction model.

[0015] The present application also provides an electronic 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, any of the above-mentioned methods for generating a prediction model of ionizable lipid molecules is implemented.

[0016] The present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for generating a prediction model for ionizable lipid molecules.

[0017] The present application also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-mentioned methods for generating a prediction model for ionizable lipid molecules.

[0018] The method for generating a prediction model of an ionizable lipid molecule provided in the present application first obtains the initial features of the ionizable lipid molecule, then determines the target features based on the initial features, the target features are high-dimensional combinations of the initial features, and then constructs a prediction model based on the target features, the prediction model is used to predict the transfection effect of the ionizable lipid molecule on the lipid nanoparticles used to carry mRNA, then obtains a training set, and finally trains the constructed prediction model based on the training set to obtain a target prediction model. In this way, the prediction of the transfection effect of the ionizable lipid molecule on the lipid nanoparticles used to carry mRNA can be achieved, the design and screening efficiency of the ionizable lipid molecules can be improved, and the research and development speed can be accelerated. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 It is a schematic diagram of a process for generating a prediction model of ionizable lipid molecules provided in this application.

[0021] Figure 2 It is a density projection process diagram provided by this application.

[0022] Figure 3 It is a schematic diagram of the molecular structure provided by this application.

[0023] Figure 4 It is a schematic diagram of spatial feature extraction provided by this application.

[0024] Figure 5 It is a schematic diagram of model verification provided by this application.

[0025] Figure 6 It is a structural schematic diagram of a device for generating a prediction model of ionizable lipid molecules provided in this application.

[0026] Figure 7 It is a structural schematic diagram of the electronic device provided by this application. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0029] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0030] At present, the design and optimization of ionizable lipid molecules mainly rely on experimental screening, which has the disadvantages of long cycle and high cost, and it is difficult to efficiently develop carriers that meet the needs of specific mRNA nanomedicines. Therefore, how to quickly predict the transfection effect of mRNA-LNPs after the construction of ionizable lipid molecules has become an important problem that needs to be solved urgently.

[0031] In response to the above problems, the embodiments of the present application provide a method, device and equipment for generating a prediction model of ionizable lipid molecules. The embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

[0032] See also Figure 1 , Figure 1 The present application provides a flow chart of a method for generating a prediction model of ionizable lipid molecules. The prediction model generation method comprises the following steps.

[0033] S101, obtaining initial features of ionizable lipid molecules.

[0034] Among them, the initial features may include chemical structure features and spatial features. For example, the chemical structure features include the total number of carbon atoms in the head group of the ionizable lipid molecule (the head group indicates the hydrophilic part surrounded by the connecting group), the total number of nitrogen atoms and oxygen atoms in the ionizable lipid molecule, the total number of functional groups that modify the head group (such as -CH2-, -OH, -CH3, each functional group is counted as one unit), the head group modification type (if a hydrophilic group is included, the value is 1, otherwise it is 0) and the total number of tails in the ionizable lipid molecule. In a specific implementation, the spatial features may include molecular length, molecular width, aspect ratio, cone angle cotangent value of the molecular head group, symmetry of the molecular head group, etc.

[0035] It should be noted that the molecules in this scheme all indicate ionizable lipid molecules used to construct lipid nanoparticles carrying mRNA.

[0036] S102: Determine a target feature according to the initial feature, where the target feature is a high-dimensional combination of the initial features.

[0037] Therein, the target feature may include multiple ones, and each target feature may be a mathematical operation combination of multiple sample initial features.

[0038] S103, constructing a prediction model according to the target feature, wherein the prediction model is used to predict the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA.

[0039] After obtaining the target features, the linear fitting coefficients corresponding to each target feature can also be obtained. In particular, a sparse linear model can be constructed by the sparse optimization (SO) method, and the final prediction model can be in the form of: P = c1D1+c2D2+c3D3+...+c n D n +c0 Among them, c1, c2, c3, c n represents the linear fitting coefficient, c0 is a constant, D1, D2, D3, D n , are target features, and P is the sample band property. In particular, n can be 5, that is, the constructed prediction model is constructed by 5 target features.

[0040] S104, obtaining a training set.

[0041] The training set includes multiple experimental data, each of which includes the initial characteristics of an ionizable lipid molecule and a sample band property. The sample band property is associated with the actual transfection effect of lipid nanoparticles constructed by the ionizable lipid molecule for carrying mRNA. In the specific implementation, the sample band property P = log10(ytrue ), y true The initial characteristics of the sample are the above initial characteristics, including spatial characteristics and chemical structure characteristics.

