Disease key factor mining method combined with multi-omics data, computer device and storage medium

Abstract

This invention discloses a method, computer device, and storage medium for mining key disease factors by combining multi-omics data. The method includes acquiring multiple omics data related to a specific disease; acquiring a first matrix and a second matrix related to the disease factors under investigation from the omics data; establishing a joint sparse optimization regularization model; and solving the joint sparse optimization regularization model to obtain key disease factors. This invention can obtain key disease factors for a specific disease, reducing the number of disease factors to be analyzed and improving efficiency. By integrating multi-omics data related to a specific disease to extract high-quality omics information, it treats all omics components of the lesions during the disease's development as a complete target group, realizing the quantification process of the regulatory effect of key factors on disease targets, and effectively improving the prediction accuracy of key factors. This invention has wide applications in the field of bioinformatics.

Description

Technical Field

[0001] This invention relates to the field of bioinformatics, and in particular to a method, computer device, and storage medium for mining key disease factors by combining multi-omics data. Background Technology

[0002] As scientific research deepens, our understanding of diseases has become more precise at the molecular level, giving rise to molecular biomedicine. Simultaneously, we have gradually realized that many complex diseases do not originate from abnormalities in a single molecule, but rather from the interactions of intricate biological regulatory networks. Therefore, accurately identifying key disease factors from vast and complex biological networks has become a research hotspot and important task in the biomedical field in recent years. Disease-related factors, including transcription factors, protein-coding RNA, non-coding RNA (ncRNA), and microRNA (miRNA), can serve as biomarkers or targets for diseases. Analyzing disease-related factors can not only reveal the mechanisms of disease development and progression but also provide a reliable basis for formulating targeted prevention and treatment strategies.

[0003] However, due to the complex biological networks and interactions behind diseases, there are numerous factors associated with a disease. If all factors related to a disease are included in the processing scope when studying a disease, it will generate a huge amount of data processing, which is inefficient and time-consuming. Summary of the Invention

[0004] To address the technical problem of processing numerous factors in current disease analysis, the present invention aims to provide a method, computer device, and storage medium for mining key disease factors by combining multi-omics data.

[0005] On one hand, embodiments of the present invention include a method for mining key disease factors by combining multi-omics data, the method comprising the following steps:

[0006] Acquire multi-omics data related to a specific disease;

[0007] Several disease factors to be investigated are identified;

[0008] Obtain the first matrix and the second matrix corresponding to each of the omics data respectively; the first matrix is ​​an omics data matrix containing information on each of the disease factors to be investigated, and the second matrix is ​​an omics data matrix containing information on disease factors other than the disease factors to be investigated.

[0009] A joint sparse optimization regularization model is established based on the first matrix and the second matrix;

[0010] The joint sparse optimization regularization model is solved to obtain the key factors of the disease.

[0011] Furthermore, the acquisition of multiple omics data related to a specific disease includes, but is not limited to:

[0012] Acquire transcriptomics data; the transcriptomics data includes conventional transcriptome sequencing data and single-cell transcriptome sequencing data corresponding to the specific disease;

[0013] Acquire genomic data; the genomic data includes single nucleotide variant data, insertion and deletion data, and copy number variation data corresponding to the specific disease;

[0014] Obtain epigenetic data; the epigenetic data includes DNA methylation data, histone modification data, and chromatin immunoprecipitation data corresponding to the specific disease;

[0015] Acquire proteomics data; the proteomics data includes protein expression data and post-translational modification data corresponding to the specific disease;

[0016] Obtain metabolomics data; the metabolomics data includes metabolite concentration data corresponding to the specific disease.

[0017] Further, the step of obtaining the first matrix and the second matrix corresponding to each of the omics data includes:

[0018] Obtain m samples or cells;

[0019] Each sample or cell yields its own n disease factors to be investigated and k disease factors that do not belong to the disease factors to be investigated.

