Method, apparatus, device, and storage medium for predicting liquid chromatography retention time

Abstract

The application provides a liquid chromatography retention time prediction method, device, equipment and storage medium, the method comprises the following steps: obtaining sample data of a to-be-tested small molecule compound, the sample data comprising test condition data required to be used in a preset chromatography test condition and physical and chemical property parameters of the to-be-tested small molecule compound; wherein the small molecule compound is a metabolite; inputting the sample data into a pre-trained retention time prediction model to obtain a predicted retention time of the to-be-tested small molecule compound; wherein the retention time prediction model comprises a support vector regression sub-model and a regression tree sub-model. The application can accurately predict the retention time of a small molecule compound under different liquid chromatography conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biological analysis, and in particular to a liquid chromatography retention time prediction method, device, equipment and storage medium. BACKGROUND

[0002] Metabolomics is a new discipline for qualitative and quantitative analysis of small molecule metabolites in a biological or cellular. In the field of food safety, it is a hot field to find relevant biomarkers of agricultural and veterinary drugs in animals and plants by using metabolomics. The test sample is mainly the extract of cells and tissues of animals and plants.

[0003] The common research method of metabolomics is metabolite fingerprint analysis, that is, using a liquid chromatograph to compare the metabolites in different samples. In essence, metabolite fingerprint analysis compares the retention times of metabolites in different individuals to understand the structures of different compounds, and ultimately establishes a complete analysis method for identifying the characteristics of these different compounds. Retention time refers to the time elapsed from the start of sample injection to the appearance of the peak top of a certain component in chromatography. Under certain chromatographic operating conditions, any small molecule compound has a certain retention time, which can be used as the basis for qualitative analysis.

[0004] However, due to the variability of the structure and method of the liquid chromatograph, the retention time of the same compound obtained under different instruments and different chromatographic conditions is quite different. The deviation between the predicted value and the measured value of the retention time obtained by using the current prediction model is often large, therefore, there is an urgent need for a liquid chromatography retention time prediction method for small molecule compounds, so that the retention time can be accurately predicted under various different liquid chromatography conditions. SUMMARY

[0005] The embodiments of the present application provide a liquid chromatography retention time prediction method, device, equipment and storage medium to solve the problem that the predicted retention time obtained by the current retention time prediction model has a large deviation from the measured value.

[0006] In a first aspect, the embodiments of the present application provide a liquid chromatography retention time prediction method, comprising:

[0007] Obtaining sample data of a small molecule compound to be tested, the sample data including test condition data required to be used in a preset chromatography test condition and physical and chemical property parameters of the small molecule compound to be tested; wherein the small molecule compound is a metabolite;

[0008] Inputting the sample data into a pre-trained retention time prediction model to obtain the predicted retention time of the small molecule compound to be tested; wherein the retention time prediction model includes a support vector regression sub-model and a regression tree sub-model.

[0009] In a possible implementation, the sample data is input into the pre-trained retention time prediction model to obtain the predicted retention time of the small-molecule compound to be tested, including:

[0010] The sample data is input into the trained support vector regression sub-model to obtain the first predicted retention time;

[0011] The sample data is input into the trained regression tree sub-model to obtain the second predicted retention time;

[0012] The predicted retention time of the small-molecule compound to be tested is determined based on the first predicted retention time and the second predicted retention time.

[0013] In a possible implementation, the predicted retention time of the small-molecule compound to be tested is determined based on the first predicted retention time and the second predicted retention time, including:

[0014] The average of the first predicted retention time and the second predicted retention time is determined as the predicted retention time of the small-molecule compound to be tested.

[0015] In a possible implementation, the kernel function of the support vector regression sub-model is a Gaussian kernel function, and the algorithm used by the regression tree sub-model is a classification and regression tree algorithm.

[0016] In a possible implementation, the sample data is input into the pre-trained retention time prediction model, including:

[0017] The sample data is stored in a comma-separated value file format, and the stored data is standardized to obtain standard sample data;

[0018] The standard sample data is input into the pre-trained retention time prediction model.