[0042] S105, training the constructed prediction model according to the training set to obtain a target prediction model.

[0043] It can be seen that in the embodiment of the present application, the initial features of the ionizable lipid molecules are first obtained, and then the target features are determined according to the initial features, the target features are high-dimensional combinations of the initial features, and then a prediction model is constructed according to the target features, the prediction model is used to predict the transfection effect of the ionizable lipid molecules on the lipid nanoparticles used to carry mRNA, and then a training set is obtained, and finally the constructed prediction model is trained according to the training set to obtain a target prediction model. In this way, the prediction of the transfection effect of the ionizable lipid molecules on the lipid nanoparticles used to carry mRNA can be achieved, the design and screening efficiency of the ionizable lipid molecules can be improved, and the research and development speed can be accelerated.

[0044] In a possible embodiment, the initial features include multiple chemical structure features and multiple spatial features, and determining the target features based on the initial features includes: combining the initial features through mathematical operations to obtain multiple candidate features; combining the features with correlations higher than a preset value to obtain multiple feature combinations; and determining the candidate feature included in the feature combination with the best prediction effect among the multiple feature combinations as the target feature.

[0045] The feature with a transfection effect correlation higher than a preset value may refer to a feature with the highest transfection effect correlation, which may include multiple features. The target feature is the candidate feature with the highest transfection effect correlation and has the best combination effect.

[0046] In a specific implementation, the initial features can be combined through mathematical operations (such as +, -, ×, ÷, log, exp, etc.) to generate high-dimensional candidate features. Then, the Sure Independence Screening (SIS) method is used to recursively select features whose correlation with the transfection effect or the above-mentioned P is higher than a preset value. In particular, the feature whose correlation is higher than the preset value can refer to the candidate feature with the highest correlation.

[0047] For example, the target features include D1=exp(-|A y1i -A y2i | / R y )、D2=(L i -W xi ) / A x1 、D3=(|Ay1i -A y2i |×R xi ) 2 、D4=|N N -N O | / R x 、D5=N head / N c-head ×A y1i , where A y1i is the cone angle of the molecular head group at the second boundary under the YZ plane projection, A y2i is the other cone angle of the molecular head group at the second boundary under the YZ plane projection, R y is the ratio of the molecular length to the width corresponding to the first boundary under YZ plane projection, L i is the molecular length, W xi A is the length of the second boundary under XZ plane projection, x1 is the cone angle of the molecular head group at the first boundary under XZ plane projection, R xi is the ratio of the molecular length to the width corresponding to the second boundary under XZ plane projection, N N is the total number of nitrogen atoms in the molecule, N O is the total number of oxygen atoms in the molecule, R x is the ratio of the molecular length to the width corresponding to the first boundary under XZ plane projection, N head is the total number of functional groups modifying the head group, N c-head is the total number of carbon atoms in the molecular head group. That is, the above subscript x represents the feature projected from the XZ plane, and the subscript y represents the feature projected from the YZ plane. The subscript i represents the feature projected from the second boundary (intramolecular contour S i ) to extract features. If the subscript i is not marked, the features are extracted according to the first boundary (molecule outer contour S).

[0048] It can be seen that in this embodiment, high-dimensional candidate features are first generated based on the initial features, and then the target features are selected based on the correlation with the variables, which can improve the prediction accuracy of the constructed prediction model.

[0049] In a possible embodiment, the constructed prediction model is trained according to the training set to obtain a target prediction model, including: selecting multiple experimental data as sub-training sets from the training set to obtain multiple sub-training sets; training the constructed prediction model according to the multiple sub-training sets respectively to obtain multiple trained prediction models; obtaining a test set, the test set including multiple experimental data; obtaining the actual transfection effect of each experimental data in the multiple experimental data included in the test set; obtaining the predicted transfection effect of each experimental data predicted by the trained prediction model; and selecting a target prediction model from the multiple trained prediction models according to the actual transfection effect and the predicted conversion effect.

[0050] The training data in the sub-training set can be randomly selected from the training set, and the training data included in each sub-training can be 90% of the training data included in the training set. That is, 90% of the training data can be randomly selected from the training set as the sub-training set. Each sub-training set can be trained to obtain a prediction model, that is, multiple molecules are randomly selected from the training set as sub-training sets for training, and each different selection can obtain a prediction model.

