Age estimation method based on multi-dimensional data

Abstract

The invention relates to an age estimation method based on multi-dimensional data, in particular to a data set establishing method, an age estimation model training method, an age estimation method, an age estimation model or device, electronic equipment, a nonvolatile computer readable storage medium and application thereof. The method can more accurately estimate the age of Asian people, especially the age of East Asia or Southeast Asia, and can be used for evaluating senescence-related information.

Description

Technical Field

[0001] The present invention relates to the biomedical field, and in particular to an age estimation method based on multidimensional data. Background Art

[0002] Aging is a complex process affected by multiple factors, including genetics and the environment, and is considered to be the root cause of many chronic diseases (such as cardiovascular diseases, neurodegenerative diseases, diabetes, or chronic obstructive pulmonary disease). Therefore, quantifying the aging rate will help evaluate disease risk and health level and provide a basis for personalized medicine. In addition, a variety of aging intervention strategies have been reported to slow down the aging rate to a certain extent, such as dietary restrictions and exercise, and a method to evaluate the degree of aging (i.e., biological age) is the basis for verifying the effectiveness of aging intervention measures. However, aging is a highly heterogeneous process, both in the population and within individuals, and has no absolute correlation with chronological age (i.e., the year from birth to the time of recording), that is, human biological age is not equal to chronological age. Therefore, the accurate evaluation of aging has become an emerging key scientific issue in aging research, and the development of tools and methods that can effectively evaluate the degree of aging is of great significance for understanding and intervening in aging.

[0003] There are two main methods for evaluating the degree of aging. The first is to collect relevant biological data from the population, conduct correlation analysis on the collected data and the individual's chronological age, and obtain detection indicators that are highly correlated with age, which are used to preliminarily evaluate the degree of aging of the individual. The second is to establish a prediction model for the aging clock. Its technical route can be basically summarized as follows: Figure 1 ): 1) Collect relevant biological data (such as whole blood DNA methylation, blood cell transcriptome, blood protein / metabolome, etc.) in the population; 2) Based on the collected data matrix, use chronological age as a label to establish an aging clock prediction model for chronological age; 3) Apply the prediction model in the validation set to obtain the predicted age of different individuals to evaluate their aging rate. However, the existing methods are basically based on single-level information, such as phenotype, transcription, epigenetic, protein, etc. The inherent heterogeneity of aging makes these single-level evaluation methods have certain limitations. In addition, most of the current studies are based on population data in European and American countries, and aging may be affected by environmental, ethnic and other factors. If the existing models are directly applied to Asian populations, it may affect the accuracy of the prediction results. In addition, the readability of the prediction results is limited, and it is difficult to directly combine the predicted age with specific biological functions. Summary of the invention

[0004] The present invention establishes an aging evaluation system based on the Chinese population through multidimensional data, which can more accurately evaluate the biological age and specific biological functions of the subjects.

[0005] In a first aspect, the present invention provides a method for establishing a data set, comprising:

[0006] (1) performing clinical phenotype testing on multiple subjects to obtain a clinical phenotype measurement data set;

[0007] (2) collecting biological samples from the multiple subjects, performing the following analysis on the biological samples, and obtaining a gene expression dataset, a protein abundance dataset, and a metabolite abundance dataset based on the analysis results:

[0008] 1) Transcriptome sequencing to obtain the expression level of each gene in the biological sample;

[0009] 2) Proteomic quantitative analysis to obtain the abundance of each protein in the biological sample;

[0010] 3) Quantitative metabolomics analysis to obtain the abundance of each metabolite in the biological sample;

[0011] (3) The four candidate datasets in (1) and (2) are combined, and age is used as a training label to construct the dataset for training an age estimation model.

[0012] In some embodiments, the clinical phenotype detection includes a physical examination (for example, one or more of a general examination, an internal medicine examination, a surgical examination, an ultrasound examination, a body composition analysis, a routine blood test, glycosylated hemoglobin, six items of sex hormones, a routine urine test, a routine stool test, a lung function test, a bone density test, an electrocardiogram, and a biochemical analysis), and / or a mobility test (for example, one or more of a single-leg stand, a grip strength test, a 30-second sit-and-stand test, a Purdue pegboard test, and a stepping test).

[0013] In some embodiments, the biological sample is selected from blood (eg, whole blood, plasma, or serum), urine, saliva, cells, and tissue extracts.

[0014] In some embodiments, the gene expression dataset is obtained by the following method:

[0015] i) obtaining immune cells from the biological sample;

[0016] ii) performing transcriptome sequencing on the immune cells;

[0017] iii) based on the transcriptome sequencing results, obtaining the expression values ​​of all genes in the immune cells.

[0018] In some embodiments, the immune cells are subjected to tissue transcriptome sequencing, single cell transcriptome sequencing, microarray detection, or flow cytometric analysis.

[0019] In some embodiments, RNA is extracted from the immune cells, an RNA-seq sequencing library is constructed (for example, using a VAHTS Universal V6 RNA-seq Library Prep Kit), and paired-end sequencing is performed (for example, using an Illumina Hisses X-Ten sequencing platform to perform the paired-end sequencing) to obtain a sequencing file with a read length of 150 bp.

[0020] In some embodiments, the sequencing file is processed as follows: removing adapters, aligning with a reference genome (eg, hg19 human reference genome), and quantifying gene expression.

[0021] In some embodiments, the method further comprises performing a normalization process on the gene expression value, for example, using a TPM, FPKM or RPKM conversion method to perform the normalization process; preferably, using a TPM conversion method to perform the normalization process to obtain a TPM gene expression value dataset:

[0022]

[0023] Here, n is the number of all genes.