[0020] For any i-th omics data, data corresponding to all the disease factors to be investigated are extracted from the omics data to form the first matrix A. i Among them, A i ∈R m×n , i = 1, 2, ... t, where t is the number of types of the omics data;

[0021] For any i-th omics data point, data corresponding to disease factors other than the disease factor to be investigated are extracted from the omics data to form the second matrix B. i Among them, B i ∈R m×k , i = 1, 2, ... t, where t is the number of types of the omics data.

[0022] Further, the step of establishing a joint sparse optimization regularization model based on the first matrix and the second matrix includes:

[0023] According to the formula

[0024] A i X i =B i +ε i

[0025] Establish the joint sparse optimization regularization model; where i = 1, 2, ..., t, t is the number of categories of the omics data, A i B is the first matrix corresponding to the i-th omics data. i X is the second matrix corresponding to the i-th omics data. i Let A be the first matrix. i The disease factors to be investigated contained in the second matrix B i The relationship matrix of disease factors included, ε i This is the noise matrix.

[0026] Furthermore, solving the joint sparse optimization regularization model to obtain key disease factors includes:

[0027] Use L p,q The regularization method solves the joint sparse optimization regularization model to determine the relation matrix X. i ;

[0028] Using the aforementioned relation matrix X i The disease factors to be investigated corresponding to the zero item in the middle are considered as key factors of the disease.

[0029] Furthermore, the use of L p,q Solving the joint sparse optimization regularization model using regularization methods includes:

[0030] Establish equations

[0031]

[0032] Where p≥1, 0≤q≤1; X=[X1,X2,…,X… i ,…,X t ], where λ is a constant coefficient.

[0033] The equations were solved using the proximal gradient algorithm to obtain the relation matrix X. i .

[0034] Furthermore, the method of solving the equation using the proximal gradient algorithm includes:

[0035] The process is executed in several iterations in sequence; in any k-th iteration:

[0036] When the k-th iteration is the 1st iteration, set The matrix is ​​zero. When the k-th iteration is not the 1st iteration, the result is calculated based on the (k-1)-th iteration. set up

[0037] According to the formula

[0038]

[0039] Z = [Z1, Z2, ..., Z i ,…,Z t Determine the calculation result of the k-th iteration process. Where i = 1, 2, ..., t; v k It is a step-size sequence.

[0040] when Convergence, according to Determine the relation matrix X i Conversely, if the condition is not met, the (k+1)th iteration process is executed.

[0041] Furthermore, p≥1, 0≤q≤1. Specifically, we can set p=1 or 2, q=0, 1 / 2, 2 / 3 or 1, thus forming combinations such as p=1 and q=0, p=1 and q=1 / 2, p=1 and q=2 / 3, p=1 and q=1, p=2 and q=0, p=2 and q=1 / 2, p=2 and q=2 / 3, p=2 and q=1, which correspond to L respectively. 1,0 L 1,1 / 2 L 1,2 / 3 L 1,1 L 2,0 L 2,1 / 2 L 2,2 / 3 L 2,1 Models such as...

[0042] On the other hand, embodiments of the present invention also include a computer device, including a memory and a processor, the memory for storing at least one program, and the processor for loading at least one program to execute the disease key factor mining method combining multi-omics data in the embodiments.

[0043] On the other hand, embodiments of the present invention also include a computer-readable storage medium storing a processor-executable program, which, when executed by a processor, is used to perform the disease key factor mining method combining multi-omics data in the embodiments.