[0051] It can be seen that in this embodiment, multiple prediction models are obtained based on different sub-training sets, and then a target training model is selected from the multiple trained prediction models based on the transfection effect, which can enhance the prediction accuracy of the final prediction model.

[0052] In a possible embodiment, selecting a target prediction model from the multiple trained prediction models according to the actual transfection effect and the predicted conversion effect includes: determining a prediction type of each trained prediction model for each experimental data according to the actual transfection effect and the predicted transfection effect; determining the accuracy, precision and recall of each trained prediction model according to the prediction type; scoring each trained prediction model according to the accuracy and the recall to obtain a target score; and selecting a target prediction model from the multiple trained prediction models according to the accuracy, the precision, the recall and the target score.

[0053] When obtaining the target score, you can first obtain the target product of the precision and recall rate corresponding to the currently trained prediction model, and then obtain the target sum of the precision and recall rate, and obtain the target score based on the target product and the target sum. The target score (F1) can be calculated using the following formula: F1=(2×Precision×Recall) / (Precision+Recall), Among them, Precision is the precision of the current trained prediction model, and Recall is the recall rate of the current trained prediction model.

[0054] It can be seen that in this embodiment, the target prediction model is screened based on the accuracy, precision, recall and target score, which can ensure the prediction accuracy of the target prediction model obtained after the screening.

[0055] In a possible embodiment, the prediction types include four types, wherein the first prediction type is used to indicate that the actual transfection effect is the first effect, and the predicted transfection effect is the first effect, the second prediction type is used to indicate that the actual transfection effect is the first effect, and the predicted transfection effect is the second effect, the third prediction type is used to indicate that the actual transfection effect is the second effect, and the predicted transfection effect is the first effect, and the fourth prediction type is used to indicate that the actual transfection effect is the second effect, and the predicted transfection effect is the second effect, and the first effect is better than the second effect; the accuracy, precision and recall rate of each trained prediction model are determined according to the prediction type of each experimental data , including: performing the following operations for each trained prediction model: determining the accuracy rate according to the number of the first prediction type and the fourth prediction type corresponding to the currently trained prediction model, and the total number of prediction types corresponding to the currently trained prediction model; determining the precision rate according to the number of the first prediction type corresponding to the currently trained prediction model, and the total number of the first prediction type and the third prediction type corresponding to the currently trained prediction model; determining the recall rate according to the number of the first prediction type corresponding to the currently trained prediction model, and the total number of the first prediction type and the second prediction type corresponding to the currently trained prediction model.

[0056] Among them, the transfection effect can be divided into the first effect and the second effect according to the value corresponding to the transfection effect, that is, if the transfection efficiency is higher than the threshold, the transfection efficiency is considered to be the first effect, otherwise it is the second effect. The threshold can be 10 5 , that is, if y pre >10 5 For example, the first effect is used to indicate high efficiency, and the second effect is used to indicate low efficiency.

[0057] Accuracy, precision, and recall can be calculated using the following formulas: Accuracy: Accuracy = (TP+TN) / (TP+FN+FP+TN), Precision: Precision = TP / (TP+FP), Recall rate: Recall = TP / (TP+FN), Among them, TP (True Positive) indicates the number of the first prediction type included in the current training model, FN (False Negative) indicates the number of the second prediction type included in the current training model, FP (False Positive) indicates the number of the third prediction type included in the current training model, and TN (True Negative) indicates the number of the fourth prediction type included in the current training model. That is, if the sub-training set corresponding to the current training model includes 90 experimental data of ionizable lipid molecules, the total number of the first prediction type is 50, and the total number of the fourth prediction type is 35, then the corresponding accuracy is 94%.

[0058] It can be seen that in this embodiment, the accuracy, precision and recall are determined based on the number of prediction types, which can ensure the prediction accuracy of the target prediction model obtained after screening.

[0059] In a possible embodiment, the initial features include spatial features, and obtaining the initial features of the ionizable lipid molecules includes: obtaining multiple molecular conformations of the ionizable lipid molecules; aligning the molecular structures of the multiple molecular conformations respectively; density projecting the aligned multiple molecular structures to obtain a density map; and extracting the spatial features of the ionizable lipid molecules based on the density map.

