A molecular structure design method for coal flotation collector based on 3D-QSAR model

Abstract

The present invention discloses a method for designing a coal flotation collector molecular structure based on a 3D-QSAR model, comprising the following steps: (1) conducting a fine-grained coal slime flotation test to obtain a clean coal combustible recovery rate, and constructing a 3D-QSAR model reflecting the relationship between the collector structure and the flotation performance; (2) using Sybyl-X molecular simulation software to divide the collector molecules in step (1) into a training set and a test set according to a 3:1 number ratio, and using the rigid structure of the template molecule as the parent core structure to perform molecular superposition on the collector molecules in the training set; (3) constructing a 3D-QSAR model; (4) verifying the consistency and predictability of the 3D-QSAR model; (5) determining an optimal 3D-QSAR model; (6) determining a plurality of collector molecular structures to be selected from a ZINC database according to the parent core structure; and (7) completing the sorting of the agent performance according to the activity prediction value, and determining the optimal collector molecular structure, without the need to construct coal molecules, thereby simplifying the steps and shortening the research cycle.

Description

Technical Field

[0001] The invention relates to the technical field of coal flotation, and in particular to a method for designing the molecular structure of a coal flotation collector based on a 3D-QSAR model. Background Art

[0002] Flotation uses the difference in wettability of mineral surfaces to achieve the separation and enrichment of target minerals, which is one of the effective methods for treating fine-grained coal slime in coal preparation plants. Flotation reagents can regulate the hydrophilicity and hydrophobicity of mineral surfaces in a directional manner to achieve the purpose of strengthening the flotation process. The collectors widely used in flotation sites are mainly non-polar hydrocarbon oils. With the deepening of coal resource utilization, the properties of flotation feed are becoming increasingly complex. Non-polar hydrocarbon oils have low flotation efficiency and serious reagent waste. In order to develop efficient collectors, researchers introduce polar groups or add surfactants to non-polar hydrocarbon oils to obtain better flotation indicators. However, it should be pointed out that the introduction of polar groups makes the cost of flotation reagents relatively high, and it is easy to increase the recovery of gangue minerals and deteriorate selectivity. The coal quality of coal slimes in different coal preparation plants is often different, and they show different adaptability to the same collector. This has changed the development model of coal slime flotation collectors from traditional empirical trials to precise design. It is urgent to develop directional design technology for flotation collectors suitable for different difficult-to-float coals based on the physical and chemical properties of coal surfaces.

[0003] Drawing on the quantitative structure-activity relationship method in drug design, by explaining the relationship between the activity and structure of the collector molecule, efficient collector molecules can be designed. Quantitative structure-activity relationship uses theoretical calculations and statistical analysis tools to study the relationship between the structure of a compound and its drug activity. It is mainly divided into two-dimensional quantitative structure-activity relationship (2D-QSAR) and three-dimensional quantitative structure-activity relationship (3D-QSAR). Since the 1980s, three-dimensional quantitative structure-activity relationship has gradually replaced the status of two-dimensional quantitative structure-activity relationship and become one of the main methods for rational drug design based on mechanism. Compared with 2D-QSAR, 3D-QSAR can better reflect the non-bonded interaction mode in the process of the interaction between small drug molecules and their macromolecular targets. Therefore, 3D-QSAR contains more information and clearer physical meaning.

[0004] According to the different molecular superposition methods in the process of building quantitative structure-activity relationship models, 3D-QSAR models are divided into ligand-based 3D-QSAR models and receptor-based 3D-QSAR models. The establishment of receptor-based models relies on the structural analysis of 3D receptor macromolecules, which has many disadvantages such as inaccurate receptor molecule construction and long time consumption. Summary of the invention

[0005] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to provide a method for designing the molecular structure of coal flotation collector based on a 3D-QSAR model.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0007] The present invention provides a method for designing the molecular structure of a coal flotation collector based on a 3D-QSAR model, comprising the following steps:

[0008] (1) A series of agents with the same parent core structure were selected from the drug compound database to conduct fine coal slime flotation tests, and the flotation products were analyzed and processed to obtain the recovery rate of the combustible matter of the clean coal. The collector molecule with the highest recovery rate of the combustible matter of the flotation clean coal was used as a template molecule to construct a 3D-QSAR model reflecting the relationship between the collector structure and the flotation performance;

