A bioinformatics analysis system for human tissue lipid metabolomics data
By introducing biochemical homology constraints and the EYM theorem to achieve optimal low-rank orthogonal loading, the problem of neglecting lipid molecular structure homology in traditional discriminant analysis is solved, enabling a more comprehensive analysis of human tissue lipid metabolomics data.
Patent Information
- Application Number
- CN202610821696.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-09
- Publication Date
- 2026-07-10
AI Technical Summary
Traditional orthogonal partial least squares discriminant analysis methods neglect the tight structural homology of lipid molecules in the biosynthetic pathway in human tissue lipid metabolomics studies, resulting in incomplete identification of core regulatory nodes.
By introducing biochemical homology structure constraints and EYM theorem, optimal low-rank optimization is performed on orthogonal loads to extract and compensate for latent variance. The rank-one structural principal axis vector is extracted through biochemical homology physical penalty matrix and singular value decomposition. An orthogonal variance compensation factor sequence is constructed, and a final state compensation variable projection importance sequence is generated to extract the core driving factors of targeted lipids.
This improves upon the problem of incomplete identification of core regulatory nodes caused by neglecting underlying regulatory information in traditional methods, thereby enhancing the accuracy and comprehensiveness of lipid metabolomics data analysis.
Smart Images
Figure CN122369559A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of bioinformatics data processing technology, and in particular to a bioinformatics analysis system for human tissue lipid metabolomics data. Background Technology
[0002] In the field of human tissue lipid metabolomics research, orthogonal partial least squares discriminant analysis (OPLS-DA) is a commonly used method to identify biomarkers with clinical phenotypic differences. This method separates phenotypic-related variance from irrelevant variance by projecting a high-dimensional lipid dataset onto a prediction subspace and an orthogonal subspace.
[0003] However, the inventors have discovered at least the following technical problems in the related technologies: Traditional analysis procedures usually discard the variance in the orthogonal subspace as pure random noise, without taking into account the strict structural homology of lipid molecules in the biosynthetic pathway. This results in a large number of latent structural regulatory metabolites that have no direct linear correlation with clinical phenotypic labels but play a supporting role in metabolic pathways being forcibly ignored by the model, thus causing incomplete identification of core regulatory nodes. Summary of the Invention
[0004] To address the shortcomings of existing technologies, one objective of this application is to provide a bioinformatics analysis system for human tissue lipid metabolomics data, which at least solves the problem that traditional discriminant analysis mostly uses the method of stripping and discarding orthogonal variances, which ignores the underlying regulatory information with physical correlations, resulting in incomplete identification of core regulatory nodes.
[0005] To achieve the above objectives, some embodiments of this application provide the following aspects:
[0006] In a first aspect, some embodiments of this application provide a bioinformatics analysis system for human tissue lipid metabolomics data, including the following modules:
[0007] The data deconstruction and initial importance assessment module is used to input the original lipid mass spectrometry dataset into the orthogonal partial least squares discriminant model for deconstruction, and extract the initial variable projection importance sequence and orthogonal loading matrix.
[0008] The biochemical homology physical distance mapping module is used to generate biochemical physical distances based on the absolute difference between carbon chain length and double bond number, and constructs a biochemical homology physical penalty matrix through an exponential decay function.
[0009] The orthogonal structure coupling and approximation optimization module is used to calculate the inner product of the orthogonal load matrix to generate an orthogonal covariance matrix, and to generate a structural orthogonal coupling matrix by performing a Hadamard product between the orthogonal covariance matrix and the biochemical homology physics penalty matrix. The module then performs singular value decomposition on the structural orthogonal coupling matrix according to the Eckart-Yang-Mirsky theorem to extract the rank-one structural principal axis vector.
[0010] The feature variance compensation and driving node output module is used to construct an orthogonal variance compensation factor sequence based on the rank-one structure principal axis vector, inject the orthogonal variance compensation factor sequence into the initial variable projection importance sequence to generate a final state compensation variable projection importance sequence, and extract the target lipid core driving factor based on the final state compensation variable projection importance sequence.
