High-resolution mass spectrum-deep learning driven metabonomics data automatic analysis method

Abstract

The invention relates to a high-resolution mass spectrum-deep learning driven metabonomics data automatic analysis method, and belongs to the technical field of data science. The method comprises the following steps: acquiring an original mass spectrum data file through a liquid chromatogram-high resolution mass spectrum coupling technology; performing peak extraction, peak alignment and de-noising processing on the original mass spectrum data file through an automatic pre-processing module, and generating a standardized metabolite characteristic quantitative matrix based on a characteristic extraction module; based on the standardized metabolite characteristic quantitative matrix and secondary mass spectrum information in the original mass spectrum data file, automatically identifying metabolite characteristics through an adaptive identification model; and for the identified metabolites and the corresponding metabolite characteristic quantitative matrix, automatically matching optimal results in a database, and obtaining corresponding molecular formula compositions and compound names. The metabolite identification speed and precision are obviously improved, and errors possibly caused by manual intervention are reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of data science, and particularly relates to a high-resolution mass spectrometry-deep learning driven metabolomics data automatic analysis method. BACKGROUND

[0002] Current metabolomics research widely adopts liquid chromatography-high resolution mass spectrometry (LC-HRMS) technology, and its data analysis process includes: Raw data preprocessing: peak extraction and alignment are performed through commercial software (such as XCMS, MS-DIAL), but the algorithm depends on fixed parameters, and the retention time drift needs to be manually adjusted; Metabolite identification: relies on manual comparison of experimental MS / MS spectrum and public standard database (such as HMDB, MassBank), and the matching result needs to be reviewed by experts for isomer and isomer; Downstream analysis: use independent tools (such as MetaboAnalyst) for pathway enrichment, which is disconnected with the data in the identification link.

[0003] The existing technical problems at this stage are: 1. Insufficient efficiency and accuracy of preprocessing: the traditional peak alignment algorithm (such as LOESS) is sensitive to non-linear drift of retention time, and needs to optimize parameters repeatedly; the denoising process lacks standardized quality control (such as not forced RSD filtering), which leads to low reproducibility of features mixed into the quantitative matrix.

[0004] 2. Low degree of metabolite identification automation: the existing method cannot jointly use the primary quantitative characteristics (parent ion intensity / isotope pattern) and secondary structure information, leading to the dependence of manual inspection of ion abundance ratio for distinguishing similar compounds (such as structural isomers), and the low-abundance metabolites are missed due to poor MS / MS spectrum quality.

[0005] 3. Database coverage and matching reliability defects: the public spectrum library has limited coverage (especially new metabolites), and the virtual fragment spectrum is not integrated; the matching result lacks multi-dimensional verification (such as isotope distribution, parent ion intensity), and the false matching rate is high; there is no adaptive processing mechanism (such as expanding the mass-to-charge ratio tolerance range) for low confidence results.

[0006] 4. Analysis process fragmentation: preprocessing, identification and pathway analysis need to be operated across platforms, and manual data transfer introduces errors; there is no end-to-end automatic output (from raw data to compound name / pathway map).

[0007] 5. Model generalization and computational efficiency bottleneck: traditional identification model cannot incrementally learn, and new data types need to be retrained; CPU processing of massive mass spectrometry data is slow, and it is difficult to meet the high-throughput demand of samples. SUMMARY

[0008] To solve the above problems existing in the prior art, the application provides a high-resolution mass spectrometry-depth learning driven metabolomics data automatic analysis method; the purpose of the application can be achieved by the following technical scheme, comprising: S1: obtaining an original mass spectrum data file through liquid chromatography-high resolution mass spectrometry; performing peak extraction, peak alignment, and denoising processing on the original mass spectrum data file through an automatic preprocessing module, and generating a standardized metabolite feature quantitative matrix based on a feature extraction module; S2: based on the standardized metabolite feature quantitative matrix and the secondary mass spectrum information in the original mass spectrum data file, automatically identifying metabolite features through an adaptive recognition model; S3: automatically matching the best result in the database for the identified metabolites and the corresponding metabolite feature quantitative matrix, and obtaining the corresponding molecular formula composition and compound name.