[0009] (2) using Sybyl-X molecular simulation software, the collector molecules in step (1) are divided into a training set and a test set according to a number ratio of 3:1, and the rigid structure of the template molecule is used as the parent core structure to perform molecular superposition on the collector molecules in the training set;

[0010] (3) The superimposed molecular clusters are placed in a three-dimensional grid and the structural information is collected using a probe. The flotation combustible recovery rate is used as the performance indicator, and a 3D-QSAR model is constructed using the partial least squares method;

[0011] (4) Verify the consistency and predictiveness of the 3D-QSAR model established in step (3);

[0012] (5) Use the test set verified in step (4) to perform external prediction ability test on multiple established models and obtain the prediction correlation coefficient r 2 pred , used to test the predictive ability of the model; the cross-validation coefficient q 2 and multiple correlation coefficient r 2 As an indicator, determine the optimal 3D-QSAR model;

[0013] (6) based on the template molecule of step (1), a three-dimensional isopotential map of the 3D-QSAR model is drawn, and according to the molecular structure modification hint provided by the three-dimensional isopotential map, the molecular structure design idea is analyzed, and multiple collector molecular structures to be selected are determined from the ZINC database according to the parent core structure;

[0014] (7) Using the 3D-QSAR model determined in step (5) to predict the activity of the collector molecules to be selected in step (6), ranking the agent performance according to the activity prediction values, and determining the optimal collector molecular structure.

[0015] Preferably, in step (1), when conducting the fine coal slime flotation test, the collector is selected based on its structural characteristics and the drug compound database of Lipinski's "Five Principles of Drug-like Agents".

[0016] Preferably, step (1) uses the following formula to calculate the recovery rate of clean coal combustibles:

[0017]

[0018] In the formula, γ j Indicates the clean coal yield, A j Ay represents the ash content of flotation clean coal, and Ay represents the ash content of flotation tailings.

[0019] Preferably, in step (2), the Tripos force field is used for geometry optimization before molecular superposition, and the energy is minimized and conformation search is performed by the conjugation method and the Powell energy gradient method to obtain the best stable conformation of the active molecule.

[0020] Preferably, in step (4), when verifying the consistency and predictability of the model, the main component value affecting the molecular structure of the collector is first set to 10, and the leave-one-out method is selected for cross-validation to obtain the cross-validation coefficient q 2 And the best principal component value, then reset the principal component value, use the non-cross-validation method to test the overall significance of the model, and obtain the model's multiple correlation coefficient r 2 , standard deviation SEE, Fischer detection value F and the contribution of different molecular fields to the model are used to evaluate the robustness, reliability and predictiveness of the constructed model.

[0021] Preferably, in step (4), when verifying the consistency and predictability of the model, the closer the r value is to 1, the smaller the SEE value is and the larger the F value is, the closer the predicted value of the collector performance is to the experimental value, and the better the fitting ability of the model is; the calculation formulas of the three indicators are as follows:

[0022]

[0023]

[0024]

[0025] Among them, Y pred , Y obs , and Y mean They represent the calculated value, experimental value and average value of the target attribute respectively, n is the number of samples, and k is the number of variables.

[0026] Preferably, in step (5), q 2 The calculation formula is as follows:

[0027]

[0028] Where Y obs , Ypre , Y mean They represent the experimental value, predicted value, and average experimental value of the target attribute in the training set respectively;

[0029] r 2 pred The calculation formula is as follows:

[0030]

[0031] Among them, PRESS is the residual sum of squares between the predicted activity value and the experimental activity value of each collector in the test set; SD is the residual sum of squares between the predicted activity value of each collector in the test set and the average experimental activity of each collector in the training set.

[0032] Preferably, in step (5), when determining the optimal 3D-QSAR model, when q 2 When it is greater than 0.5, it indicates that the statistical significance of the model is very obvious; the non-cross-validation coefficient r 2 When it is greater than 0.9, it indicates that the model itself has good consistency.