[0044] The beneficial effects of this invention are as follows: The disease key factor mining method combining multi-omics data in the embodiments can obtain the disease key factors of a specific disease, reduce the number of disease factors to be analyzed when studying and analyzing a specific disease, and improve efficiency; the disease key factor mining method combining multi-omics data in the embodiments extracts high-quality omics information by integrating multi-omics data related to a specific disease, and regards all omics components of the lesions in the process of disease occurrence and development as a complete target group, ensuring that all omics components that lead to the lesions are used as targets of inferred key transcription factors, and finally realizes the quantification process of the regulatory effect of key factors on disease targets, effectively improving the prediction accuracy of key factors. Attached Figure Description

[0045] Figure 1 This is a schematic diagram illustrating the steps of the disease key factor mining method that combines multi-omics data in the embodiment. Detailed Implementation

[0046] Key disease factors can be identified through various experimental techniques and computational methods. Experimental techniques include high-throughput omics methods such as transcriptomics, genomics, epigenetics, proteomics, and metabolomics, which can extract data on diseases and various factors. Computational methods utilize mathematical models and computer algorithms to mine correlations between diseases and various factors from large-scale data. In the past, various omics data were often used independently to extract biological information, lacking comprehensive or fusion analysis. Although research methods analyzing single omics data have yielded many results, they still limit a comprehensive understanding of the complexity of diseases. In fact, differences at the genomic level do not necessarily translate into the final manifestation of disease, which highlights the complex biological networks and interactions behind diseases and underscores the importance of cross-omics analysis. Therefore, to fully utilize the advantages of omics technologies and achieve a more comprehensive understanding of human diseases, new computational methods are needed to integrate and analyze multiple types of omics data. Our method can easily and efficiently integrate and analyze multiple types of omics data, further accurately and rapidly extracting key factors closely related to specific diseases from massive amounts of information. This will provide strong data support and scientific evidence for disease research, prevention, and treatment.

[0047] Based on the above principles, this embodiment provides a method for mining key disease factors by combining multi-omics data. (Refer to...) Figure 1 The method for mining key disease factors by combining multi-omics data includes the following steps:

[0048] S1. Acquire multiple omics data related to a specific disease;

[0049] S2. Define several disease factors to be investigated;

[0050] S3. Obtain the first and second matrices corresponding to each omics data respectively;

[0051] S4. Establish a joint sparse optimization regularization model based on the first and second matrices;

[0052] S5. Solve the joint sparse optimization regularization model to obtain the key factors of the disease.

[0053] The task of steps S1-S5 is to discover key disease factors associated with a specific disease, where the specific disease is the one to be studied and analyzed. The key disease factors obtained are a small number of factors that can cover most of the omics changes of the specific disease.

[0054] Steps S1-S5 can be executed by a computer.

[0055] In step S1, t omics data related to a specific disease are acquired, including the first omics data, the second omics data, the third omics data, ..., the tth omics data.

[0056] Specifically, when performing step S1, the first omics data obtained can be transcriptomics data, which includes conventional transcriptome sequencing data and single-cell transcriptome sequencing data corresponding to a specific disease; the second omics data obtained can be genomics data, which includes single nucleotide variant data, insertion and deletion data, and copy number variation data corresponding to a specific disease; the third omics data obtained can be epigenetics data, which includes DNA methylation data, histone modification data, and chromatin immunoprecipitation data corresponding to a specific disease; the fourth omics data obtained can be proteomics data, which includes protein expression data and post-translational modification data corresponding to a specific disease; and the fifth omics data obtained can be metabolomics data, which includes metabolite concentration data corresponding to a specific disease.

[0057] In step S2, several disease factors to be investigated are defined. From all disease factors related to a specific disease, the disease factors most likely to be key disease factors can be preliminarily screened as the disease factors to be investigated. In this embodiment, the disease factors are essentially biomolecules such as transcription factors, protein-coding RNA, non-coding RNA, and microRNAs, which can serve as disease biomarkers or targets.

[0058] In step S3, the first matrix and the second matrix corresponding to each of the omics data determined in step S1, such as the first omics data, the second omics data, the third omics data, ... the t-th omics data, are obtained respectively.

[0059] For example, for the first omics data, the transcriptomics data containing information about each disease factor to be investigated are extracted from the first omics data to form the first matrix corresponding to the first omics data; the transcriptomics data containing information about other disease factors (not belonging to the disease factors to be investigated) are extracted from the first omics data to form the second matrix corresponding to the first omics data.