[0009] Specifically, the database matching in S3 includes the following steps: Based on the mass-to-charge ratio and the retention time of the metabolite feature quantitative matrix, preliminary screening is performed in the local and cloud metabolite standard library; Real-time comparison is performed between the experimental spectrum and the standard spectrum of the candidate compound screened out, and the spectrum similarity score is calculated through a deep learning model; According to the similarity score, the parent ion, the consistency of the ion intensity ratio, and the isotopic distribution fitting degree, a comprehensive confidence score is generated; The matching result with a comprehensive confidence score higher than a preset threshold is automatically selected, and the molecular formula and the compound name are output.

[0010] Specifically, the peak alignment in S1 adopts a dynamic time warping algorithm based on retention time correction, and the denoising processing includes blank sample background deduction and mixed sample relative standard deviation filtering.

[0011] Specifically, the input data of the adaptive recognition model in S2 includes: The parent ion pair intensity ratio and the isotopic distribution pattern extracted from the metabolite feature quantitative matrix; The sub-ion intensity sequence obtained by vectorizing the secondary mass spectrum information.

[0012] Further, the deep learning model fuses the primary mass spectrum feature vector and the secondary spectrum feature vector through a dot product vector algorithm.

[0013] Further, the metabolite standard library includes a self-built standard sample experimental library, a public database, and an algorithm prediction virtual fragment spectrum library, wherein the virtual fragment spectrum is generated by a spectrum prediction tool.

[0014] Specifically, the adaptive recognition model in S2 deploys an online updating mechanism, and automatically triggers incremental training based on user feedback of identification result error samples.

[0015] Specifically, S4 further comprises: based on the identification result, automatically generating a metabolic pathway visualization atlas and labeling the enrichment significance of the differential metabolites in the pathway.

[0016] Specifically, when the similarity score of the optimal result is lower than the preset threshold, the mass-to-charge ratio error tolerance range is expanded and the isomer distinguishing module is called to analyze the characteristic ion abundance ratio.

[0017] The method of the present application has significant advantages. In the pretreatment link, based on the dynamic time warping algorithm and standardized denoising processing of retention time correction, the efficiency and accuracy of pretreatment are effectively improved, avoiding the sensitivity problem of traditional algorithms to nonlinear drift of retention time and the defect of lack of standardized control of denoising. In the aspect of metabolite identification, the adaptive recognition model can jointly use the primary quantitative features and secondary structure information to improve the discrimination ability of similar compounds and reduce the need for manual inspection, while the online updating mechanism ensures that the model can continuously learn and optimize. In the database matching process, the comprehensive confidence score is generated by combining multiple dimensions of information, increasing the reliability of the matching result, and having an adaptive processing mechanism for low confidence results. In addition, the entire analysis process realizes end-to-end automation, directly outputs the compound name and metabolic pathway visualization atlas from the raw data, avoiding errors introduced by data fragmentation and manual data transmission. In summary, the high-resolution mass spectrometry - deep learning driven metabolomics data automatic analysis method of the present application provides an efficient, accurate and highly automated solution for metabolomics research. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to facilitate the understanding of those skilled in the art, the present application will be further described below in conjunction with the drawings.

[0019] Fig. 1 A flowchart of the high-resolution mass spectrometry - deep learning driven metabolomics data automatic analysis method of the present application.

[0020] Fig. 2 A data processing operation diagram of the present application. DETAILED DESCRIPTION

[0021] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific embodiments, structures, features and effects according to the present application are described in detail below in conjunction with the drawings and preferred embodiments.

[0022] Please refer to Figs. 1-2A high-resolution mass spectrometry-deep learning driven metabolomics data automatic analysis method, comprising: S1: obtaining an original mass spectrometry data file by liquid chromatography-high resolution mass spectrometry technology; performing peak extraction, peak alignment, and noise removal processing on the original mass spectrometry data file by an automatic preprocessing module, and generating a standardized metabolite feature quantitative matrix based on a feature extraction module; S2: based on the standardized metabolite feature quantitative matrix and the secondary mass spectrum information in the original mass spectrometry data file, automatically identifying metabolite features by an adaptive identification model; S3: automatically matching the best result in the database for the identified metabolites and the corresponding metabolite feature quantitative matrix, and obtaining the corresponding molecular formula composition and compound name.