[0019] In a possible implementation, the physicochemical property parameters of the small-molecule compound to be tested include: the number of carbon atoms, the number of hydrogen atoms, the number of nitrogen atoms, the number of oxygen atoms, the number of phosphorus atoms, the number of sulfur atoms, the hydrophilicity parameter of the small-molecule compound to be tested, and the lipophilicity parameter of the small-molecule compound to be tested.

[0020] The preset chromatographic test conditions include: using a high-performance liquid chromatograph or an ultra-high-performance liquid chromatograph for testing;

[0021] The test condition data required to be used in the preset chromatographic test conditions include: the instrument name used by the preset chromatographic test conditions, the column name, the column temperature value, the flow rate value, the aqueous phase name, the organic phase name, the elution time, the initial proportion of the organic phase, and the final proportion of the organic phase.

[0022] In a possible implementation, the prediction method further comprises:

[0023] obtaining training sample data, wherein the training sample data comprises data of a standard sample of a small molecule compound determined in a laboratory, data of a small molecule compound identified in a biological sample determined in a laboratory, and data of a small molecule compound collected from a PubChem database;

[0024] training the pre-constructed support vector regression sub-model and the regression tree sub-model respectively based on the training sample data to obtain a trained retention time prediction model.

[0025] In a second aspect, an embodiment of the present application provides a device for predicting liquid chromatography retention time, comprising:

[0026] a data acquisition module configured to acquire sample data of a small molecule compound to be tested, the sample data comprising test condition data required to be used in a preset chromatography test condition and physicochemical property parameters of the small molecule compound to be tested; wherein the small molecule compound is a metabolite;

[0027] a test time module configured to input the sample data into a pre-trained retention time prediction model to obtain a predicted retention time of the small molecule compound to be tested; wherein the retention time prediction model comprises a support vector regression sub-model and a regression tree sub-model.

[0028] In a possible implementation, the test time module is specifically configured to:

[0029] input the sample data into the trained support vector regression sub-model to obtain a first predicted retention time;

[0030] input the sample data into the trained regression tree sub-model to obtain a second predicted retention time;

[0031] determine the predicted retention time of the small molecule compound to be tested based on the first predicted retention time and the second predicted retention time.

[0032] In a possible implementation, the test time module is further configured to:

[0033] determine the average value of the first predicted retention time and the second predicted retention time as the predicted retention time of the small molecule compound to be tested.

[0034] In a possible implementation, the kernel function of the support vector regression sub-model is a Gaussian kernel function, and the algorithm used by the regression tree sub-model is a classification and regression tree algorithm.

[0035] In a possible implementation, the test time module is further configured to:

[0036] The sample data is stored in a comma-separated value file format, and the stored data is standardized to obtain standard sample data.

[0037] The standard sample data is input into a pre-trained retention time prediction model.

[0038] In a possible implementation, the physicochemical property parameters of the small molecule compound to be tested include: the number of carbon atoms, the number of hydrogen atoms, the number of nitrogen atoms, the number of oxygen atoms, the number of phosphorus atoms, the number of sulfur atoms, the hydrophilicity parameter of the small molecule compound to be tested, and the lipophilicity parameter of the small molecule compound to be tested.

[0039] The preset chromatographic test condition includes: using a high-performance liquid chromatograph or an ultra-high-performance liquid chromatograph for testing.

[0040] The test condition data required to be used in the preset chromatographic test condition includes: the instrument name used in the preset chromatographic test condition, the column name, the column temperature value, the flow rate value, the aqueous phase name, the organic phase name, the elution time, the initial proportion of the organic phase, and the final proportion of the organic phase.

[0041] In a possible implementation, the test time module is further configured to:

[0042] The training sample data is obtained, wherein the training sample data includes data of a small molecule compound standard substance measured in a laboratory, data of a small molecule compound identified in a biological sample measured in a laboratory, and data of a small molecule compound collected from a PubChem database.

[0043] Based on the training sample data, the pre-constructed support vector regression sub-model and the regression tree sub-model are trained respectively to obtain the trained retention time prediction model.

[0044] In a third aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the method of the first aspect or any possible implementation manner of the first aspect when executing the computer program.

[0045] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, and the computer program is executable on a processor to implement the steps of the method of the first aspect or any possible implementation manner of the first aspect.