[0060] Specifically, when performing step S3, which is to obtain the first and second matrices corresponding to each omics data respectively, the following steps can be performed:

[0061] S301. Obtain m samples or cells;

[0062] S302. Obtain n disease factors to be investigated and k disease factors that do not belong to the disease factors to be investigated from each sample or cell respectively;

[0063] S303. For any i-th omics data, extract the data corresponding to all the disease factors to be investigated from the omics data to form the first matrix A. i ;

[0064] S304. For any i-th omics data, extract data from the omics data corresponding to disease factors other than the disease factor to be investigated to form a second matrix B. i .

[0065] In step S301, samples can be taken from patients with a specific disease to obtain m samples or cells (cells will be used as an example below).

[0066] In step S302, n disease factors to be investigated and k disease factors that do not belong to the disease factors to be investigated are obtained from each cell. After performing step S302, a total of m×n disease factors to be investigated and m×k other disease factors (that is, disease factors that do not belong to the disease factors to be investigated) are obtained from m cells.

[0067] In step S303, i = 1, 2, ... t is set. For the i-th omics data, data corresponding to all the disease factors to be investigated are obtained from the i-th omics data, thereby forming the first matrix A corresponding to the i-th omics data. i The first matrix A i Satisfy A i ∈R m×n .

[0068] In step S304, i = 1, 2, ... t is set. For the i-th omics data, data corresponding to all other disease factors (i.e., disease factors that do not belong to the disease factors to be investigated) are obtained from the i-th omics data to form the second matrix B corresponding to the i-th omics data.i The second matrix B i Satisfy B i ∈R m×k .

[0069] In this embodiment, when performing step S4, which is to establish the joint sparse optimization regularization model based on the first and second matrices, the relationship between the key disease factors and other disease factors can be assumed to be roughly a linear system. Based on this assumption, the constructed joint sparse optimization (JSO) regularization model takes the following form:

[0070] A i X i =B i +ε i #(1)

[0071] Where i = 1, 2, ..., t, and t is the total number of omics data. For example, when i = 1, the obtained A1X1 = B1 + ε1 represents the joint sparse optimization regularization model corresponding to the first omics data.

[0072] In the joint sparse optimization regularization model, A i Let B be the first matrix corresponding to the i-th omics data. i Let X be the second matrix corresponding to the i-th omics data. i Let A be the first matrix. i The disease factors to be investigated and the second matrix B are included. i The relationship matrix of disease factors included, ε i This is the noise matrix.

[0073] The principle of the joint sparse optimization regularization model shown in Equation (1) is that if multiple measured signals are correlated, multiple correlated variables can be inferred simultaneously in a cooperative manner, thus supporting that they belong to the same set. Therefore, the joint sparse optimization regularization model shown in Equation (1) can utilize and integrate multiple omics data to predict key disease factors. After integrating multiple omics data, key factor prediction is to find a small number of factors that regulate the target and cover most omics changes, which can be described as a sparse optimization problem, i.e., finding an X i This allows A to be achieved using only a small number of selected key factors. i X i and B i Minimize the differences between them.

[0074] Based on the above principles, in step S5, the joint sparse optimization regularization model shown in formula (1) is solved, and the obtained X i Corresponding to the first matrix Ai Information on some disease factors to be investigated, which are those factors that can cause A i X i and B i The few key factors that minimize the differences between them can be used as the key disease factors to be obtained in this embodiment.

[0075] In this embodiment, when executing step S5, L can be used by employing a joint sparsity penalty of p and q. p,q Regularization methods are used to solve joint sparsity problems, thereby expressing the inference process of key factors.

[0076] In this embodiment, L is used in step S5. p,q When solving the joint sparse optimization regularization model using regularization methods, the following equation can be established:

[0077]

[0078] In equation (2), X = [X1, X2, ..., X i ,…,X t λ is a constant coefficient.