[0011] Secondly, some embodiments of this application also provide an electronic device, the electronic device comprising: one or more processors; and a memory storing computer program instructions, which, when executed, cause the processor to perform a bioinformatics analysis system for human tissue lipid metabolomics data as described above.
[0012] Thirdly, some embodiments of this application also provide a computer-readable medium having computer program instructions stored thereon, which can be executed by a processor to implement a bioinformatics analysis system for human tissue lipid metabolomics data as described above.
[0013] Fourthly, some embodiments of this application also provide a computer program product, including a computer program / instructions that, when executed by a processor, implement a bioinformatics analysis system for human tissue lipid metabolomics data as described above.
[0014] Compared with related technologies, the solution provided in this application introduces biochemical homology structure constraints and EYM theorem to perform optimal low-rank optimization on orthogonal loads, thereby realizing the extraction and compensation of implicit variance in orthogonal space. This improves the problem that traditional discriminant analysis mostly adopts the method of stripping and discarding orthogonal variance, which ignores the underlying regulatory information with physical correlation, resulting in incomplete identification of core regulatory nodes. Attached Figure Description
[0015] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.
[0016] Figure 1A flowchart of a bioinformatics analysis system module for human tissue lipid metabolomics data is provided for some embodiments;
[0017] Figure 2 A schematic diagram of a bioinformatics analysis system for human tissue lipid metabolomics data is provided for some embodiments;
[0018] Figure 3 A flowchart of data deconstruction and compensatory recombination of a bioinformatics analysis system for human tissue lipid metabolomics data is provided for some embodiments;
[0019] Figure 4 This is an electronic device hardware structure diagram of a bioinformatics analysis system for human tissue lipid metabolomics data, provided for some embodiments. Detailed Implementation
[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0021] First Embodiment
[0022] The first embodiment relates to a bioinformatics analysis system for human tissue lipid metabolomics data. For example... Figures 1-3 As shown, the system may include the following modules:
[0023] The data deconstruction and initial importance assessment module is used to input the original lipid mass spectrometry dataset into the orthogonal partial least squares discriminant model for deconstruction, and extract the initial variable projection importance sequence and orthogonal loading matrix.
[0024] The biochemical homology physical distance mapping module is used to generate biochemical physical distances based on the absolute difference between carbon chain length and double bond number, and constructs a biochemical homology physical penalty matrix through an exponential decay function.
[0025] The orthogonal structure coupling and approximation optimization module is used to calculate the inner product of the orthogonal load matrix to generate an orthogonal covariance matrix, and to generate a structural orthogonal coupling matrix by performing a Hadamard product between the orthogonal covariance matrix and the biochemical homology physics penalty matrix. The module then performs singular value decomposition on the structural orthogonal coupling matrix according to the Eckart-Yang-Mirsky theorem to extract the rank-one structural principal axis vector.
[0026] The feature variance compensation and driving node output module is used to construct an orthogonal variance compensation factor sequence based on the rank-one structure principal axis vector, inject the orthogonal variance compensation factor sequence into the initial variable projection importance sequence to generate a final state compensation variable projection importance sequence, and extract the target lipid core driving factor based on the final state compensation variable projection importance sequence.
[0027] By introducing biochemical homology constraints and the EYM theorem to perform optimal low-rank optimization on orthogonal loads, the implicit variance in orthogonal space is extracted and compensated. This improves the problem that traditional discriminant analysis often uses the method of stripping and discarding orthogonal variances, which ignores the underlying regulatory information with physical correlations, resulting in incomplete identification of core regulatory nodes.
[0028] The following is a detailed description of the implementation details of a bioinformatics analysis system for human tissue lipid metabolomics data in this embodiment. The following content is only for the convenience of understanding and is not necessary for implementing this solution.
