Method and System for Constructing a miRNA Marker Model for Sepsis Based on Evolutionary Algorithm

Abstract

The present invention relates to a method and system for constructing a sepsis miRNA biomarker model based on an evolutionary algorithm, belonging to the field of bioinformatics. First, dimensionality reduction processing is performed on the miRNA expression matrix. Then, a search objective function is constructed, with maximizing the training model performance and eliminating the multicollinearity among miRNA screening results as the optimization objectives. Next, an improved genetic algorithm search strategy is adopted to decompose the search objective into multiple sub-problems for solution. For example, a three-classification problem is decomposed into three binary-classification problems to separately search for the optimal miRNA combinations for identifying healthy patients, identifying critically ill but non-sepsis patients, and identifying sepsis patients. Finally, through parameter tuning, the best miRNA biomarkers are obtained and a sepsis miRNA biomarker model is constructed, such as using machine learning algorithms like random forest or support vector machine. The present invention provides an efficient and accurate method for mining sepsis miRNA biomarkers.

Description

Technical Field

[0001] The present invention belongs to the field of bioinformatics and relates to a method and system for constructing a sepsis miRNA biomarker model based on an evolutionary algorithm. Background Art

[0002] Sepsis is a systemic inflammatory response syndrome caused by infection, which develops rapidly and has a high mortality rate. In recent years, with the development of high-throughput sequencing technology, microRNA (miRNA), as a class of non-coding RNA, the change in its expression level is closely related to the occurrence and development of various diseases, including sepsis. miRNA shows great potential in the mining of sepsis biomarkers, and the change in its expression level can be used as an early biomarker for sepsis.

[0003] However, effectively mining useful biomarkers from a large amount of miRNA expression data is still a challenge. The existing miRNA biomarker screening methods mainly have the following problems:

[0004] High data dimension: The miRNA expression matrix usually contains thousands of miRNAs, with a high data dimension and large computational complexity.

[0005] Multicollinearity: There are complex interaction relationships among miRNAs, resulting in the problem of multicollinearity, which affects the generalization ability of the model.

[0006] Model performance evaluation: The index selection and calculation method for evaluating the performance of the miRNA biomarker model need further study.

[0007] Therefore, it is necessary to develop an efficient and accurate miRNA biomarker screening method to solve the deficiencies of the existing methods and improve the accuracy and efficiency of sepsis miRNA biomarker mining. Summary of the Invention

[0008] In view of this, the purpose of the present invention is to provide a method and system for constructing a sepsis miRNA biomarker model based on an evolutionary algorithm, which can mine effective biomarkers from complex miRNA expression data and construct an accurate model.

[0009] To achieve the above purpose, the present invention provides the following technical solutions:

[0010] A method for constructing a sepsis miRNA biomarker model based on an evolutionary algorithm, the method comprising the following steps:

[0011] Input the miRNA expression matrix processed by TPM; the TPM value represents the number of specific genes per million transcripts and is normalized to relative abundance;

[0012] Perform dimensionality reduction on the miRNA expression matrix for screening sepsis miRNA markers;

[0013] Construct a search objective function for sepsis miRNA markers, with maximizing the training model performance and eliminating the multicollinearity among miRNA screening results as the optimization objectives;

[0014] Adopt an improved genetic algorithm search strategy to decompose the search objective into multiple sub-problems for solution;

[0015] Through parameter tuning, obtain the best miRNA markers and construct a sepsis miRNA marker model.

[0016] Furthermore, the dimensionality reduction processing of the miRNA expression matrix consists of DESeq differential expression analysis, differential expression gene screening, and screening result representation. The specific working principle is as follows:

[0017] Use DESeq differential expression analysis to output genes with significantly different expressions;

[0018] Screen significantly different genes according to the confidence level;

[0019] Draw the corresponding differential expression volcano plot for analysis.