[0079]

[0080] In equation (2), p ≥ 1, 0 ≤ q ≤ 1. Specifically, p = 1 or 2, q = 0, 1 / 2, 2 / 3 or 1, thus forming combinations such as p = 1 and q = 0, p = 1 and q = 1 / 2, p = 1 and q = 2 / 3, p = 1 and q = 1, p = 2 and q = 0, p = 2 and q = 1 / 2, p = 2 and q = 2 / 3, p = 2 and q = 1, which correspond to L respectively. 1,0 L 1,1 / 2 L 1,2 / 3 L 1,1 L 2,0 L 2,1 / 2 L 2,2 / 3 L 2,1 Models such as...

[0081] Equation (2) can be solved using the Proximal Gradient Algorithms (PGA) to obtain the relation matrix X. i The proximal gradient algorithm is characterized by fast convergence speed, simple formula, and low computational complexity.

[0082] When performing step S5 to solve the equation using the proximal gradient algorithm, the following steps can be executed:

[0083] S501. Determine that this iteration process is the kth iteration process;

[0084] S502. If the current iteration (the k-th iteration) is the 1st iteration, then set... If the current iteration (the k-th iteration) is an iteration following the 1st iteration, then the result of the previous iteration (the (k-1)-th iteration) is obtained. As

[0085] S503. According to formula (3)

[0086]

[0087] Z = [Z1, Z2, ..., Z i ,…,Z t Determine the calculation results of this iteration process (the k-th iteration process). Where i = 1, 2, ..., t; v k It is a step-size sequence.

[0088] S504. When the result obtained in step S503 Convergence (e.g.) and If the absolute value of the difference is less than a preset threshold, then... Determined as relation matrix X i Conversely, if k = k + 1, jump to S501.

[0089] Steps S501-S504 are the steps executed in one of the iterative processes, and each iterative process executes the same steps as steps S501-S504. If the result calculated during a certain round of iterative processing... Determined as relation matrix X i Then we obtain the relation matrix X. i According to the relation matrix X i Once the key disease factors are identified, the subsequent iteration process will not be executed; otherwise, the process will jump back to S501 to start the next iteration process.

[0090] From the perspective of consistency theory, L p,q Regularization methods require only weak regularity conditions to guarantee accurate recovery. Meanwhile, L... p,qRegularization methods exhibit stronger sparsity, enabling the acquisition of sparser solutions from fewer samples. Furthermore, PGA-JSO has been demonstrated to jointly reconstruct multiple measurement signals from different measurement data. Utilizing the synchronization effects of multiple omics variable signals and the similarity of relationships between factors can improve the recovery ability and estimation accuracy of Xi, thereby accurately predicting key disease factors. Therefore, by inducing joint sparsity, the method involved in this embodiment can select a group that quantifies the moderating strength of key factors on all targets, utilizing and integrating multiple omics data to improve the prediction accuracy and efficiency of key disease factors. Prior to this work, joint sparse optimization regularization models had not been used for predicting various key disease factors.

[0091] Based on the reported ranking of driving factors among key predictive factors, an evaluation score S is introduced. T and weighted score S TW To quantify the accuracy of different methods, we use formulas (5) and (6).

[0092]

[0093] Among them, R i This is the ranking of reported driving factors among the key predictors, r i This is a ranking of reported driving factors based on the number of publications they have. Higher scores are assigned to reported driving factors that rank higher among the predicted key factors. Evaluation Score S T The scores of the reported driving factors appearing in the top 20 key factors predicted by the method are directly summed. Although the evaluation score S T We considered the equal role of all reported driver factors in the disease process, but the importance of each factor also varies. Therefore, we searched for the number of publications on these factors in relation to lung adenocarcinoma, assuming that the more researchers studying these factors, the more important the factor. The more publications a driver factor has, the higher its weight.

[0094] Referring to Table 1, the comparison shows the prediction results of key factors calculated by different methods based on copy number variation (i.e., the first omics mentioned above), transcriptome data (i.e., the second omics mentioned above), and proteome data (i.e., the third omics mentioned above) of human lung adenocarcinoma (i.e., the specific disease mentioned above).