[0029] The following sections will provide a detailed explanation of each of the above steps.
[0030] The data deconstruction and initial importance assessment module in this embodiment specifically includes:
[0031] The data destructuring and initial importance assessment module is used to perform:
[0032] Extract the absolute values of the original mass spectrum peak areas of each candidate lipid from the original lipid mass spectrometry dataset;
[0033] Perform a base-10 logarithmic transformation on the absolute values of the original mass spectrometry peak areas for each item to generate a logarithmic transformation matrix;
[0034] Extract the arithmetic mean and standard deviation of the peak area of each candidate lipid in the logarithmic transformation matrix;
[0035] The logarithmic transformation matrix is centered using the arithmetic mean of the peak area and the standard deviation to generate a standard input matrix;
[0036] Obtain the clinical phenotype label vector, and input the standard input matrix and the clinical phenotype label vector into the orthogonal partial least squares discrimination model;
[0037] In the orthogonal partial least squares discrimination model, an alternating partial least squares iterative operation is performed to decompose the global variance of the standard input matrix into a prediction subspace that is positively correlated with the clinical phenotype label vector and an orthogonal subspace that is not correlated with the clinical phenotype label vector.
[0038] Extract the prediction load vector from the prediction subspace, and generate the initial variable projection importance sequence based on the prediction load vector;
[0039] Orthogonal principal components are extracted from the orthogonal subspace, and the orthogonal loading matrix is generated based on the orthogonal principal components.
[0040] Specifically, the system can first perform nonlinear scaling on the input raw lipid mass spectrometry dataset. Specifically, it can extract the absolute values of the raw mass spectrometry peak areas of each candidate lipid in the raw lipid mass spectrometry dataset, perform a logarithmic transformation mapping to base 10, thereby pulling the skewed raw peak area data back to an approximately normal distribution surface, and reconstructing the logarithmic transformation matrix.
[0041] ;
[0042] in, Represents the first in the standard input matrix The first tissue sample The standardized abundance values of candidate lipids; Represents the absolute value of the original mass spectrum peak area in the original lipid mass spectrometry dataset; operator Represents the logarithmic function with base 10; Representing the The arithmetic mean of the peak areas of candidate lipids in all tissue samples after logarithmic transformation; Representing the The standard deviation of candidate lipids after logarithmic transformation across all tissue samples.
[0043] The technical role of performing the above-mentioned cascade preprocessing is to: compress the dynamic range difference of several orders of magnitude between high-abundance molecules and ultra-low-abundance molecules in mass spectrometry detection through logarithmic mapping; and then eliminate the system dimension differences of each candidate lipid due to different instrument response channels through centralization transformation, so that all candidate lipids can be mapped to a completely equivalent quantitative baseline for subsequent orthogonal space deconstruction.
[0044] The biochemical homology physical distance mapping module in this embodiment specifically includes:
[0045] The biochemical homology physical distance mapping module is used to perform:
[0046] Extract any two candidate lipids from the original lipid mass spectrometry dataset to form a comparison combination, and extract the carbon chain length and double bond number of any two candidate lipids in the comparison combination.
[0047] Calculate the absolute difference in carbon chain length and the absolute difference in double bond number between any two candidate lipids;
[0048] The absolute difference in carbon chain length and the absolute difference in the number of double bonds are summed to generate the biochemical physical distance between any two candidate lipids.
[0049] Introducing a biochemical difference decay constant, the biochemical physical distance is input into an exponential decay function and mapped to generate a single biochemical constraint coefficient;
[0050] Based on the arrangement dimensions of each candidate lipid, all individual biochemical constraint coefficients are arranged in matrix form to generate a biochemical homology physical penalty matrix.