[0046] The embodiment of the present application provides a liquid chromatography retention time prediction method, first, sample data of a to-be-tested small molecule compound is acquired, then, the sample data is input into a pre-trained retention time prediction model, and prediction retention time of the to-be-tested small molecule compound is obtained. Therefore, by directly inputting the sample data into the trained retention time prediction model, the prediction retention time under preset liquid chromatography conditions can be obtained, and the prediction retention time has a small difference with the real test retention time. The prediction retention time under different liquid chromatography conditions can also be obtained by changing test condition data required to be used in the preset chromatography test condition. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0048] Figure 1 is an implementation flowchart of the liquid chromatography retention time prediction method provided by the embodiment of the present application;

[0049] Figure 2A is an experimental test result and prediction result graph tested by using an independent SVR model;

[0050] Figure 2B is an experimental test result and prediction result graph tested by using an independent DTR model;

[0051] Figure 2C is an experimental test result and prediction result graph tested by using an RF model;

[0052] Figure 2D is an experimental test result and prediction result graph tested by using an SVR-DTR integrated model;

[0053] Figure 3A is a box plot of ΔRT tested by using training set data of four models;

[0054] Figure 3B is a box plot of ΔRT tested by using validation set data of four models;

[0055] Figure 3C is a box plot of ΔRT tested by using independent sample data of four models;

[0056] Figure 4 is a structure schematic diagram of the liquid chromatography retention time prediction device provided by the embodiment of the present application;

[0057] Figure 5This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0058] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0059] To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0060] As described in the background section, the retention time of the same small molecule compound is fixed under the same chromatographic conditions. Therefore, the identification result of the small molecule compound to be tested can be obtained by detecting the retention time of the standard and comparing the retention time and mass spectrometry information of the standard and the small molecule compound to be tested. However, in practical applications, due to the high cost and difficulty in preparing standards, most unknown compounds often require comparison and identification using existing databases of small molecule compounds.

[0061] In existing databases, the mass-to-charge ratio of the parent ion and information on secondary fragments of the small molecule compound being analyzed are typically compared with the database to obtain the identification result. However, because the retention times of small molecule compounds vary significantly under different chromatographic conditions, and the predicted and measured retention times obtained by current predictive models often deviate considerably, they cannot accurately characterize the retention times of small molecule compounds under different chromatographic conditions, leading to inaccurate identification results. Therefore, there is an urgent need for a predictive method that can accurately predict the liquid chromatography retention times of small molecule compounds.

[0062] To address the problems of existing technologies, embodiments of the present invention provide a method, apparatus, device, and storage medium for predicting liquid chromatography retention time. The method for predicting liquid chromatography retention time provided by the embodiments of the present invention will be described first below.

[0063] The execution entity of the method for predicting liquid chromatography retention time can be a liquid chromatography retention time prediction device. This prediction device can be an electronic device with a processor and memory, such as a mobile electronic device or a non-mobile electronic device. This invention does not impose specific limitations on the embodiments.

[0064] See Figure 1 The flowchart illustrating the method for predicting liquid chromatography retention time provided in this embodiment of the invention is shown below:

[0065] Step S110, sample data of the small molecule compound to be tested is acquired.

[0066] The sample data includes test condition data required to be used in the preset chromatographic test condition and physicochemical property parameters of the small molecule compound to be tested. It should be noted that the small molecule compound is a metabolite.

[0067] Specifically, the preset chromatographic test condition includes using a high-performance liquid chromatograph or an ultra-high-performance liquid chromatograph for testing. The test condition data required to be used in the preset chromatographic test condition can include: an instrument name used in the preset chromatographic test condition, a column name, a column temperature value, a flow rate value, an aqueous phase name, an organic phase name, an elution time, an initial proportion of the organic phase, and a final proportion of the organic phase. Of course, the test condition data can be appropriately increased or reduced according to an actual use scene, and can be increased or reduced according to an actual use condition, which is not limited herein. The user can limit the test condition to be used according to different chromatographic test conditions.