[0020] Furthermore, the search objective function for sepsis miRNA markers is as follows:

[0021]

[0022] Among them, ModelScore represents the model performance score, Penalty represents the penalty coefficient, f now represents the existing feature set, |f now | represents the number of existing features, and MaxFeature represents the maximum number of features.

[0023] Furthermore, in the search objective function for sepsis miRNA markers, the calculation of the model performance score is to use the input feature set f now , and the corresponding miRNA expression data is used to train and test a random forest model. The obtained test F1 Score is the model performance score.

[0024] Furthermore, in the decomposition of the search objective, the multi-classification problem with a large information demand is decomposed into multiple binary-classification problems with a small information demand. For the best sepsis miRNA marker combination, it is expressed in the following form:

[0025]

[0026] Among them, R best represents the best sepsis miRNA marker combination, Indicates the optimal miRNA combination for identifying healthy patients, Indicates the optimal miRNA combination for identifying critically ill but non-septic patients, Indicates the optimal miRNA combination for identifying septic patients;

[0027] According to the above formula, the search for R best is regarded as the union of the search results for and respectively.

[0028] Furthermore, the improved genetic algorithm search strategy is specifically as follows: Use three genetic iterative populations to search for and respectively. Abbreviate the binary classification problems of these three sub-search models as Discriminant HC, Discriminant PC, and Discriminant SP respectively; To allow parallel computing and accelerate the search speed, perform iterations on the three sub-search models simultaneously, and take the top m miRNA combinations of the three sub-search models as and candidate combination sets and respectively. Take the Cartesian product of all candidate combination sets to combine the "perfect combination" candidate set R group for the three-classification problem:

[0029]

[0030] Prove that the maximum number of "perfect combination" candidate sets of size m can be combined from three sub-candidate combination sets of size m 3 ;

[0031] Suppose and both contain and Then the miRNA combination with the highest score in R group must be R best ; To accelerate the search speed of the model, use R group as the search set, perform one step of crossover, genetic, and mutation on the population within the set, and take the best result as R best .

[0032] Furthermore, the parameter tuning is specifically as follows: Perform grid search on the initial feature number, population number, and crossover rate of the sub-genetic algorithm; Obtain the sepsis miRNA biomarker combination with the best score; Then put the data corresponding to the combination into the random forest model to search for the best parameters, and obtain the best model parameters of the random forest.

[0033] A sepsis miRNA biomarker model construction system based on an evolutionary algorithm, the system includes:

[0034] A data preprocessing module, which receives the input miRNA expression matrix, performs dimensionality reduction on it, and outputs potential sepsis miRNA markers;

[0035] A search objective function module, which receives potential sepsis miRNA markers, constructs a search objective function, and outputs the search objective function;

[0036] An evolutionary algorithm module, which receives the search objective function and potential sepsis miRNA markers, and adopts an improved genetic algorithm search strategy to decompose the search objective into multiple sub-problems for solution, decomposes the three-classification problem into three binary-classification problems, and respectively searches for the best miRNA combination for identifying healthy patients, identifying critically ill but non-sepsis patients, and identifying sepsis patients, and outputs the best miRNA marker combination;

[0037] A model construction module, which receives the best miRNA marker combination, and uses machine learning algorithms such as random forest or support vector machine to construct a sepsis miRNA marker model, and outputs the sepsis miRNA marker model;

[0038] A model evaluation module, which receives the sepsis miRNA marker model, and uses indicators such as accuracy or recall to evaluate the performance of the constructed sepsis miRNA marker model

[0039] A computer-readable storage medium, on which computer-executable instructions are stored for executing the described method.

[0040] .A computer program product, which contains the computer-executable code of the described method.

[0041] The beneficial effects of the present invention are as follows: The present invention provides a method for mining sepsis miRNA markers, uses the DESeq method to perform dimensionality reduction on the miRNA expression matrix, reducing the complexity of subsequent screening; takes maximizing the performance of the training model and eliminating the multicollinearity between miRNA screening results as two main optimization objectives to ensure the accuracy and compliance of the search results; by decomposing the search problem into multiple sub-problems for solution and transforming the model classification task into three sub-tasks, more information is provided for the feature screening model.