[0060] Among them, the molecular conformation can be obtained by molecular dynamics simulation, which specifically includes first constructing a system containing a single ionizable lipid molecule and a solvent in a simulation box. The solvent can be various commonly used solvents, such as water and ethanol (EtOH) in a volume ratio. In particular, this embodiment uses pure water solvent. Then consider the ionization of the ionizable lipid molecules and examine the effect of the charged state on the three-dimensional structure of the molecule. At this time, in order to ensure the electrical neutrality of the system, Na⁺ and Cl⁻ ions are introduced. This embodiment uses non-ionized molecules. Finally, the classical force field charmm36 can be used to describe the molecule, tip3p water model. Other force fields can also be used for simulation. This embodiment simulates 100 ns, selects trajectory data from 20 ns to 100 ns for analysis, extracts the molecular structure every 0.04ns, and obtains a total of 2000 molecular conformations.

[0061] In a specific implementation, when performing density projection, density projection can be performed on both the XZ and YZ planes to obtain two density maps. The density map on the XZ plane represents the side information of the molecule, and the density map on the YZ plane represents the front information of the molecule. When performing density projection, it specifically includes superimposing and averaging the density of multiple target molecular structures to obtain an initial density map; normalizing the initial density map to obtain the density map, and the density value of the central nitrogen atom included in the density map is 1. For example Figure 2 As shown, Figure 2 It is a density projection process diagram provided in this application, that is, multiple single-molecule structures are structurally superimposed and density averaged, and then a two-dimensional density projection diagram on the XZ and YZ planes is obtained.

[0062] It can be seen that in this embodiment, after the molecular structure is aligned, the spatial features are extracted based on the acquired density map, which can improve the accuracy of the extracted spatial features.

[0063] In a possible embodiment, the aligning processing of the molecular structures of the multiple molecular conformations respectively includes: performing the following operations on the molecular structure of each molecular conformation in the multiple molecular conformations: obtaining a target central atom and two reference atoms from the head group of the current molecular structure; fixing the target central atom at the target coordinates; obtaining a first vector based on the target central atom and the molecular center of mass of the current molecular structure; obtaining a second vector based on the two reference atoms; rotating the current molecular structure so that the first vector is parallel to the first coordinate axis, and the second vector is parallel to the plane corresponding to the first coordinate axis and the second coordinate axis.

[0064] The target central atom is the central nitrogen atom; the reference atom is the nitrogen atom or a carbon atom connected to the nitrogen atom.

[0065] In the specific implementation, Figure 3 As shown in the figure, firstly, the molecular dynamics simulation was performed for 100 ns, and the trajectory data from 20 ns to 100 ns were selected for analysis. The molecular structure was extracted every 0.04 ns, and a total of 2000 ionizable lipid molecular conformations were obtained. Then, when aligning the molecular structure, the central nitrogen atom N center Fixed at a specific position in the coordinate system, and then by N center Define the vector n1 to the molecular mass center Q, rotate the molecule so that n1 is parallel to the Z axis and in the same direction as the positive Z axis. The vector n2 defined by the two other nitrogen atoms N1 and N2 is parallel to the YZ plane. In particular, if the head group contains less than 3 nitrogen atoms, the two carbon atoms connected to the nitrogen atoms are used instead.

[0066] It can be seen that in this embodiment, the alignment processing efficiency can be improved by aligning the atoms in the head group and the molecular mass center.

[0067] In a possible embodiment, extracting the spatial characteristics of the ionizable lipid molecules according to the density map includes: acquiring a first boundary and a second boundary according to the density map, the density value corresponding to the first boundary belongs to a first threshold, the density value corresponding to the second boundary belongs to a second threshold, and the first threshold is less than the second threshold; extracting the spatial characteristics of the ionizable lipid molecules according to the first boundary and the second boundary, the spatial characteristics including at least one of the following: molecular length, molecular width, aspect ratio, cone angle cotangent value of the molecular head group, and symmetry of the molecular head group.

[0068] The density matrix can be first obtained based on the density map, and then the first boundary and the second boundary can be determined based on the density matrix. The first boundary S can be a boundary composed of grid points with density values ​​of 0.1 to 0.15 in the density matrix, which is the outer contour of the molecule. The second boundary S i It can be the boundary composed of density values ​​of 0.4 to 0.45 in the density matrix, which is the internal contour of the molecule.