[0051] Specifically, the system extracts any two candidate lipids from the original lipid mass spectrometry dataset to form a comparison pair, and extracts the carbon chain length and double bond number of the first and second candidate lipids in the comparison pair. Then, the system calculates the absolute difference in carbon chain length and the absolute difference in double bond number between the two molecules, and sums the two absolute differences to generate the biochemical physical distance corresponding to the two candidate lipids.
[0052] After obtaining the full distance baseline, the system introduces a biochemical difference attenuation constant as a control parameter. The calculated biochemical physical distance is input into an exponential attenuation function and nonlinearly mapped to generate a single biochemical constraint coefficient. The calculation formula for this mapping is defined as follows:
[0053] ;
[0054] in The system generates a representation of the first... Candidate lipids and the first A single biochemical constraint coefficient between candidate lipids, the value of which is between zero and one; Represents the natural constant; This represents a pre-defined biochemical difference attenuation constant, and its value must be greater than zero to ensure the convergence of the weights. Representing the Candidate lipids and the first The system calculates the biochemical physical distance between candidate lipids. Using the self-permutation dimension of each candidate lipid as the row and column index of the matrix, it aligns all the individual biochemical constraint coefficients generated by the traversal combination, and finally reconstructs and generates the biochemical homology physical penalty matrix.
[0055] The orthogonal structure coupling and approximation optimization module in this embodiment specifically includes:
[0056] The orthogonal structure coupling and approximation optimization module is used to perform the following when generating the orthogonal coupling matrix:
[0057] Obtain the transpose of the orthogonal load matrix, and perform matrix multiplication between the orthogonal load matrix and the transpose matrix to generate the orthogonal covariance matrix;
[0058] Extract the covariance matrix elements from the orthogonal covariance matrix, and extract the penalty matrix elements corresponding to the same row and column coordinate positions from the biochemical homology physical penalty matrix;
[0059] Perform the Hadamard product operation on the elements of the covariance matrix and the penalty matrix to generate the corresponding coupling elements;
[0060] Repeatedly perform the Hadamard product operation by traversing all row and column coordinate positions, and generate a structural orthogonal coupling matrix by arranging all coupled elements.
[0061] The orthogonal structure coupling and approximation optimization module is used to perform the following when extracting the principal axis vector of the rank-one structure:
[0062] Perform singular value decomposition on the structural orthogonal coupling matrix, decomposing the structural orthogonal coupling matrix into a left singular vector matrix, a singular value diagonal matrix, and a right singular vector matrix;
[0063] The target rank value for low-rank approximation of the structural orthogonal coupling matrix is set to one.
[0064] Extract the first principal singular value from the singular value diagonal matrix, and extract the first principal singular vector corresponding to the first principal singular value from the left singular vector matrix or the right singular vector matrix.
[0065] The extracted first principal singular vector is designated as the principal axis vector of the rank-one structure.
[0066] Specifically, the system first obtains the transpose of the orthogonal loading matrix, performs matrix multiplication on the orthogonal loading matrix and the transpose matrix, and calculates and generates an orthogonal covariance matrix to characterize the degree of mathematical covariance of candidate lipids in the orthogonal space.
[0067] Subsequently, the system extracts the covariance matrix elements from the orthogonal covariance matrix and the penalty matrix elements at the same row and column coordinate positions from the biochemical homology physics penalty matrix. A point-by-point Hadamard product operation is then performed on both to generate the corresponding coupling elements. This operation is repeated for all row and column coordinate positions, resulting in a structural orthogonal coupling matrix generated from the arrangement of all coupling elements. The formula for the Hadamard product operation is defined as follows:
[0068] ;
[0069] in This represents the orthogonal coupling matrix of the reconstructed structure; Represents the Hadama product operator; This represents the orthogonal covariance matrix generated by the aforementioned operations; This represents the physical penalty matrix for biochemical homology.