[0095] Table 1 Comparison of Key Factors Predicting Human Lung Adenocarcinoma

[0096]

[0097]

[0098] It should be noted that 0 in Table 1 indicates that the factor was not predicted to be among the top 20 key factors.

[0099] Table 1 lists the reported driving factors for cancer and lung adenocarcinoma, including TP53, MYC, BRAF, STAT3, ROS1, CTNNB1, SMAD3, RAC1, SMAD4, ESR1, SETDB1, ELF3, and EPAS1. The number of studies on each reported driving factor and its association with lung adenocarcinoma is listed, arranged from highest to lowest. The number of related studies can indirectly demonstrate the importance of this key factor in lung adenocarcinoma.

[0100] The prediction methods involved include GSO_L 2,0 GSO_L 2,1 JSO_L 1,0 JSO_L 1,1 / 2 JSO_L 1,2 / 3 JSO_L 1,1 JSO_L 2,0 JSO_L 2,1 / 2 JSO_L 2,2 / 3 and JSO_L 2,1 Among them, GSO_L 2,0 and GSO_L 2,1 As a current technology, JSO_L 1,0 JSO_L 1,1 / 2 JSO_L 1,2 / 3 JSO_L 1,1 JSO_L 2,0 JSO_L 2,1 / 2 JSO_L 2,2 / 3 and JSO_L 2,1 In this embodiment, L is used p,q Regularization methods are used to solve the joint sparse optimization regularization model, with cases where p=1 and q=0, etc.; GSO_L 2,0 Representative of L using group sparse optimization 2,0 Regularized models are analyzed using single-data transcriptome genomes; GSO_L 2,1 Representative of L using group sparse optimization 2,1 Regularized models are analyzed using single transcriptome data. The JSO method represents the use of joint sparse optimization regularized models with three omics datasets, where p = 1 or 2, and q = 0, 1 / 2, 2 / 3, or 1. Eight models are presented as special cases.

[0101] It should be noted that the data in the dashed box represents the prediction priority of a certain factor under a certain prediction method. The prediction priority ranges from 1 to 20, with a smaller value indicating a higher prediction priority. The two rows at the bottom of Table 1 represent the evaluation score and weight score of a certain factor under a certain prediction method, respectively. A larger value indicates a better prediction effect.

[0102] Based on the above explanation of Table 1, after applying this embodiment, the predictive effect on key factors of human lung adenocarcinoma is better than other prediction methods. This also indicates that the analysis of multi-omics data in this embodiment can achieve better predictive results compared to single-omics data mining. Specifically, L in this embodiment... 2,0 L 2,1 / 2 and L 2,2 / 3 The joint sparse optimization regularization model can achieve excellent results.

[0103] This embodiment extracts high-quality omics information by integrating multi-omics data related to specific diseases; it simultaneously constructs gene regulatory networks and predicts key factors, and treats all omics components of lesions during the disease occurrence and development process as a complete target group, ensuring that all omics components of lesions are used as targets for inferred key transcription factors, and ultimately realizes the quantification process of the regulatory effect of key factors on disease targets; effectively improving the prediction accuracy of key factors.

[0104] In this embodiment, a disease key factor mining system combining multi-omics data can be established. This system includes:

[0105] The first module is used to acquire multiple omics data related to a specific disease;

[0106] The second module is used to define several disease factors to be investigated;

[0107] The third module is used to obtain the first matrix and the second matrix corresponding to each omics data respectively; wherein, the first matrix is ​​an omics data matrix containing information on each disease factor to be investigated, and the second matrix is ​​an omics data matrix containing information on disease factors other than the disease factors to be investigated.

[0108] The fourth module is used to establish a joint sparse optimization regularization model based on the first and second matrices.

[0109] The fifth module is used to solve the joint sparse optimization regularization model to obtain key disease factors.