[0069] In the specific implementation, when performing feature extraction, spatial features can be extracted from the density maps of the XZ and YZ planes respectively. Figure 4 , an example of a specific method for extracting spatial features from a density map in the YZ plane. The molecular length includes the length L of the line segment separated by the boundary S of the first preset straight line in the density map and the length L of the line segment separated by the boundary S i The length of the partition L i , the first preset straight line may be Y=3nm. Wherein, L is completely identical in the XZ and YZ plane projections.

[0070] Taking the YZ plane projection as an example, the molecular width includes the length W of the line segment separated by the boundary S of the second preset straight line in the density map. y and bounded by S i The length of the separator W yi , the second preset straight line can be Z=2.25nm. Then the corresponding aspect ratios R and R can be obtained based on the molecular length and width i , R y =L / W y , R yi =L i / W yi .

[0071] Taking YZ plane projection as an example, when obtaining the cone angle cotangent value, Figure 4 (a) in the figure indicates the cotangent value of the cone angle with the boundary S. That is, the cotangent value A of the cone angle on the YZ plane y1 Refers to the cotangent value of ∠OH-Q1, that is, A y1=cot(∠OH-Q1), where H can be the intersection of the first preset straight line and the boundary S, Q1 and Q2 are the intersections of the third preset straight line (e.g., straight line Y=2.76 nm) and the fourth preset straight line (e.g., straight line Y=3.24 nm) and the boundary S. Another cone angle cotangent value A of the head group y2 Refers to the cotangent value of ∠OH-Q2, that is, A y2 =cot(∠OH-Q2). Similarly, Figure 4 (b) is used to indicate the acquisition and boundary S i The cone angle cotangent value. That is, the cone angle cotangent value A in the Y plane y1i Refers to ∠OH-Q1 i The cotangent value of A y1i =cot(∠OH-Q1 i ), where H can be the first preset straight line and the boundary S i Q1 and Q2 are the intersection points of the third preset straight line (for example, straight line Y=2.76 nm) and the fourth preset straight line (for example, straight line Y=3.24 nm) with the boundary S i The other cone angle cotangent value of the head base is A y2i Refers to ∠OH-Q2 i The cotangent value of A y2i =cot(∠OH-Q2 i ). In particular, A1>A2, then similarly A 1i >A 2i The symmetry of the molecular head group is |A1-A2| and |A 1i -A 2i ∣.

[0072] It can be seen that in this embodiment, the accuracy of the extracted spatial features can be improved by extracting the spatial features based on the acquired density map.

[0073] In a possible embodiment, after training the constructed prediction model according to the training set to obtain the target prediction model, the method further includes: obtaining the target spatial characteristics and target chemical structure characteristics of the target ionizable lipid molecule to be predicted; inputting the target spatial characteristics and the target chemical structure characteristics into the target prediction model to obtain a prediction result, and the prediction result is used to indicate the transfection effect of lipid nanoparticles for carrying mRNA constructed based on the target ionizable lipid molecules.

[0074] Among them, the target spatial characteristics and target chemical structure characteristics can be the spatial characteristics and chemical structure characteristics described above.

[0075] See below Figure 5 , verify the target prediction model constructed in this application.

[0076] Six ionizable lipids with different structures were randomly selected and their chemical structures (two-dimensional structures) were input into the prediction model. Figure 5 As shown in the output, the performance of ionizable lipids P1-P5 is "highly efficient" and they are predicted to be candidate lipid molecules for mRNA delivery. In contrast, the performance of P6 is classified as an "inefficient" molecule. To verify these predictions, these six ionizable lipid molecules were synthesized and purified, and used to prepare LNPs loaded with Luciferase mRNA. Figure 5 The results of in vitro and in vivo tests showed that their mRNA delivery efficiency was consistent with the prediction of the machine learning model, with lipids P1 to P4 having high mRNA transfection efficiency and lipid P6 having low transfection efficiency.

[0077] It can be seen that the above model significantly improves the stability and accuracy of prediction, and provides an efficient tool for the design and optimization of ionizable lipid molecules. When using this model in the future, it is only necessary to obtain the spatial characteristics and chemical structure characteristics of the ionizable lipid molecules that need to predict the transfection effect. By inputting the relevant characteristics into the final model, the predicted transfection effect can be calculated and whether it is an "efficient" molecule can be determined.

[0078] A prediction model generation device provided by the present application is described below. The prediction model generation device described below corresponds to the prediction model generation method described above.