[0068] The physicochemical property parameters of the small molecule compound to be tested mainly refer to the number of chemical elements contained in the small molecule compound to be tested, a hydrophilicity parameter of the small molecule compound to be tested, and a lipophilicity parameter of the small molecule compound to be tested. Specifically, the physicochemical property parameters of the small molecule compound to be tested can include: the number of carbon atoms, the number of hydrogen atoms, the number of nitrogen atoms, the number of oxygen atoms, the number of phosphorus atoms, the number of sulfur atoms, a hydrophilicity parameter LogS of the small molecule compound to be tested, and a lipophilicity parameter LogP of the small molecule compound to be tested. The hydrophilicity parameter LogS of the small molecule compound to be tested and the lipophilicity parameter LogP of the small molecule compound to be tested can be queried in the HMDB database.

[0069] The sample data is determined by the contribution rates of the parameters. It can be known by calculating the correlation coefficients between the required parameters that the correlation coefficients between the parameters of the sample data are small, the independence between the parameters in the sample data is large, and the accuracy of the prediction result is improved.

[0070] Step S120, the sample data is input into a retention time prediction model trained in advance to obtain a predicted retention time of the small molecule compound to be tested.

[0071] The retention time prediction model includes a support vector regression sub-model and a regression tree sub-model.

[0072] The support vector regression sub-model is a regression algorithm based on a support vector machine, and is used to find a higher-dimensional hyperplane to distinguish different types of liquid chromatography tests, and to perform regression prediction based on the hyperplane to return a floating-point number as a prediction value. For the sake of subsequent description, the support vector regression sub-model is referred to as an SVR sub-model. In the SVR sub-model, a kernel function is crucial to the establishment of the hyperplane. In the present application, the SVR sub-model uses a Gaussian kernel function, and specifically, ‘rbf’ can be used as the kernel function to establish the SVR sub-model.

[0073] The regression tree sub-model is a regression algorithm based on a decision tree, i.e., a classification and regression tree (CART) algorithm. The decision tree calculates the Gini value of each variable in the data set by binary classification, and establishes classification nodes in the order of the Gini value. Then, the original decision tree is pruned and optimized by post-pruning, and finally the final decision tree is generated. For the sake of subsequent description, the regression tree sub-model is referred to as a DTR sub-model. The DTR sub-model performs regression prediction based on the generated decision tree, and returns a floating-point number as a prediction value.

[0074] In some embodiments, the sample data of the to-be-tested small molecule compound obtained needs to be preprocessed first. Specifically, the sample data is stored in a comma-separated value (csv) format file, where the columns are the parameters needed to use the sample data, and the rows are the numerical values of the corresponding parameters.

[0075] After the sample data is stored, the sample data needs to be standardized. The standardization is performed by using mean and variance. The standard sample data is obtained. The specific processing formula is:

[0076]

[0077] where x is the value of a single data, μ is the mean value of the corresponding column, and σ is the standard deviation of the corresponding column.

[0078] Then, the processed standard sample data is input into the pre-trained retention time prediction model.

[0079] In some embodiments, the retention time prediction model has been mentioned above to include the support vector regression sub-model and the regression tree sub-model, and the specific use process of the model is as follows:

[0080] First, the sample data is input into the trained support vector regression sub-model to obtain the first predicted retention time.

[0081] Then, the sample data is input into the trained regression tree sub-model to obtain a second predicted retention time.

[0082] Finally, based on the first predicted retention time and the second predicted retention time, a predicted retention time of the small-molecule compound to be tested is determined.

[0083] Specifically, the average of the first predicted retention time and the second predicted retention time can be determined as the predicted retention time of the small-molecule compound to be tested.

[0084] It should be noted here that the retention time prediction model is an integration of the support vector regression sub-model and the regression tree sub-model, and the integration method is mean voting.

[0085] The retention time prediction model integrated with the support vector regression SVR sub-model and the regression tree DTR sub-model will be described in detail as follows:

[0086] The support vector regression SVR sub-model and the regression tree DTR sub-model are both modeled using the scikit-learn 1.0.1 package in Python 3.8.

[0087] First, the svm.SVR() function in scikit-learn is used to establish the SVR sub-model,

[0088] The parameter settings are as follows: kernel='rbf', degree=3, gamma='scale', coef0=0.0, tol=0.001, C=1.0, epsilon=0.1, shrinking=True, cache_size=200, verbose=False, max_iter=-1.