[0033] The beneficial effects of the present invention are:

[0034] 1. The present invention characterizes the relationship between the flotation index and the three-dimensional structure of the reagent through COMFA isopotential diagram and COMSIA isopotential diagram, which can more vividly analyze the influence of the reagent structure on the flotation result and provide a scientific basis for the design and development of coal flotation reagents.

[0035] 2. The present invention obtains the main way to optimize the reagent structure by analyzing the potential energy diagram of the collector molecules, avoiding the huge workload caused by blindly conducting collector tests during flotation collector screening and design, and can greatly improve the flotation test results through a small number of flotation tests in a short time.

[0036] 3. The present invention adopts ligand-based drug design. Compared with receptor-based drug design, it does not need to construct coal molecules, simplifies the steps, shortens the research cycle, and is not affected by problems such as unrealistic coal molecule construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only 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.

[0038] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0040] like Figure 1 As shown, this embodiment provides a method for designing the molecular structure of a coal flotation collector based on a 3D-QSAR model, taking the flotation combustible recovery rate as an indicator for evaluating the performance of the collector, and comprising the following steps:

[0041] Step 1: According to the structural characteristics of the collector and the "five principles of drug-like" proposed by Lipinski, a series of agents with the same parent core structure are selected from the drug compound database to conduct fine coal slime flotation tests, and the flotation products are analyzed and processed to obtain the recovery rate of the combustible body of the clean coal. The collector molecule with the highest recovery rate of the combustible body of the flotation clean coal is used as a template molecule to construct a 3D-QSAR model that reflects the relationship between the collector structure and the flotation performance;

[0042] Lipinski's five drug-like principles are: molecular weight less than 500; number of hydrogen bond donors less than 5; number of hydrogen bond acceptors less than 10; lipid-water partition coefficient value less than 5; number of rotatable bonds less than 7;

[0043] In the preparation stage of the coal slime flotation experiment, the coal slime to be floated and upgraded is pre-screened, and the screened materials are crushed, ground, inspected and screened to ensure that the particle size meets the flotation feed requirements. The coal sample is placed in an oven for low-temperature drying for 24 hours, and then mixed and bagged. During the flotation test, the coal sample is added to the flotation machine, slurried for 2 minutes, and then the collector and frother are added, slurried for 2 minutes and 30 seconds respectively, followed by flotation. The floated products are filtered, dried, and burned. The recovery rate of the clean coal combustible body is calculated using the following formula:

[0044]

[0045] In the formula, γ j Indicates the clean coal yield, A j represents the ash content of flotation clean coal, and Ay represents the ash content of flotation tailings;

[0046] Step 2: using Sybyl-X molecular simulation software, the collector molecules in step 1 are divided into a training set and a test set according to a quantity ratio of 3:1, and the rigid structure of the template molecule is used as the parent core structure to perform molecular superposition on the collector molecules in the training set;

[0047] Before the collector molecules are superimposed, the Tripos force field is used for geometry optimization, and the energy is minimized and the conformation search is performed through the conjugation method and the Powell energy gradient method to obtain the best stable conformation of the active molecule;

[0048] The collector molecules have the same hydrocarbon core structure and the 3D conformation of the receptor is unknown, so a common skeleton superposition method is used. The common skeleton superposition method can ensure that the orientation of the molecular force field of the collector molecular structure in the three-dimensional grid is consistent;

[0049] Step 3: Place the superimposed molecular clusters into a three-dimensional grid and use a probe to collect structural information. Take the flotation combustible recovery rate as the performance indicator and construct a 3D-QSAR model, namely, the COMFA model and the COMSIA model, by partial least squares method;

[0050] The constructed COMFA model is the sum of one or both of the steric field and the electrostatic field, and the COMSIA model is the sum of one or more of the steric field, the electrostatic field, the hydrophobic field, the hydrogen bond donor field, and the hydrogen bond acceptor field;

[0051] In the COMFA model, the steric field and electrostatic field are calculated by Lennard-Jones and Coulombic potential functions, respectively. The calculation process is as follows: the van der Waals radius of each force field is set to 0.152nm, the superimposed molecular clusters are placed in a three-dimensional grid with a grid margin of 1nm×1nm×1nm, and a sp3-hybridized C atom with a positive charge is used as a probe to detect the size and distribution of each molecular field on the grid points around the superimposed molecules. The thresholds (energy cutoffs) of the electrostatic field and steric field are set to 30kcal / mol, the grid step size is 0.02nm, and the attenuation factor α is 0.2.