[0079] See also Figure 6 The ionizable lipid molecule prediction model generation device 600 includes: a first acquisition unit 601, used to acquire the initial features of the ionizable lipid molecules; a determination unit 602, used to determine the target features according to the initial features, wherein the target features are high-dimensional combinations of the initial features; a construction unit 603, used to construct a prediction model according to the target features, wherein the prediction model is used to predict the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA; a second acquisition unit 604, used to acquire a training set; and a training unit 605, used to train the constructed prediction model according to the training set to obtain a target prediction model.

[0080] In a possible embodiment, the initial features include multiple chemical structure features and multiple spatial features. In determining the target features based on the initial features, the determination unit 602 is specifically used to: combine the initial features through mathematical operations to obtain multiple candidate features; recursively select features whose correlation with the transfection effect is higher than a preset value from the candidate features; combine the features whose correlation is higher than the preset value to obtain multiple feature combinations; and determine the candidate feature included in the feature combination with the best prediction effect among the multiple feature combinations as the target feature.

[0081] In a possible embodiment, in terms of training the constructed prediction model according to the training set to obtain a target prediction model, the training unit 605 is specifically used to: select multiple experimental data from the training set as sub-training sets to obtain multiple sub-training sets; train the constructed prediction model according to the multiple sub-training sets respectively to obtain multiple trained prediction models; obtain a test set, which includes multiple experimental data; obtain the actual transfection effect of each experimental data in the multiple experimental data included in the test set; obtain the predicted transfection effect of each experimental data predicted by the trained prediction model; and select a target prediction model from the multiple trained prediction models according to the actual transfection effect and the predicted conversion effect.

[0082] In a possible embodiment, in terms of selecting a target prediction model from the multiple trained prediction models according to the actual transfection effect and the predicted conversion effect, the training unit 605 is specifically used to: determine the prediction type of each trained prediction model for each experimental data according to the actual transfection effect and the predicted transfection effect; determine the accuracy, precision and recall of each trained prediction model according to the prediction type; score each trained prediction model according to the accuracy and the recall to obtain a target score; and select a target prediction model from the multiple trained prediction models according to the accuracy, the precision, the recall and the target score.

[0083] In a possible embodiment, the prediction types include four types, wherein the first prediction type is used to indicate that the actual transfection effect is the first effect, and the predicted transfection effect is the first effect, the second prediction type is used to indicate that the actual transfection effect is the first effect, and the predicted transfection effect is the second effect, the third prediction type is used to indicate that the actual transfection effect is the second effect, and the predicted transfection effect is the first effect, and the fourth prediction type is used to indicate that the actual transfection effect is the second effect, and the predicted transfection effect is the second effect, the first effect is better than the second effect, and in terms of determining the accuracy, precision and recall rate of each trained prediction model according to the prediction type of each experimental data, the training The training unit 605 is specifically used to perform the following operations for each trained prediction model: determine the accuracy rate according to the number of the first prediction type and the fourth prediction type corresponding to the currently trained prediction model, and the total number of prediction types corresponding to the currently trained prediction model; determine the precision rate according to the number of the first prediction type corresponding to the currently trained prediction model, and the total number of the first prediction type and the third prediction type corresponding to the currently trained prediction model; determine the recall rate according to the number of the first prediction type corresponding to the currently trained prediction model, and the total number of the first prediction type and the second prediction type corresponding to the currently trained prediction model.

[0084] In a possible embodiment, the initial characteristics of the sample include spatial characteristics. In terms of obtaining the initial characteristics of the ionizable lipid molecules, the first acquisition unit 601 is specifically used to: obtain multiple molecular conformations of the ionizable lipid molecules; align the molecular structures of the multiple molecular conformations respectively; perform density projection on the multiple molecular structures after the alignment to obtain a density map; and extract the spatial characteristics of the ionizable lipid molecules based on the density map.

[0085] In a possible embodiment, in the respectively aligning processing of the molecular structures of the multiple molecular conformations, the first acquisition unit 601 is specifically used to: perform the following operations for the molecular structure of each molecular conformation in the multiple molecular conformations: obtain a target central atom and two reference atoms from the head group of the current molecular structure; fix the target central atom at the target coordinates; obtain a first vector based on the target central atom and the molecular center of mass of the current molecular structure; obtain a second vector based on the two reference atoms; rotate the current molecular structure so that the first vector is parallel to the first coordinate axis, and the second vector is parallel to the plane corresponding to the first coordinate axis and the second coordinate axis.