[0042] Other advantages, objectives and features of the present invention will be described to some extent in the subsequent specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following specification. Brief Description of the Drawings

[0043] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below with reference to the accompanying drawings, where:

[0044] Figure 1 It is a flowchart of a miRNA expression data preprocessing method;

[0045] Figure 2 It is a flowchart of a sepsis miRNA biomarker mining method based on a genetic algorithm. Specific embodiments

[0046] The following specific examples illustrate the embodiments of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the drawings provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0047] Among them, the accompanying drawings are only for illustrative purposes, showing only schematic diagrams, not physical diagrams, and should not be construed as a limitation to the present invention; in order to better illustrate the embodiments of the present invention, some components in the drawings will be omitted, enlarged or reduced, which do not represent the dimensions of actual products; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0048] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", "front", "rear", etc. indicating the orientation or positional relationship, they are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, the terms describing the positional relationship in the accompanying drawings are only for illustrative purposes and should not be construed as a limitation to the present invention. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to specific circumstances.

[0049] For hospital clinical data, this embodiment provides a miRNA biomarker mining method based on a genetic algorithm.

[0050] The sepsis miRNA biomarker mining method based on a genetic algorithm includes the following steps:

[0051] S1. Construct a data set;

[0052] S101. Use the DESeq method to reduce the dimension of the miRNA expression matrix to obtain miRNAs with significant differences.

[0053] S102. Construct a data set composed of miRNA expression data and clinical results.

[0054] S103. Divide the data set into a training set and a test set.

[0055] Furthermore, construct an improved evolutionary algorithm search strategy.

[0056] S2. Set the search objective function for sepsis miRNA markers as:

[0057]

[0058] where ModelScore represents the model performance score, Penalty represents the penalty coefficient, f now represents the existing feature set, |f now | represents the number of existing features, and MaxFeature represents the maximum number of features.

[0059] S3. Construct an evolutionary algorithm search strategy according to the specific problem to obtain the best sepsis search model.

[0060] S301. Convert the three-class classification problem of healthy population HC, non-sepsis patients with disease PC, and sepsis patients SP into three binary classification problems, namely, determine whether the sample is HC, determine whether the sample is PC, and determine whether the sample is SP.

[0061] S302. Construct an evolutionary algorithm search strategy according to the problem division result, and use three genetic iterative populations to search and Abbreviate the binary classification problems of these three sub-search models as determining HC, determining PC, and determining SP respectively. To allow parallel computing and speed up the search, this paper iterates the three sub-search models simultaneously, and takes the first m miRNA combinations of the three sub-search models as and the candidate combination sets and Take the Cartesian product of all candidate combination sets to combine the "perfect combination" candidate set R group .

[0062]

[0063] It can be proved that the three sub-candidate combination sets of size m can combine at most a "perfect combination" candidate set of size m 3 .

[0064] Suppose and both contain and Then the miRNA combination with the highest score in R group must be R best . However, due to the limitations of the sub-search model itself, and may be randomly distributed in the components of and in each set, which will slow down the search speed of the search model. Therefore, in order to speed up the search speed of the model, this paper uses R group as the search set, performs one crossover, genetic, and mutation step on the population within the set, and takes the best result as R best .

[0065] S303. Set the specific sepsis search parameter range, use the training set to search, and obtain the best sepsis miRNA biomarker combination found.

[0066] S4. Through parameter tuning, obtain the sepsis miRNA biomarker model;

[0067] Specifically: (1) Use the grid search method and the training set to search for the best parameters of the random forest, and obtain the sepsis miRNA biomarker model with the best performance on the training set; (2) Use the test set to further test the performance of the sepsis miRNA biomarker model.

[0068] Example 1: Construction of a sepsis miRNA biomarker model based on hospital clinical data

[0069] Scenario: A hospital hopes to use miRNA expression data to construct a sepsis model to identify patients earlier and take treatment measures.