[0023] In this embodiment, the automatic preprocessing module uses a dynamic time warping algorithm based on retention time correction for peak alignment, which can effectively handle the nonlinear drift of retention time and avoid the problems sensitive to traditional algorithms. In the noise removal process, blank sample deduction and relative standard deviation filtering are strictly implemented to ensure the high quality of the quantitative matrix. After receiving the parent ion intensity ratio extracted from the metabolite feature quantitative matrix and the isotope distribution pattern, as well as the vectorized secondary mass spectrum sub-ion intensity sequence, the adaptive identification model can accurately identify the metabolite features. Its online updating mechanism can trigger incremental training in time according to user feedback of identification errors, so that the model always maintains high accuracy and adaptability.

[0024] In the database matching step, first, based on the mass-to-charge ratio and retention time of the metabolite feature quantitative matrix, perform preliminary screening in the self-built standard sample library, public database, and algorithm prediction virtual fragment spectrum library. The self-built standard sample library combines the data of standard samples in actual research, the public database ensures a wide coverage range, and the virtual fragment spectrum is calculated by a spectrum prediction tool to make up for the lack of coverage of the public spectrum library. After real-time comparison of the experimental spectrum with the reference spectrum of the candidate compounds, the deep learning model fuses the primary mass spectrometry feature vector and the secondary spectrum feature vector by the dot product vector algorithm to calculate the spectrum similarity score. Then, according to the similarity score, ion intensity ratio consistency, and isotope distribution coincidence, a comprehensive confidence score is generated, and the matching result above the preset threshold is output to obtain the molecular formula and compound name. If the similarity score of the best matching result is lower than the preset threshold, the system will expand the mass-to-charge ratio error tolerance range and call the isomer differentiation module to analyze the feature ion abundance ratio to ensure effective processing of low confidence results.

[0025] After the metabolite identification is completed, the system automatically generates a metabolic pathway visualization map based on the identification results and labels the enrichment significance of the differential metabolites in the pathway. Through these steps, the method of the application truly realizes the end-to-end automation of metabolomics data from raw data to final analysis results, providing great convenience and efficient solutions for metabolomics research.

[0026] Specifically, the database matching in S3 includes the following steps: Based on the mass-to-charge ratio and retention time of the metabolite feature quantitative matrix, preliminary screening is performed in the local and cloud metabolite standard library; Real-time comparison of experimental spectra with standard spectra of candidate compounds is performed, and a spectrum similarity score is calculated by a deep learning model; A comprehensive confidence score is generated according to the similarity score, ion intensity ratio consistency of parent ions and daughter ions, and isotope distribution fit; The matching results with a comprehensive confidence score higher than a preset threshold are automatically selected, and the molecular formula and compound name are output.

[0027] Specifically, the peak alignment in S1 uses a dynamic time warping algorithm based on retention time correction, and the denoising process includes blank sample background subtraction and mixed sample relative standard deviation filtering.

[0028] Specifically, the input data of the adaptive recognition model in S2 includes: The parent ion pair intensity ratio and isotope distribution pattern extracted from the metabolite feature quantitative matrix; The sub-ion intensity sequence obtained by vectorizing the secondary mass spectrum information.

[0029] Further, the deep learning model fuses the primary mass spectrum feature vector and the secondary spectrum feature vector through a dot product vector algorithm.

[0030] Further, the metabolite standard library includes a self-built standard sample library, a public database, and a virtual fragment spectrum library predicted by an algorithm, wherein the virtual fragment spectrum is generated by a spectrum prediction tool.

[0031] Specifically, the adaptive recognition model in S2 is deployed with an online update mechanism, and automatically triggers incremental training based on user feedback of incorrect samples.

[0032] Specifically, S4 further includes: based on the identification results, automatically generating a metabolic pathway visualization map, and labeling the enrichment significance of differential metabolites in the pathway.

[0033] Specifically, when the similarity score of the best result is lower than the preset threshold, the mass-to-charge ratio error tolerance range is expanded and the isomer differentiation module is called to analyze the characteristic ion abundance ratio.

[0034] In this embodiment, human serum metabolomics analysis is taken as an example, and 50 cases of healthy people / 50 cases of liver cancer patients are collected; Liquid chromatography and high-resolution mass spectrometry are collected by instruments; Real-time data preprocessing is performed, ion peaks with signal-to-noise ratio>3 are extracted, an initial feature list is generated, dynamic time warping algorithm (DTW) is used to correct retention time drift, and maximum tolerance deviation is set; denoising processing is used to remove background pollution features, delete features with RSD>30% in QC samples, and retain high reproducibility metabolites; finally, output standardized metabolite feature quantitative matrix (including multi-dimensional features).