[0070] After obtaining the structural orthogonal coupling matrix, the system performs singular value decomposition on it, forcibly deconstructing it into a left singular vector matrix, a singular value diagonal matrix, and a right singular vector matrix. At this point, the system sets the target rank value for low-rank approximation of the structural orthogonal coupling matrix to one, extracts the first principal singular value located in the first row and first column of the singular value diagonal matrix, and extracts the one-dimensional column vector corresponding to this first principal singular value from the left singular vector matrix, thus designating it as the rank-one structural principal axis vector with the minimum global error.
[0071] The feature variance compensation and driving node output module in this embodiment specifically includes:
[0072] The feature variance compensation and driving node output module is used to perform the following when constructing the orthogonal variance compensation factor sequence:
[0073] Extract the component values corresponding to each candidate lipid dimension from the principal axis vector of the rank-one structure;
[0074] Perform square norm operation on each component value;
[0075] The principal singular values generated by performing singular value decomposition on the structural orthogonal coupling matrix are extracted. The numerical results output by the square norm operation are then multiplied by the principal singular values to generate the orthogonal variance compensation factor corresponding to a single candidate lipid.
[0076] Based on the arrangement dimension of the initial variable projection importance sequence, the orthogonal variance compensation factors corresponding to each candidate lipid are sorted and combined to generate an orthogonal variance compensation factor sequence.
[0077] The feature variance compensation and driving node output module is used to perform the following when performing weight adjustment:
[0078] Introduce a compensating gain coefficient and perform a scaling multiplication operation on the compensating gain coefficient and the orthogonal variance compensating factor sequence;
[0079] The orthogonal variance compensation factor sequence after scaling multiplication is linearly added to the initial variable projection importance sequence in the corresponding dimension to generate the final state compensation variable projection importance sequence.
[0080] Based on the magnitude of each value in the final state compensation variable projection importance sequence, all candidate lipids are sorted in descending order;
[0081] Candidate lipids that are sorted in descending order and ranked above a preset importance threshold are extracted and output as the core driving factors of targeted lipids.
[0082] Specifically, the system extracts the component values corresponding to each candidate lipid dimension from the principal axis vector of the rank-one structure, and performs a square norm operation on each component value to eliminate the sign uncertainty of the singular vector in spatial orientation. Next, the system obtains the first principal singular value generated by the aforementioned singular value decomposition, performs a scalar multiplication operation between the numerical result output by the square norm operation and this first principal singular value, and generates an orthogonal variance compensation factor corresponding to a single candidate lipid. The calculation formula for the variance compensation factor is defined as follows:
[0083] ;
[0084] in The generated representation is for the first Orthogonal variance compensation factor for candidate lipids; The first principal singular value obtained by performing singular value decomposition on the orthogonal coupling matrix of the structure represents the total variance energy intensity of the structural regulation features in the orthogonal space. Represents the corresponding rank-one principal axis vector in the rank-one structure The system calculates the component values of each candidate lipid dimension based on the arrangement dimensions of the initial variable projection importance sequence, sorts and combines the calculation results of each component, and generates a complete orthogonal variance compensation factor sequence.
[0085] In the weight adjustment phase at the end of the system, a compensatory gain coefficient is introduced as a proportional adjustment operator. The value of this compensatory gain coefficient is set in the real number range greater than zero. The system performs a scaling multiplication operation on the compensatory gain coefficient and each element in the orthogonal variance compensatory factor sequence, and then performs a linear summation on the corresponding dimension of the scaled sequence and the initial variable projection importance sequence, thereby assigning each lipid molecule a final-state compensatory variable projection importance after orthogonal energy rescue. Finally, based on the magnitude of each value in the final-state compensatory variable projection importance, the system sorts all candidate lipids in descending order, and selects candidate lipids whose sorting position is higher than a preset importance threshold as the final targeted lipid core driving factor output.
[0086] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed in this application; however, this does not mean that other units are absent from this embodiment.
[0087] Furthermore, some embodiments of this application also provide an electronic device. The electronic device can be various forms of digital computer, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, mainframe computers, cellular phones, smartphones, wearable devices, and other similar computing devices.