[0070] Steps:

[0071] Data collection: Collect miRNA expression data of sepsis patients and non-sepsis patients from the hospital and perform TPM normalization processing.

[0072] Data preprocessing: Use the DESeq method to perform differential expression analysis on the miRNA expression matrix after TPM processing, and screen out miRNAs with significant differences.

[0073] Model construction:

[0074] Search objective function: Set the maximization of the model performance score (such as the F1Score of the random forest) and the feature number limit as the optimization objective.

[0075] Genetic algorithm: An improved genetic algorithm is used to decompose the three-class classification problem into three binary-class classification problems (discriminating HC, discriminating PC, discriminating SP), and the optimal miRNA combination is searched for each respectively.

[0076] Parameter tuning: The grid search method is used to optimize the parameters of the random forest model to obtain the optimal sepsis miRNA biomarker model.

[0077] Model evaluation: The performance of the model is evaluated using the test set, such as indicators like accuracy and recall rate.

[0078] Example 2: Construction of a sepsis miRNA biomarker model based on a public dataset

[0079] Scenario: Researchers hope to explore the potential applications of sepsis miRNA biomarkers using publicly available miRNA expression datasets.

[0080] Steps:

[0081] Data download: Download the miRNA expression data of sepsis patients and non-sepsis patients from public databases (such as GEO, TCGA), and perform TPM normalization processing.

[0082] Data preprocessing: Use the DESeq method to perform differential expression analysis on the miRNA expression matrix after TPM processing, and screen out miRNAs with significant differences.

[0083] Model construction:

[0084] Search for the objective function: Set the maximization of the model performance score (such as the accuracy of the support vector machine) and the feature number limit as the optimization objectives.

[0085] Genetic algorithm: An improved genetic algorithm is used to decompose the three-class classification problem into three binary-class classification problems, and the optimal miRNA combination is searched for each respectively.

[0086] Parameter tuning: Use the cross-validation method to optimize the parameters of the support vector machine model to obtain the optimal sepsis miRNA biomarker model.

[0087] Model evaluation: The performance of the model is evaluated using the test set, such as indicators like accuracy and recall rate.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the purpose and scope of the present technical solution, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for constructing a sepsis miRNA biomarker model based on an evolutionary algorithm, characterized in that: The method includes the following steps: Input the processed miRNA expression matrix of transcripts per million (TPM); Perform dimensionality reduction on the miRNA expression matrix for screening sepsis miRNA markers; Construct a search objective function for sepsis miRNA markers, with maximizing the training model performance and eliminating the multicollinearity among miRNA screening results as the optimization objectives; the search objective function for sepsis miRNA markers is: Among them, ModelScore represents the model performance score, Penalty represents the penalty coefficient, and f now represents the existing feature set, |f now | represents the number of existing features, and MaxFeature represents the maximum number of features; Adopt an improved genetic algorithm search strategy to decompose the search objective into multiple sub-problems for solution; in the decomposition of the search objective, the multi-classification problem with large information demand is decomposed into multiple binary-classification problems with small information demand. For the best combination of sepsis miRNA markers, it is expressed in the following form: Among them, R best represents the optimal combination of sepsis miRNA markers, represents the optimal miRNA combination for identifying healthy patients, represents the optimal miRNA combination for identifying critically ill but non-sepsis patients, represents the optimal miRNA combination for identifying sepsis patients; According to the above formula, the search for R best is regarded as the union of the search results for and respectively; Through parameter tuning, obtain the best miRNA markers and construct a sepsis miRNA marker model.

2. The method for constructing a sepsis miRNA biomarker model based on an evolutionary algorithm according to claim 1, wherein: The dimensionality reduction processing of the miRNA expression matrix consists of DESeq differential expression analysis, screening of differentially expressed genes, and representation of screening results. The specific working principle is: Use DESeq for differential gene expression analysis based on the negative binomial distribution to output genes with significantly different expressions; Screen significantly differentially expressed genes according to the confidence level; Draw the corresponding differential expression volcano plot for analysis.