[0086] In a possible embodiment, in terms of extracting the spatial characteristics of the ionizable lipid molecules based on the density map, the first acquisition unit 601 is specifically used to: acquire a first boundary and a second boundary based on the density map, the density value corresponding to the first boundary belongs to a first threshold, the density value corresponding to the second boundary belongs to a second threshold, and the first threshold is less than the second threshold; extract the spatial characteristics of the ionizable lipid molecules based on the first boundary and the second boundary, the spatial characteristics including at least one of the following: molecular length, molecular width, aspect ratio, cone angle cotangent value of the molecular head group, and symmetry of the molecular head group.

[0087] In a possible embodiment, the prediction model generating device 600 also includes a prediction unit. After the prediction model constructed according to the training set is trained to obtain the target prediction model, the prediction unit is used to: obtain the target spatial characteristics and target chemical structure characteristics of the target ionizable lipid molecule to be predicted; input the target spatial characteristics and the target chemical structure characteristics into the target prediction model to obtain a prediction result, and the prediction result is used to indicate the transfection effect of lipid nanoparticles for carrying mRNA constructed based on the target ionizable lipid molecules.

[0088] See also Figure 7 , Figure 7 Schematic diagram of the structure of the electronic device provided by this application. Figure 7 As shown, the electronic device may include: a processor 710, a communication interface 720, a memory 730 and a communication bus 740, wherein the processor 710, the communication interface 720 and the memory 730 communicate with each other through the communication bus 740. The processor 710 may call the logic instructions in the memory 730 to execute the method for generating a prediction model of ionizable lipid molecules, the method comprising: obtaining the initial features of the ionizable lipid molecules; determining the target features according to the initial features, the target features being a high-dimensional combination of the initial features; constructing a prediction model according to the target features, the prediction model being used to predict the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA; obtaining a training set; and training the constructed prediction model according to the training set to obtain a target prediction model.

[0089] In addition, the logic instructions in the above-mentioned memory 730 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0090] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the method for generating a prediction model of ionizable lipid molecules provided by the above-mentioned methods, the method comprising: obtaining initial characteristics of ionizable lipid molecules; determining target characteristics based on the initial characteristics, wherein the target characteristics are a high-dimensional combination of the initial characteristics; constructing a prediction model based on the target characteristics, wherein the prediction model is used to predict the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA; obtaining a training set; and training the constructed prediction model based on the training set to obtain a target prediction model.

[0091] On the other hand, the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for generating a prediction model for ionizable lipid molecules, the method comprising: obtaining initial features of ionizable lipid molecules; determining target features based on the initial features, wherein the target features are high-dimensional combinations of the initial features; constructing a prediction model based on the target features, wherein the prediction model is used to predict the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA; obtaining a training set; and training the constructed prediction model based on the training set to obtain a target prediction model.

[0092] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0093] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating a prediction model of ionizable lipid molecules, characterized in that: include: Obtaining initial characteristics of ionizable lipid molecules; Determine a target feature according to the initial feature, wherein the target feature is a high-dimensional combination of the initial features; Constructing a prediction model according to the target feature, wherein the prediction model is used to predict the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA; Get the training set; The constructed prediction model is trained according to the training set to obtain a target prediction model.

2. The method according to claim 1, characterized in that: The initial features include multiple chemical structure features and multiple spatial features, and determining the target features according to the initial features includes: Combining the initial features through mathematical operations to obtain multiple candidate features; Recursively select features from the candidate features, the features having a correlation with the transfection effect higher than a preset value; Combining the features whose correlation is higher than a preset value to obtain a plurality of feature combinations; Determine the candidate feature included in the feature combination with the best corresponding prediction effect among the multiple feature combinations as the target feature.

3. The method according to claim 1, characterized in that: The constructed prediction model is trained according to the training set to obtain a target prediction model, including: Selecting multiple experimental data from the training set as sub-training sets to obtain multiple sub-training sets; Training the constructed prediction models according to the multiple sub-training sets respectively to obtain multiple trained prediction models; Acquire a test set, wherein the test set includes a plurality of experimental data; Obtaining an actual transfection effect of each experimental data in the plurality of experimental data included in the test set; Obtaining the predicted transfection effect of each experimental data predicted by the trained prediction model; A target prediction model is selected from the plurality of trained prediction models according to the actual transfection effect and the predicted conversion effect.