[0088] The electronic device includes: one or more processors; and a memory storing computer program instructions, which, when executed, cause the processor to perform all the module functions covered by the bioinformatics analysis system for human tissue lipid metabolomics data provided in any one or more of the above embodiments.
[0089] Figure 4 An exemplary structural diagram of the electronic device is disclosed. The electronic device includes one or more processors 1101, a memory 1102, an input device 1103, and an output device 1104. The various components are interconnected via a bus or other means (the diagram shows an example of bus connection). The processor 1101 can be used to execute instructions stored in the memory 1102 to control the overall operation of the electronic device. The memory 1102 may include a program storage area and a data storage area, wherein the program storage area stores the operating system and applications required for at least one function; the data storage area stores data created according to the use of the electronic device, etc. The memory 1102 may include high-speed random access memory and may also include non-transitory memory, such as disk storage devices, flash memory devices, or other non-transitory solid-state storage devices. In some embodiments, the memory 1102 may also include storage resources located remotely to the processor and accessible via a network.
[0090] Input device 1103 can be used to receive input numerical or character information or user operation signals, such as a touch screen, keypad, mouse, trackpad, touchpad, indicator, one or more mouse buttons, trackball, joystick, etc. Output device 1104 may include display devices (such as liquid crystal displays, light-emitting diode displays, plasma displays, and optional touch screens), auxiliary lighting devices (such as LEDs), and haptic feedback devices (such as vibration motors), etc.
[0091] To facilitate user interaction, the electronic device may be configured to include a display device (such as an LCD or CRT monitor) and input devices such as a keyboard and pointing devices (e.g., a mouse or touchpad). Feedback can be any form of sensory feedback (e.g., visual feedback, auditory feedback); input may also be received via voice, touch, or other means.
[0092] This application also relates to a computer-readable medium storing a computer program / instructions thereon, which, when executed by a processor, implements all the module functions covered by the bioinformatics analysis system for human tissue lipid metabolomics data provided in any one or more of the above embodiments. This computer-readable medium may be a memory included in an electronic device, or it may be a standalone storage medium not assembled into the device.
[0093] It should be noted that the computer-readable medium described in this application may be a computer-readable signal medium, a computer-readable storage medium, or a combination of both. Examples include, but are not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. Specific examples of storage media may include, but are not limited to, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory, optical fibers, portable CD-ROMs, optical storage devices, magnetic storage devices, etc., or any suitable combination thereof.
[0094] Computer-readable media may store one or more programs that can be used by or in conjunction with an instruction execution system. The media may be permanent or non-permanent, removable or non-removable, and may store information by any method or technology, including computer-readable instructions, data structures, program modules, or other data.
[0095] The computer program code used to implement the operations of this application can be written in any combination of one or more programming languages, including object-oriented programming languages (such as Java, Smalltalk, and C++) and conventional procedural programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer, partially on a remote computer, or entirely on a remote computer or server. The remote computer can be connected to the user's computer via any network (including a local area network or a wide area network) or can be connected to an external computer.
[0096] In the above embodiments, the functions can be implemented in whole or in part by software, hardware, firmware, or any combination thereof, for example, by using application-specific integrated circuits, general-purpose computers, or other similar hardware devices. In some embodiments, the software program of this application can be executed by a processor to implement the modules or functions; it can also be implemented by hardware, for example as a circuit that works in conjunction with the processor to execute the modules or functions.
[0097] This application also provides a computer program product, including one or more computer programs / instructions, which, when executed by a processor, generate all or part of the processes or functions described in this application. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium or transmitted from one storage medium to another via wired (e.g., DSL) or wireless (e.g., wireless, microwave) means. The computer-readable storage medium may be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive).