3. The method for constructing a sepsis miRNA biomarker model based on an evolutionary algorithm according to claim 1, characterized in that: In the sepsis miRNA biomarker search objective function, the model performance score is calculated by using the input feature set f now , and training and testing a random forest model with the corresponding miRNA expression data, and the F1 Score obtained from the test is the model performance score.

4. The method for constructing a sepsis miRNA biomarker model based on an evolutionary algorithm according to claim 1, characterized in that: The specific improved genetic algorithm search strategy is as follows: three genetic iterative populations are used to search respectively and The binary classification problems of these three sub-search models are respectively referred to as discriminant HC, discriminant PC, and discriminant SP; to allow parallel operations and speed up the search, the three sub-search models are iterated simultaneously, and the first m miRNA combinations of the three sub-search models are respectively taken as and the candidate combination sets and The Cartesian product is performed on all candidate combination sets to combine the candidate set R of "perfect combinations" for the three-classification problem group : Prove that the maximum number of candidate combination sets of three sub-candidate combination sets of size m can combine into a candidate set of "perfect combination" of size m 3 ; Suppose and both contain and Then the miRNA combination with the highest score in R group must be R best ; To speed up the search speed of the model, use R group as the search set, perform one crossover, genetic, and mutation step on the population within the set, and take the best result as R best .

5. The method for constructing a sepsis miRNA biomarker model based on an evolutionary algorithm according to claim 3, characterized in that: The specific parameter tuning is as follows: perform grid search on the initial number of features, population size, and crossover rate of the genetic algorithm; obtain the best combination of sepsis miRNA markers with the best score; then put the data corresponding to the combination into the random forest model to search for the best parameters and obtain the best model parameters of the random forest.

6. A sepsis miRNA biomarker model construction system based on an evolutionary algorithm, characterized in that: The system includes: A data preprocessing module that receives the input miRNA expression matrix, performs dimensionality reduction on it, and outputs potential sepsis miRNA markers; A search objective function module that receives potential sepsis miRNA markers, constructs a search objective function, and outputs the search objective function; the search objective function is: Among them, ModelScore represents the model performance score, Penalty represents the penalty coefficient, and f now represents the existing feature set, and |f now | represents the number of existing features, and MaxFeature represents the maximum number of features; An evolutionary algorithm module that receives the search objective function and potential sepsis miRNA markers, and adopts an improved genetic algorithm search strategy to decompose the search objective into multiple sub-problems for solution, decomposes the three-classification problem into three binary-classification problems, and separately searches for the best miRNA combinations for identifying healthy patients, identifying critically ill but non-sepsis patients, and identifying sepsis patients, and outputs the best combination of miRNA markers; in the decomposition of the search objective, the multi-classification problem with large information demand is decomposed into multiple binary-classification problems with small information demand. For the best combination of sepsis miRNA markers, it is expressed in the following form: Among them, R best represents the optimal combination of sepsis miRNA markers, represents the optimal miRNA combination for identifying healthy patients, represents the optimal miRNA combination for identifying critically ill but non-sepsis patients, represents the optimal miRNA combination for identifying sepsis patients; According to the above formula, the search for R best is regarded as the union of the search results for and respectively; A model construction module that receives the best combination of miRNA markers and uses machine learning algorithms such as random forest or support vector machine to construct a sepsis miRNA marker model and outputs the sepsis miRNA marker model; A model evaluation module that receives the sepsis miRNA marker model and uses indicators such as accuracy or recall to evaluate the performance of the constructed sepsis miRNA marker model.

7. A computer-readable storage medium, characterized in that: It stores computer-executable instructions for executing the method described in any one of claims 1 to 5.

8. A computer program product, characterized in that: Computer-executable code including the method according to any one of claims 1 to 5.

Images