4. The method according to claim 3, characterized in that The step of selecting a target prediction model from the plurality of trained prediction models according to the actual transfection effect and the predicted conversion effect comprises: Determining the prediction type of each trained prediction model for each experimental data according to the actual transfection effect and the predicted transfection effect; Determining the accuracy, precision and recall of each trained prediction model according to the prediction type; Scoring each trained prediction model according to the accuracy and the recall to obtain a target score; A target prediction model is selected from the multiple trained prediction models according to the accuracy, the precision, the recall and the target score.

5. The method according to claim 4, characterized in that The prediction types include four types: the first prediction type is used to indicate that the actual transfection effect is the first effect, and the predicted transfection effect is the first effect; the second prediction type is used to indicate that the actual transfection effect is the first effect, and the predicted transfection effect is the second effect; the third prediction type is used to indicate that the actual transfection effect is the second effect, and the predicted transfection effect is the first effect; the fourth prediction type is used to indicate that the actual transfection effect is the second effect, and the predicted transfection effect is the second effect; the first effect is better than the second effect; Determining the accuracy, precision and recall of each trained prediction model according to the prediction type of each experimental data includes: For each trained prediction model, the following operations are performed: Determining the accuracy rate according to the number of the first prediction type and the fourth prediction type corresponding to the currently trained prediction model and the total number of prediction types corresponding to the currently trained prediction model; Determine the precision rate according to the number of the first prediction type corresponding to the currently trained prediction model and the total number of the first prediction type and the third prediction type corresponding to the currently trained prediction model; The recall rate is determined according to the number of first prediction types corresponding to the currently trained prediction model and the total number of first prediction types and second prediction types corresponding to the currently trained prediction model.

6. The method according to any one of claims 1 to 5, characterized in that: The initial features include spatial features, and the initial features of the ionizable lipid molecules are obtained, including: acquiring a plurality of molecular conformations of the ionizable lipid molecule; Performing alignment processing on the molecular structures of the multiple molecular conformations respectively; Density projection is performed on the multiple molecular structures after alignment to obtain a density map; The spatial features of the ionizable lipid molecules are extracted according to the density map.

7. The method according to claim 6, characterized in that The aligning of the molecular structures of the plurality of molecular conformations respectively comprises: The following operations are performed for the molecular structure of each molecular conformation in the plurality of molecular conformations: Get a target central atom and two reference atoms from the head group of the current molecular structure; Fixing the target central atom at a target coordinate; Acquire a first vector according to the target central atom and the molecular mass center of the current molecular structure; Obtaining a second vector based on the two reference atoms; The current molecular structure is rotated so that the first vector is parallel to the first coordinate axis, and the second vector is parallel to the plane corresponding to the first coordinate axis and the second coordinate axis.

8. The method according to claim 6, characterized in that The step of extracting the spatial characteristics of the ionizable lipid molecules according to the density map comprises: Acquire a first boundary and a second boundary according to the density map, wherein a density value corresponding to the first boundary belongs to a first threshold, a density value corresponding to the second boundary belongs to a second threshold, and the first threshold is less than the second threshold; The spatial features of the ionizable lipid molecules are extracted according to the first boundary and the second boundary, and the spatial features include at least one of the following: Molecular length, molecular width, aspect ratio, cotangent of the cone angle of the molecular head group and the symmetry of the molecular head group.

9. The method according to claim 1, characterized in that: After training the constructed prediction model according to the training set to obtain the target prediction model, the method further includes: Obtaining target spatial features and target chemical structure features of the target ionizable lipid molecule to be predicted; The target spatial features and the target chemical structure features are input into the target prediction model to obtain a prediction result, which is used to indicate the transfection effect of lipid nanoparticles for carrying mRNA constructed based on the target ionizable lipid molecules.

10. A device for generating a prediction model of ionizable lipid molecules, characterized in that: include: A first acquisition unit, used to acquire initial characteristics of ionizable lipid molecules; A determination unit, configured to determine a target feature according to the initial feature, wherein the target feature is a high-dimensional combination of the initial features; A construction unit, for constructing a prediction model according to the target feature, wherein the prediction model is used to predict the transfection effect of the ionizable lipid molecules on lipid nanoparticles for carrying mRNA; A second acquisition unit, used for acquiring a training set; The training unit is used to train the constructed prediction model according to the training set to obtain a target prediction model.

Images