[0035] Metabolite identification is performed, and model input is constructed: Primary features: extract parent ion intensity ratio (such as [M+H]+ / [M+Na]+) and 13C isotope distribution pattern; Secondary spectrum: MS / MS sub-ion intensity sequence is vectorized into a 256-dimensional tensor; A variety of metabolite features are identified; Database matching and result optimization: through self-built standard sample library, public databases such as HMDB and theoretical prediction database, candidate compounds are preliminarily screened, and comprehensive score is used for judgment, and the pseudo code for comprehensive score calculation is: { #Pseudo code: comprehensive confidence score formula final_score = 0.5*cosine_similarity(exp_ms2, ref_ms2)#spectrum similarity + 0.25*ion_ratio_consistency#ion intensity ratio consistency + 0.25*isotope_pattern_match#isotope distribution consistency } Based on the comprehensive score, metabolites with high confidence are output, and the molecular formula and compound name of the database are associated; Downstream analysis is performed by screening differential metabolites, KEGG pathway analysis shows that glycine / serine metabolism is significantly enriched, and a pathway map is automatically generated and labeled with differential metabolites.

[0036] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application in any form. Although the present application has been disclosed with the preferred embodiments as above, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, as long as the changes or modifications do not deviate from the technical solution of the present application. Any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application still belongs to the scope of the technical solution of the present application.

Claims

1. A high-resolution mass spectrometry-deep learning driven automatic analysis method for metabolomics data, characterized in that: include: S1: The original mass spectrometry data files were obtained by liquid chromatography-high resolution mass spectrometry; Performing peak extraction, peak alignment, and denoising on the raw mass spectrometry data file through an automated preprocessing module, and generating a standardized metabolite feature quantitative matrix based on a feature extraction module; S2: Based on the standardized metabolite feature quantitative matrix and the secondary mass spectrometry spectrum information in the original mass spectrometry data file, automatically identifying the metabolite features through an adaptive recognition model; S3: For the identified metabolites and the corresponding metabolite feature quantitative matrix, the best results are automatically matched in the database, and the corresponding molecular formula composition and compound name are obtained.

2. The method according to claim 1, characterized in that Database matching in S3 involves the following steps: Based on the mass-to-charge ratio and retention time of the metabolite feature quantitative matrix, a preliminary screening is performed in the local and cloud metabolite standard library; Compare the experimental spectra with the standard spectra of the primary screening candidate compounds in real time, and calculate the spectrum similarity score through the deep learning model; Generate a comprehensive confidence score based on the similarity score, ion intensity ratio consistency and isotope distribution consistency; Automatically select matching results with a comprehensive confidence score higher than the preset threshold and output the molecular formula and compound name.

3. The method according to claim 1, characterized in that Peak alignment in S1 uses a dynamic time warping algorithm based on retention time correction, and the denoising process includes blank sample background subtraction and mixed sample relative standard deviation filtering.

4. The method according to claim 1, wherein The adaptive recognition model input data in S2 includes: Fragment peak area values ​​and isotope distribution patterns extracted from the metabolite feature quantification matrix; The fragment ion intensity sequence of the secondary mass spectrometry information is vectorized.

5. The method according to claim 2, characterized in that The deep learning model fuses the primary mass spectrum feature vector and the secondary spectrum feature vector through the dot vector product algorithm.

6. The method according to claim 2, characterized in that The metabolite standard library includes a self-built standard experimental library, a public database and an algorithm-predicted virtual fragment spectrum library, wherein the virtual fragment spectrum is calculated and generated by a spectrum prediction tool.

7. The method according to claim 1, characterized in that The adaptive recognition model in S2 deploys an online update mechanism, which automatically triggers incremental training based on erroneous samples of identification results fed back by users.

8. The method according to claim 1, characterized in that S4 further includes: automatically generating a metabolic pathway visualization map based on the identification results, and annotating the enrichment significance of differential metabolites in the pathway.

9. The method according to claim 1, characterized in that When the similarity score of the best result is lower than a preset threshold, the mass-to-charge ratio error tolerance range is expanded and the isomer differentiation module is called to analyze the characteristic ion abundance ratio.

Images