[0052] In the construction of COMSIA model, the hydrophobic parameter, hydrogen bond acceptor parameter and hydrogen bond donor parameter are all +1, and the Gaussian function is used to evaluate the distance between the probe atom and each atom in the collector molecule in the calculation of the molecular field, so there is no need to set a cutoff value. The recovery rate of the flotation combustible body corresponding to the collector is taken as the dependent variable, and the chemical structure or physicochemical parameters of the reagent are taken as the independent variable. The partial least squares method that considers the information of both the independent variable and the dependent variable is used to construct a mathematical regression model;

[0053] Step 4: Verify the consistency and predictability of the 3D-QSAR model established in step 3. First, set the main component value affecting the molecular structure of the collector to 10, select the leave-one-out method for cross-validation, and obtain the cross-validation coefficient q 2 And the best principal component value, then reset the principal component value, use the non-cross-validation method to test the overall significance of the model, and obtain the model's multiple correlation coefficient r 2, standard deviation SEE, Fischer detection value F and the contribution of different molecular fields to the model are used to evaluate the robustness, reliability and predictiveness of the constructed model;

[0054] The fitting correlation coefficient (r), standard deviation (SEE) and Fisher test value (F) were used to evaluate the fitting ability of the 3D-QSAR model. The closer the r value is to 1, the smaller the SEE value is and the larger the F value is, the closer the predicted value of the collector performance is to the experimental value, and the better the fitting ability of the model is. The calculation formulas of these three indicators are as follows:

[0055]

[0056]

[0057]

[0058] Among them, Y pred , Y obs , and Y mean They represent the calculated value, experimental value and average value of the target attribute, n is the number of samples, and k is the number of variables;

[0059] Step 5: Use the test set to test the external prediction ability of multiple established models and obtain the prediction correlation coefficient r 2 pred , used to test the predictive ability of the model; the cross-validation coefficient q 2 and multiple correlation coefficient r 2 As an indicator, determine the optimal 3D-QSAR model;

[0060] The cross-validation correlation coefficient (q 2 , r 2 cv ) and the square of the predicted correlation coefficient (r 2 pred ) to evaluate the predictive ability of the 3D-QSAR model, where q 2 and r 2 refers to the internal predictive ability of the model, r 2 pred It refers to the use of the test set to test the external prediction ability of the constructed model. The calculation formula is as follows:

[0061]

[0062] Where Y obs , Y pre , Y mean They represent the experimental value, predicted value, and average of the experimental value of the target attribute in the training set, respectively.

[0063]

[0064] Among them, PRESS (predicted residual sum of squares) is the residual sum of squares between the predicted activity value and the experimental activity value of each collector in the test set. SD (squared deviation) is the residual sum of squares between the predicted activity value of each collector in the test set and the average experimental activity of each collector in the training set;

[0065] The best 3D-QSAR model was selected by the cross-validation coefficient q2 and the non-cross-validation coefficient r2. When q2>0.5, it indicated that the statistical significance of the model was very obvious. When the non-cross-validation coefficient r2 was greater than 0.9, it indicated that the model itself had good consistency.

[0066] Step 6: Based on the template molecule, a three-dimensional isopotential map of the 3D-QSAR model is drawn. According to the molecular structure modification tips provided by the three-dimensional isopotential map, the molecular structure design ideas are analyzed, and multiple collector molecular structures to be selected are determined from the ZINC database according to the parent core structure;

[0067] Step 7: Use the determined 3D-QSAR model to predict the activity of the selected collector molecules, rank the agent performance according to the activity prediction values, and determine the optimal collector molecular structure.