[0110] By using a disease key factor mining system that integrates multi-omics data, a disease key factor mining method that integrates multi-omics data can be implemented. Specifically, step S1 in the disease key factor mining method that integrates multi-omics data can be executed by the first module of the system, step S2 by the second module, step S3 by the third module, step S4 by the fourth module, and step S5 by the fifth module.

[0111] By running a disease key factor mining system that combines multi-omics data, the technical effects of disease key factor mining methods that combine multi-omics data can be achieved.

[0112] A computer program that executes the disease key factor mining method combining multi-omics data in this embodiment can be written into a computer device or storage medium. When the computer program is read out and run, the disease key factor mining method combining multi-omics data in this embodiment is executed, thereby achieving the same technical effect as the disease key factor mining method combining multi-omics data in the embodiment.

[0113] It should be noted that, unless otherwise specified, when a feature is referred to as "fixed" or "connected" to another feature, it can be directly fixed or connected to the other feature, or indirectly fixed or connected to the other feature. Furthermore, the descriptions of "upper," "lower," "left," and "right" used in this disclosure are only relative to the relative positional relationships of the components of this disclosure in the accompanying drawings. The singular forms "a" and "the" used in this disclosure are also intended to include the plural forms, unless the context clearly indicates otherwise. Moreover, unless otherwise defined, all technical and scientific terms used in this embodiment have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this embodiment specification is only for describing particular embodiments and is not intended to limit the invention. The term "and / or" as used in this embodiment includes any combination of one or more of the associated listed items.

[0114] It should be understood that although the terms first, second, third, etc., may be used to describe various elements in this disclosure, these elements should not be limited to these terms. These terms are only used to distinguish elements of the same type from each other. For example, a first element may also be referred to as a second element without departing from the scope of this disclosure, and similarly, a second element may also be referred to as a first element. The use of any and all instances or exemplary language (“e.g.,” “such as,” etc.) provided in this embodiment is intended only to better illustrate embodiments of the invention and, unless otherwise required, does not impose a limitation on the scope of the invention.

[0115] It should be recognized that embodiments of the present invention can be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium. The method can be implemented using standard programming techniques—including a non-transitory computer-readable storage medium configured with a computer program, wherein such a storage medium causes the computer to operate in a specific and predefined manner—according to the methods and drawings described in the specific embodiments. Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system. However, if desired, the program can be implemented in assembly or machine language. In any case, the language can be a compiled or interpreted language. Furthermore, for this purpose, the program can run on a programmed application-specific integrated circuit (ASIC).

[0116] Furthermore, the procedures described in this embodiment can be performed in any suitable order unless otherwise indicated by this embodiment or clearly contradicted by the context. The procedures (or variations and / or combinations thereof) described in this embodiment can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. A computer program includes multiple instructions executable by one or more processors.

[0117] Furthermore, the method can be implemented in any suitable type of computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices, etc. Aspects of the invention can be implemented as machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it is readable by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein. Furthermore, the machine-readable code, or portions thereof, can be transmitted via wired or wireless networks. The invention of this embodiment includes these and other different types of non-transitory computer-readable storage media when such media comprises instructions or programs that implement the steps above in conjunction with a microprocessor or other data processor. When programmed according to the methods and techniques of the invention, the invention also includes the computer itself.

[0118] A computer program can be applied to input data to perform the functions of this embodiment, thereby transforming the input data to generate output data stored in non-volatile memory. The output information can also be applied to one or more output devices, such as a display. In a preferred embodiment of the invention, the transformed data represents physical and tangible objects, including specific visual depictions of physical and tangible objects generated on the display.

[0119] The above are merely preferred embodiments of the present invention. The present invention is not limited to the above-described embodiments. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention, as long as they achieve the technical effects of the present invention by the same means, should be included within the scope of protection of the present invention. Within the scope of protection of the present invention, the technical solutions and / or implementation methods can have various modifications and variations.