[0098] The flowcharts or block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of devices, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-specific system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0099] The scope of this application is defined by the appended claims rather than the foregoing description, and is therefore intended to encompass all variations falling within the meaning and scope of equivalents of the claims. No reference numerals in the claims should be construed as limiting the scope of the claims. Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a device claim may also be implemented by a single unit or device in software or hardware. Terms such as "first," "second," etc., are used only for distinguishing descriptions and do not indicate any particular order, nor should they be construed as indicating or implying relative importance.
[0100] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily made by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims, and the above embodiments should be regarded as exemplary and non-limiting.
Claims
1. A bioinformatics analysis system for human tissue lipid metabolomics data, characterized in that, Includes the following modules: The data deconstruction and initial importance assessment module is used to input the original lipid mass spectrometry dataset into the orthogonal partial least squares discrimination model for deconstruction, and extract the initial variable projection importance sequence and orthogonal loading matrix for each candidate lipid. The biochemical homology physical distance mapping module is used to generate biochemical physical distances based on the absolute difference between the carbon chain length and the number of double bonds of each candidate lipid, and to construct a biochemical homology physical penalty matrix through an exponential decay function. The orthogonal structure coupling and approximation optimization module is used to calculate the inner product of the orthogonal load matrix to generate an orthogonal covariance matrix, perform Hadamard product on the orthogonal covariance matrix and the biochemical homology physics penalty matrix to generate a structural orthogonal coupling matrix, and perform singular value decomposition on the structural orthogonal coupling matrix to perform low-rank approximation and extract the rank-one structural principal axis vector. The feature variance compensation and driving node output module is used to construct an orthogonal variance compensation factor sequence based on the rank-one structure principal axis vector, inject the orthogonal variance compensation factor sequence into the initial variable projection importance sequence, perform compensation recombination through weight adjustment, generate a final state compensation variable projection importance sequence, and extract the target lipid core driving factor based on the final state compensation variable projection importance sequence.
2. The bioinformatics analysis system for human tissue lipid metabolomics data according to claim 1, characterized in that, The data deconstruction and initial importance assessment module is used to perform: Extract the absolute values of the original mass spectrum peak areas of each candidate lipid from the original lipid mass spectrometry dataset; Perform a base-10 logarithmic transformation on the absolute values of the original mass spectrometry peak areas for each item to generate a logarithmic transformation matrix; Extract the arithmetic mean and standard deviation of the peak area of each candidate lipid in the logarithmic transformation matrix; The logarithmic transformation matrix is centered using the arithmetic mean of the peak area and the standard deviation to generate a standard input matrix; Obtain the clinical phenotype label vector, and input the standard input matrix and the clinical phenotype label vector into the orthogonal partial least squares discrimination model; In the orthogonal partial least squares discrimination model, an alternating partial least squares iterative operation is performed to decompose the global variance of the standard input matrix into a prediction subspace that is positively correlated with the clinical phenotype label vector and an orthogonal subspace that is not correlated with the clinical phenotype label vector. Extract the prediction load vector from the prediction subspace, and generate the initial variable projection importance sequence based on the prediction load vector; Orthogonal principal components are extracted from the orthogonal subspace, and the orthogonal loading matrix is generated based on the orthogonal principal components.
3. The bioinformatics analysis system for human tissue lipid metabolomics data according to claim 1, characterized in that, The biochemical homology physical distance mapping module is used to perform: From the original lipid mass spectrometry dataset, any two candidate lipids are extracted to form a comparison combination, and the carbon chain length and double bond number of any two candidate lipids in the comparison combination are extracted. Calculate the absolute difference in carbon chain length and the absolute difference in the number of double bonds between any two candidate lipids; The absolute difference in carbon chain length and the absolute difference in double bond number are summed to generate the biochemical physical distance corresponding to any two candidate lipids. A biochemical difference decay constant is introduced, and the biochemical physical distance is input into the exponential decay function for mapping to generate a single biochemical constraint coefficient; Based on the arrangement dimensions of each candidate lipid, all the single biochemical constraint coefficients are arranged in matrix form to generate the biochemical homology physical penalty matrix.