[0068] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for designing the molecular structure of coal flotation collector based on 3D-QSAR model, characterized in that: The following steps are involved: (1) Select reagents with the same parent core structure to conduct coal slime flotation tests, analyze the flotation products, and use the collector with the highest recovery rate of combustible matter in flotation clean coal as a template molecule to construct a 3D-QSAR model that reflects the relationship between collector structure and performance; (2) Using molecular simulation software, the collectors in step (1) are divided into a training set and a test set according to a ratio of 3:1, and the rigid structure of the template molecule is used as the parent core structure to perform molecular superposition on the collector molecules in the training set; (3) The superimposed molecular clusters are placed in a three-dimensional grid and the structural information is collected using a probe. The combustible recovery rate is used as a performance indicator, and a 3D-QSAR model is constructed using the partial least squares method; (4) Verify the consistency and predictiveness of the 3D-QSAR model established in step (3); (5) Use the test set verified in step (4) to conduct external prediction test on multiple models to test the prediction ability of the models; use the cross-validation coefficient q 2 and multiple correlation coefficient r 2 As an indicator, determine the optimal 3D-QSAR model; (6) Based on the template molecule of step (1), draw a three-dimensional isopotential diagram of the 3D-QSAR model, and determine the molecular structure to be selected from the collector database according to the molecular structure modification prompts in the diagram and the parent core structure; (7) The 3D-QSAR model determined in step (5) is used to predict the activity of the collector molecules to be selected in step (6), and the performance of the agents is ranked according to the predicted activity values, thereby determining the optimal collector molecular structure, i.e., the collector molecular structure with the largest predicted activity value.

2. A method for designing the molecular structure of a coal flotation collector based on a 3D-QSAR model as claimed in claim 1, characterized in that: Step (1) When conducting the flotation test of fine coal slime, the collector should be selected based on its structural characteristics and Lipinski's "five principles of drug-likeness" and the drug compound database.

3. The method for designing the molecular structure of coal flotation collector based on 3D-QSAR model according to claim 1, characterized in that: Step (1) uses the following formula to calculate the recovery rate of clean coal combustibles: In the formula, γ j represents the clean coal yield, A j Indicates the ash content of flotation clean coal, Ay Indicates the ash content of flotation tailings.

4. A method for designing the molecular structure of a coal flotation collector based on a 3D-QSAR model as claimed in claim 1, characterized in that: Step (2) Before molecular alignment, the Tripos force field is used to perform geometry optimization, and the energy is minimized and a conformation search is performed using the conjugation method and the Powell energy gradient method to obtain the optimal stable conformation of the active molecule.

5. The method for designing the molecular structure of coal flotation collector based on 3D-QSAR model according to claim 1, characterized in that: In step (4), when verifying the consistency and predictability of the model, the main component value affecting the molecular structure of the collector is first set to 10, and the leave-one-out method is selected for cross-validation to obtain the cross-validation coefficient q 2 And the best principal component value, then reset the principal component value, use the non-cross-validation method to test the overall significance of the model, and obtain the model's multiple correlation coefficient r 2 , standard deviation SEE, Fischer detection value F and the contribution of different molecular fields to the model are used to evaluate the robustness, reliability and predictiveness of the constructed model.

6. A method for designing the molecular structure of a coal flotation collector based on a 3D-QSAR model as claimed in claim 5, characterized in that: Step (4) When verifying the consistency and predictability of the model, the closer the r value is to 1, the smaller the SEE value is and the larger the F value is, the closer the predicted value of the collector performance is to the experimental value, and the better the model fitting ability is. The calculation formulas for the three indicators are as follows: Among them, Y pred , Y obs , and Y mean They represent the calculated value, experimental value and average value of the target attribute respectively, n is the number of samples, and k is the number of variables.

7. The method for designing the molecular structure of coal flotation collector based on 3D-QSAR model according to claim 1, characterized in that: In step (5), 2 The calculation formula is as follows: Where Y obs , Y pre , Y mean They represent the experimental value, predicted value, and average experimental value of the target attribute in the training set respectively; r 2 pred The calculation formula is as follows: Among them, PRESS is the residual sum of squares between the predicted activity value and the experimental activity value of each collector in the test set; SD is the residual sum of squares between the predicted activity value of each collector in the test set and the average experimental activity of each collector in the training set.

8. The method for designing the molecular structure of coal flotation collector based on 3D-QSAR model according to claim 1, characterized in that: Step (5) When determining the optimal 3D-QSAR model, when q 2 When it is greater than 0.5, it indicates that the statistical significance of the model is very obvious; the non-cross-validation coefficient r 2 When it is greater than 0.9, it indicates that the model itself has good consistency.

Images