Claims

1. A method for mining key disease factors by combining multi-omics data, characterized in that, The method for mining key disease factors by combining multi-omics data includes: Acquire multi-omics data related to a specific disease; Several disease factors to be investigated are identified; Obtain the first matrix and the second matrix corresponding to each of the omics data respectively; the first matrix is ​​an omics data matrix containing information on each of the disease factors to be investigated, and the second matrix is ​​an omics data matrix containing information on disease factors other than the disease factors to be investigated. A joint sparse optimization regularization model is established based on the first matrix and the second matrix; The joint sparse optimization regularization model is solved to obtain key disease factors; The step of establishing a joint sparse optimization regularization model based on the first matrix and the second matrix includes: According to the formula Establish the joint sparse optimization regularization model; wherein, , The number of types of the omics data. For the first The first matrix corresponding to each of the omics data. For the first The second matrix corresponding to each of the aforementioned omics data For the first matrix The disease factors to be investigated contained in the second matrix The matrix containing the relationships between disease factors. This is the noise matrix; Solving the joint sparse optimization regularization model to obtain key disease factors includes: use The regularization method solves the joint sparse optimization regularization model to determine the relation matrix. ; With the aforementioned relation matrix The disease factors to be investigated corresponding to the zero item in the non-Chinese section are considered as key factors of the disease. The use Solving the joint sparse optimization regularization model using regularization methods includes: Establish equations in, ; , The constant coefficients, , ; The equations were solved using the proximal gradient algorithm to obtain the relation matrix. ; The method of solving the equation using the proximal gradient algorithm includes: The process is executed in several rounds of iteration; in any given round... k During the round of iteration: When the k The first iteration is the first iteration process, which is set as follows: For a zero matrix, when the first... k The round of iteration is not the first round of iteration, according to the... k Calculation results of -1 round of iteration set up , According to the formula Determine the first k Calculation results of the round iteration process ,in, ; It is a step-size sequence. ; when Convergence, according to Determine the relation matrix Conversely, execute the first k +1 round of iterations.

2. The method for mining key disease factors combining multi-omics data according to claim 1, characterized in that, The acquisition of multiple omics data related to a specific disease includes: Acquire transcriptomics data; the transcriptomics data includes conventional transcriptome sequencing data and single-cell transcriptome sequencing data corresponding to the specific disease; Acquire genomic data; the genomic data includes single nucleotide variant data, insertion and deletion data, and copy number variation data corresponding to the specific disease; Obtain epigenetic data; the epigenetic data includes DNA methylation data, histone modification data, and chromatin immunoprecipitation data corresponding to the specific disease; Acquire proteomics data; the proteomics data includes protein expression data and post-translational modification data corresponding to the specific disease; Obtain metabolomics data; the metabolomics data includes metabolite concentration data corresponding to the specific disease.

3. The method for mining key disease factors by combining multi-omics data according to claim 1, characterized in that, The step of obtaining the first matrix and the second matrix corresponding to each of the omics data includes: Get One sample or cell; Obtain the corresponding samples or cells from each sample. The disease factors to be investigated and One disease factor that does not belong to the disease factors to be investigated; For any i The first matrix is ​​formed by extracting data corresponding to all the disease factors to be investigated from the omics data. ;in, , , The number of types of the omics data; For any i The second matrix is ​​formed by extracting data from the omics data corresponding to disease factors other than the disease factor to be investigated. ;in, , , The number of types of the omics data.

4. The method for mining key disease factors by combining multi-omics data according to claim 1, characterized in that, 。 5. A computer device, characterized in that, It includes a memory and a processor, the memory being used to store at least one program, and the processor being used to load at least one program to execute the disease key factor mining method combining multi-omics data as described in any one of claims 1-4.

6. A computer-readable storage medium storing a processor-executable program, characterized in that, The processor-executable program, when executed by the processor, is used to perform the disease key factor mining method combining multi-omics data as described in any one of claims 1-4.

Images