[0089] Then, the DecisionTreeRegressor() function in scikit-learn is used to establish the DTR sub-model, and the parameter settings are as follows: criterion='squared_error', splitter='best', max_depth=None, min_samples_split=2, min_samples_leaf=1, min_weight_fraction_leaf=0.0, max_features=None, random_state=None, max_leaf_nodes=None, min_impurity_decrease=0.0, ccp_alpha=0.0.

[0090] Finally, the SVR sub-model and the DTR sub-model are combined to build a retention time prediction model using the VotingRegressor() function in scikit-learn. For convenience of subsequent description, the retention time prediction model is referred to as the SVR-DTR model.

[0091] The parameter settings are as follows: estimators = [('gb', svr), ('rf', dtr)], weights = None, n_jobs = None, verbose = False.

[0092] After the retention time prediction model, i.e., the SVR-DTR model, is built, the model can be trained and tested.

[0093] The training and testing process of the retention time prediction model, i.e., the SVR-DTR model, is described below.

[0094] The test sample data comes from the data of small molecule compound standards measured in the laboratory, the data of small molecule compounds identified in biological samples measured in the laboratory, and the data of small molecule compounds collected from the PubChem database. The test sample data includes 12 liquid chromatography conditions, of which 4 are high-performance liquid chromatography conditions. The total number of test samples is 2779 retention time data, which are collected from 206 small molecule compounds after different conditions of liquid chromatography.

[0095] Specifically, the 206 small molecule compounds include 111 independent samples and 95 model samples. It should be noted that an independent sample refers to a small molecule compound sample with only one retention time data collected, and a model sample refers to a small molecule compound sample with more than one retention time data collected, totaling 2668 retention time data. In the model samples, stratified random sampling is used to divide the model samples into a training set and a validation set in a ratio of 7.5:2.5, with 2001 samples in the training set and 667 samples in the validation set. Among the 2779 test sample data, part of the data is collected from the PubChem database, and the other part is based on laboratory-determined small molecule compound data. All the collected test sample data are stored as csv format files, with the columns representing the test parameters of the test sample data and the rows representing the numerical values of the corresponding parameters.

[0096] Since the small molecule compound data collected from the PubChem database may have different degrees of missing values, such as the absence of a certain test condition required in the preset chromatography test conditions, the missing data can be first processed by interpolation. The specific interpolation process is as follows: in the total sample, the average value of the column at the missing value position is calculated, and the calculated average value is inserted into the missing value position.

[0097] After the missing values of the test sample data are supplemented, the data after the missing values are supplemented need to be normalized, that is, mean-variance normalization is adopted.

[0098] After the test sample data are processed by the missing value processing and the normalization processing, the processed independent sample data, the training set data and the validation set data are formed.

[0099] The training set data are input into the retention time prediction model, that is, the SVR-DTR model, which is constructed in advance, and the support vector regression sub-model and the regression tree sub-model in the retention time prediction model are trained respectively to obtain the trained retention time prediction model.

[0100] After the retention time prediction model, that is, the SVR-DTR model is trained, the independent sample data and the validation set data are respectively input into the retention time prediction model, that is, the SVR-DTR model for testing.

[0101] In order to verify the accuracy of the constructed retention time prediction model, that is, the SVR-DTR model, the SVR-DTR model, the single SVR model, the single DTR model and the random forest model, that is, the RF model are used to predict the retention time of the same test sample data, so as to verify the accuracy of the retention time prediction of the retention time prediction model, that is, the SVR-DTR model constructed by the application.

[0102] In the embodiment of the application, the ten-fold cross-validation method is adopted to verify the accuracy of the SVR-DTR model, the single SVR model, the single DTR model and the RF model. The regression prediction model is used as the model for verification, R 2 and ΔRT are used as evaluation indexes, R 2 is a decision coefficient, and ΔRT is an absolute difference value between the experimental retention time and the model predicted retention time.