4. The bioinformatics analysis system for human tissue lipid metabolomics data according to claim 1, characterized in that, The orthogonal structure coupling and approximation optimization module is used to perform the following when generating the orthogonal structure coupling matrix: Obtain the transpose of the orthogonal load matrix, and perform matrix multiplication between the orthogonal load matrix and the transpose matrix to generate the orthogonal covariance matrix; Extract the covariance matrix elements from the orthogonal covariance matrix, and extract the penalty matrix elements corresponding to the same row and column coordinate positions from the biochemical homology physical penalty matrix; The Hadamard product operation is performed on the elements of the covariance matrix and the elements of the penalty matrix to generate corresponding coupling elements; Repeatedly perform the Hadamard product operation by traversing all row and column coordinate positions, and generate the structural orthogonal coupling matrix by arranging all the coupling elements.
5. The bioinformatics analysis system for human tissue lipid metabolomics data according to claim 1, characterized in that, The orthogonal structure coupling and approximation optimization module is used to perform the following when extracting the principal axis vector of the rank-one structure: Perform the singular value decomposition on the structural orthogonal coupling matrix to decompose the structural orthogonal coupling matrix into a left singular vector matrix, a singular value diagonal matrix, and a right singular vector matrix; The target rank value for low-rank approximation of the orthogonal coupling matrix of the structure is set to one. Extract the first principal singular value from the singular value diagonal matrix, and extract the first principal singular vector corresponding to the first principal singular value from the left singular vector matrix or the right singular vector matrix; The extracted first principal singular vector is designated as the principal axis vector of the rank-one structure.
6. The bioinformatics analysis system for human tissue lipid metabolomics data according to claim 5, characterized in that, The feature variance compensation and driving node output module is used to perform the following when constructing the orthogonal variance compensation factor sequence: Extract the component values corresponding to each candidate lipid dimension from the principal axis vector of the rank-one structure; Perform square norm operation on each of the component values; Extract the principal singular values generated by performing the singular value decomposition on the orthogonal coupling matrix of the structure, and perform scalar multiplication operation between the numerical result output by the square norm operation and the principal singular values to generate the orthogonal variance compensation factor corresponding to a single candidate lipid. Based on the arrangement dimension of the initial variable projection importance sequence, the orthogonal variance compensation factors corresponding to each candidate lipid are sorted and combined to generate the orthogonal variance compensation factor sequence.
7. The bioinformatics analysis system for human tissue lipid metabolomics data according to claim 1, characterized in that, The feature variance compensation and driving node output module, when performing the weight adjustment, is used to perform: A compensation gain coefficient is introduced, and the compensation gain coefficient is multiplied by the orthogonal variance compensation factor sequence by scaling operation; The orthogonal variance compensation factor sequence, which has undergone scaling multiplication, is linearly added to the initial variable projection importance sequence along the corresponding dimension to generate the final state compensation variable projection importance sequence. Based on the magnitude of each value in the final state compensation variable projection importance sequence, all candidate lipids are sorted in descending order. Candidate lipids that are sorted in descending order and whose importance is higher than a preset threshold are extracted and output as the core driving factor of the targeted lipids.
8. An electronic device, characterized in that, The electronic device includes: one or more processors; and a memory storing computer program instructions, which, when executed, cause the processor to perform all the module functions covered by the bioinformatics analysis system for human tissue lipid metabolomics data as described in any one of claims 1 to 7.
9. A computer-readable medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement all the module functions covered by the bioinformatics analysis system for human tissue lipid metabolomics data as described in any one of claims 1 to 7.
10. A computer program product comprising a computer program / instructions, characterized in that, When executed by a processor, the computer program / instructions implement all the module functions covered by the bioinformatics analysis system for human tissue lipid metabolomics data as described in any one of claims 1 to 7.