[0103] In the formula, R 2 is calculated as follows:

[0104]

[0105] In the formula, R 2 , y i is a retention time value of the experimental test, is a retention time value predicted by the model, is an average value of all experimental test retention times. R 2 The closer the R 2 value is to 1, the better the model prediction ability is.

[0106] The calculation formula of ΔRT is as follows:

[0107] ΔRT = |RTexp -RT pre |

[0108] In the formula of ΔRT, RT exp is the retention time obtained by experimental test, and RT pre is the retention time predicted by the model. The smaller the value of ΔRT, the smaller the deviation of the model prediction.

[0109] In the 10-fold cross-validation method, the corresponding R 2 value is obtained in each test. The R 2 average value of the 10 results is used as an estimate of the accuracy of the algorithm.

[0110] The independent sample data, training set data and validation set data in the above test sample data are respectively input into the SVR-DTR model, the single SVR model, the single DTR model and the RF model, and the results of the verification are shown in Table 1A and Table 1B as follows:

[0111] Table 1A

[0112]

[0113] Table 1B

[0114]

[0115] Figures 2A-2D is a comparison chart of the experimental test results and the prediction results of the single SVR model, the single DTR model, the RF model and the SVR-DTR integrated model respectively.

[0116] Figures 3A-3C is a box plot of ΔRT of the training set data, the validation set data and the independent sample data of the four models respectively, wherein, Figures 3A-3C the dashed line in the box plot represents the median of ΔRT, and the triangle represents the mean of ΔRT.

[0117] From Table 1A-1B, Figures 2A-2D and Figures 3A-3C it can be seen that there is little difference between the validation set data and the R 2 value of the 10-fold cross-validation of the same model, and it can be known that the above four models do not have overfitting phenomenon, so the above four models can be used for the next step of analysis.

[0118] In the training set data and the validation set data, the R 2 value of the SVR-DTR integrated model is slightly lower than that of the DTR model and the RF model, and the R 2 value of the SVR model is the lowest. In the independent sample data, the R 2The value is much higher than the other three models. The trend of ΔRT value also follows the same pattern as R. 2 Similarly, in the training and validation sets, the mean and median ΔRT of the SVR-DTR ensemble model are slightly higher than those of the DTR and RF models, but in the independent dataset, the mean and median ΔRT of the SVR-DTR ensemble model are the smallest among the four models.

[0119] It should be noted that in the RF model mentioned above, the number of sub-learners in the RF model is set to 100, but the SVR-DTR ensemble model only uses 1 SVR sub-learner and 1 DTR sub-learner. The number of sub-learners in the SVR-DTR ensemble model is much lower than that in the RF model, and the diversity of sub-learners is higher than that in the RF model.

[0120] Therefore, the results above show that the overall performance of the SVR-DTR ensemble model is superior to that of the SVR model, DTR model, and RF model.

[0121] The prediction method provided in this invention first acquires sample data of the small molecule compound to be tested. Then, the sample data is input into a pre-trained retention time prediction model to obtain the predicted retention time of the small molecule compound. Thus, by directly inputting the sample data into the trained retention time prediction model, the predicted retention time under preset liquid chromatography conditions can be obtained, and the difference between the obtained predicted retention time and the actual test retention time is small. Furthermore, the predicted retention time under different liquid chromatography conditions can be obtained by changing the test condition data required in the preset chromatographic test conditions.

[0122] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0123] Based on the liquid chromatography retention time prediction method provided in the above embodiments, the present invention also provides a specific implementation of a liquid chromatography retention time prediction device applied to the liquid chromatography retention time prediction method. Please refer to the following embodiments.

[0124] like Figure 4 As shown, a liquid chromatography retention time prediction device 400 is provided, the device comprising:

[0125] The data acquisition module 410 is used to acquire sample data of the small molecule compound to be tested. The sample data includes the test condition data required in the preset chromatographic test conditions and the physicochemical property parameters of the small molecule compound to be tested; wherein, the small molecule compound is a metabolite.

[0126] The test time module 420 is configured to input the sample data into the pre-trained retention time prediction model to obtain a predicted retention time of the small molecule compound to be tested.

[0127] In a possible implementation, the test time module 420 is specifically configured to:

[0128] input the sample data into the trained support vector regression sub-model to obtain a first predicted retention time;

[0129] input the sample data into the trained regression tree sub-model to obtain a second predicted retention time;

[0130] determine the predicted retention time of the small molecule compound to be tested based on the first predicted retention time and the second predicted retention time.

[0131] In a possible implementation, the test time module 420 is further configured to:

[0132] determine the average value of the first predicted retention time and the second predicted retention time as the predicted retention time of the small molecule compound to be tested.

[0133] In a possible implementation, the kernel function of the support vector regression sub-model is a Gaussian kernel function, and the algorithm used by the regression tree sub-model is a classification and regression tree algorithm.

[0134] In a possible implementation, the test time module 420 is further configured to:

[0135] store the sample data in a comma-separated value file format, and perform standardization processing on the stored data to obtain standard sample data;

[0136] input the standard sample data into the pre-trained retention time prediction model.

[0137] In a possible implementation, the physicochemical property parameters of the small molecule compound to be tested include: the number of carbon atoms, the number of hydrogen atoms, the number of nitrogen atoms, the number of oxygen atoms, the number of phosphorus atoms, the number of sulfur atoms, the hydrophilicity parameter of the small molecule compound to be tested, and the lipophilicity parameter of the small molecule compound to be tested.

[0138] The preset chromatographic test condition includes: using a high-performance liquid chromatograph or an ultra-high-performance liquid chromatograph for testing;

[0139] The test condition data required to be used in the preset chromatographic test condition includes: an instrument name, a column name, a column temperature value, a flow rate value, an aqueous phase name, an organic phase name, an elution time, an initial proportion of the organic phase, and a final proportion of the organic phase.

[0140] In a possible implementation, the test time module 420 is further configured to:

[0141] obtain training sample data, wherein the training sample data comprises data of a standard sample of a small molecule compound determined by a laboratory, data of a small molecule compound identified in a biological sample determined by a laboratory, and data of a small molecule compound collected from a PubChem database;

[0142] train the pre-constructed support vector regression sub-model and the regression tree sub-model respectively based on the training sample data to obtain a trained retention time prediction model.

[0143] Figure 5 is a schematic diagram of an electronic device provided by an embodiment of the present application. As shown in Figure 5 the electronic device 5 of this embodiment includes a processor 50, a memory 51, and a computer program 52 stored in the memory 51 and executable on the processor 50. The processor 50 implements the steps in each of the above-mentioned liquid chromatography retention time prediction method embodiments when executing the computer program 52, such as Figure 1 steps 110 to 120 shown in the figure. Alternatively, the processor 50 implements the functions of each module in each of the above-mentioned device embodiments when executing the computer program 52, such as Figure 4 the functions of the modules 410 to 420 shown in the figure.

[0144] For example, the computer program 52 can be divided into one or more modules, which are stored in the memory 51 and executed by the processor 50 to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program 52 in the electronic device 5. For example, the computer program 52 can be divided into Figure 4 the modules 410 to 420 shown in the figure.

[0145] The electronic device 5 can include, but is not limited to, the processor 50 and the memory 51. Those skilled in the art can understand that Figure 5 the electronic device 5 is only an example and does not constitute a limitation on the electronic device 5, and can include more or fewer components than those shown, or combine certain components, or different components, for example, the electronic device can also include an input / output device, a network access device, a bus, etc.

[0146] The processor 50 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0147] The memory 51 can be an internal storage unit of the electronic device 5, such as a hard disk or a memory of the electronic device 5. The memory 51 can also be an external storage device of the electronic device 5, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 51 can also include both the internal storage unit and the external storage device of the electronic device 5. The memory 51 is used to store the computer program and other programs and data required by the electronic device. The memory 51 can also be used to temporarily store data that has been output or will be output.

[0148] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to complete all or part of the above described functions. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit, and the integrated unit can be realized in the form of hardware or in the form of software. In addition, the specific names of each functional unit and module are only for easy distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0149] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0150] Those skilled in the art can understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0151] In the embodiments provided by the present application, it should be understood that the disclosed apparatus / equipment and method can be implemented in other ways. For example, the apparatus / equipment embodiments described above are merely schematic. The division of the modules or units is merely a logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling or direct coupling or communication connection between the units can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or in other forms.

[0152] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. they can be located in one place, or distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0153] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0154] The integrated module / unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. The computer program can implement the steps of each of the above-mentioned liquid chromatography retention time prediction method embodiments when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the contents included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0155] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for predicting retention time in liquid chromatography, characterized in that, include: The sample data of the small molecule compound to be tested is obtained. The sample data includes the test condition data required in the preset chromatographic test conditions and the physicochemical property parameters of the small molecule compound to be tested. The small molecule compound is a metabolite. The physicochemical property parameters of the small molecule compound to be tested include: the number of carbon atoms, hydrogen atoms, nitrogen atoms, oxygen atoms, phosphorus atoms, and sulfur atoms of the small molecule compound to be tested; the hydrophilicity parameter of the small molecule compound to be tested; and the lipophilicity parameter of the small molecule compound to be tested. The test condition data required in the preset chromatographic test conditions include: the name of the instrument used in the preset chromatographic test conditions; the name of the chromatographic column; the column temperature; the flow rate; the name of the aqueous phase; the name of the organic phase; the elution time; and the initial and final proportions of the organic phase. The sample data is input into the trained support vector regression sub-model to obtain the first prediction retention time; The sample data is input into the trained regression sub-model to obtain the second prediction retention time; The average of the first predicted retention time and the second predicted retention time is determined as the predicted retention time of the small molecule compound to be tested. The retention time prediction model includes a support vector regression sub-model and a regression tree sub-model. The kernel function of the support vector regression sub-model is a Gaussian kernel function, and the algorithm used in the regression tree sub-model is the classification regression tree algorithm.

2. The prediction method as described in claim 1, characterized in that, The step of inputting the sample data into a pre-trained retention time prediction model includes: The sample data is stored in a comma-separated value file format, and the stored data is standardized to obtain standard sample data. The standard sample data is input into the pre-trained retention time prediction model.

3. The prediction method as described in claim 2, characterized in that, The preset chromatographic test conditions include: testing using a high-performance liquid chromatograph or an ultra-high-performance liquid chromatograph.

4. The prediction method as described in claim 1, characterized in that, The method further includes: Acquire training sample data, wherein the training sample data includes data of small molecule compound standards determined in the laboratory, data of small molecule compounds identified in biological samples determined in the laboratory, and data of small molecule compounds collected from the PubChem database; Based on the training sample data, the pre-constructed support vector regression sub-model and the regression tree sub-model are trained respectively to obtain the trained retention time prediction model.

5. A device for predicting retention time in liquid chromatography, characterized in that, include: The data acquisition module is used to acquire sample data of the small molecule compound to be tested. The sample data includes the test condition data required in the preset chromatographic test conditions and the physicochemical property parameters of the small molecule compound to be tested. The small molecule compound is a metabolite. The physicochemical property parameters of the small molecule compound to be tested include: the number of carbon atoms, hydrogen atoms, nitrogen atoms, oxygen atoms, phosphorus atoms, and sulfur atoms of the small molecule compound to be tested; the hydrophilicity parameter of the small molecule compound to be tested; and the lipophilicity parameter of the small molecule compound to be tested. The test condition data required in the preset chromatographic test conditions includes: the name of the instrument used in the preset chromatographic test conditions; the name of the chromatographic column; the column temperature; the flow rate; the name of the aqueous phase; the name of the organic phase; the elution time; and the initial and final proportions of the organic phase. The test time module is used to input the sample data into a trained support vector regression sub-model to obtain a first predicted retention time; input the sample data into a trained regression tree sub-model to obtain a second predicted retention time; and determine the average of the first predicted retention time and the second predicted retention time as the predicted retention time of the small molecule compound to be tested. The retention time prediction model includes a support vector regression sub-model and a regression tree sub-model. The kernel function of the support vector regression sub-model is a Gaussian kernel function, and the algorithm used in the regression tree sub-model is the classification regression tree algorithm.

6. An electronic device, characterized in that, The method includes a memory and a processor, the memory being used to store a computer program, and the processor being used to call and run the computer program stored in the memory to perform the method as described in any one of claims 1 to 4.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 4.

Images