[0024] In some embodiments, the protein abundance dataset is obtained by:

[0025] i) isolating and quantifying proteins from the biological sample;

[0026] ii) enzymatically hydrolyzing the protein to obtain a peptide sample;

[0027] iii) analyzing the peptide sample by liquid chromatography-mass spectrometry and collecting mass spectrometry data;

[0028] iv) performing qualitative analysis based on the mass spectrometry data (eg, importing the mass spectrometry data into MaxQuant, matching it with the SwissProt human database, and screening results with FDR < 1%) to screen the mass spectrometry data for successful qualitative analysis.

[0029] In some embodiments, prior to quantifying the proteins, high abundance proteins are removed (e.g., using Pierce TM Top 14 Abundant Protein Depletion Spin Columns Kit (Thermo Scientific) to remove high-abundance proteins).

[0030] In some embodiments, before the liquid chromatography-mass spectrometry analysis is performed, a preliminary separation step is also included, for example, using an Agilent 300Extend C18 column, selecting a 4.6×250 column to separate the sample.

[0031] In some embodiments, the liquid chromatography-mass spectrometry analysis comprises the following steps:

[0032] Construction of Spectral Library: The peptide samples are separated by liquid chromatography, and then mass spectrometry data is collected to construct a spectral library; preferably, the peptides are dissolved in liquid chromatography mobile phase A and separated using an EASY-nLC 1200 ultra-high performance liquid phase system, wherein the liquid phase parameters are: mobile phase A is an aqueous solution containing 0.1% formic acid and 2% acetonitrile; mobile phase B is an acetonitrile solution containing 0.1% formic acid; liquid phase gradient settings: 0-55min, 4% to 20% B; 55-67min, 20% to 32% B; 67-71min, 32% to 80% B; 71-75min, 80% B, and the flow rate is maintained at 500nL / min; after the peptides are separated by the ultra-high performance liquid phase system, they are injected into the Capillary ion source for ionization and then into the Orbitrap Exploris TM 480 mass spectrometer for data acquisition;

[0033] DIA data acquisition: The liquid phase parameters were kept the same as those used in library construction. The peptides were separated by ultra-high performance liquid phase system and then analyzed by Orbitrap Exploris TM The peptide precursor ion and its secondary fragments were detected and analyzed using TOF. The data acquisition mode used the data-independent parallel accumulation serial fragmentation (dia-PASEF) mode. The primary mass spectrometry scanning range was set to 390-810 m / z. After one primary mass spectrum was acquired, 10 PASEF mode acquisitions were performed. The secondary mass spectrometry scanning range was 100-1700 m / z, with every 25 m / z as a window.

[0034] In some embodiments, the mass spectrometry data of step iii) is imported into MaxQuant and matched with the SwissProt human database, and the results with FDR<1% are screened. In some embodiments, the following screening parameters are set: i. the cleavage enzyme is Trypsin / P; ii. a maximum of 2 missed cleavages are allowed; iii. for precursor ions, the masstolerance is 20 ppm and 5 ppm in the first and main searches, respectively; iv. for fragment ions, the mass tolerance is 0.02 Da; v. the fixed modification is carbamidomethylation on Cys; vi. the variable modification is acetylation of the N-terminus of the protein and oxidation of Met.

[0035] In some embodiments, the method further includes normalizing the successfully characterized mass spectrometry data, such as dividing the original intensity value of the successfully characterized mass spectrometry data by the median value of all protein abundance values ​​in each biological sample.

[0036] In some embodiments, the metabolite abundance dataset is obtained by the following method:

[0037] i) analyzing the biological sample by liquid chromatography-mass spectrometry and collecting mass spectrometry data;

[0038] ii) performing qualitative analysis based on the mass spectrum data, and screening the mass spectrum data for successful qualitative analysis.

[0039] In some embodiments, before performing the liquid chromatography-mass spectrometry analysis, the biological sample is also processed to remove proteins from the biological sample (plasma sample) (for example, using organic solvents such as acetonitrile, methanol, etc., or neutral salts such as ammonium sulfate, sodium sulfate, etc. to dehydrate and precipitate proteins).

[0040] In some embodiments, the liquid chromatography-mass spectrometry analysis is performed by coupling high performance liquid chromatography (UHPLC) (1290 Infinity LC from Agilent Technologies) with a quadrupole time-of-flight mass spectrometer (AB Sciex TripleTOF 6600). For hydrophilic interaction liquid chromatography (HILIC) separation, a 2.1 mm × 100 mm ACQUIYUPLC BEH 1.7 μm column (Waters, Ireland) is used to analyze the sample. For the first MS / MS acquisition, the instrument was set to acquire over an m / z range of 60-1000 Da, and the time for the product ion scan was set to 0.2 seconds / spectrum. The m / z range for the second MS / MS acquisition was 25-1000 Da, and the time was 0.05 seconds / spectrum. The product ion scan used information-dependent acquisition (IDA) and selected high sensitivity mode. In short, the dissociation potential (DP) was fixed at ±60 V, excluding isotopes within the 4 Da range. The collision energy (CE) was fixed at 35 V, ±15 eV. Ten candidate ions were monitored per cycle.

[0041] In some embodiments, the qualitative analysis in step ii) comprises the following steps:

[0042] a) using ProteoWizard MSConvert (version 3.0.6428) to convert the mass spectrometry data obtained in step i) into an MzXML file;

[0043] b) Peak screening was performed using XCMS software (version 3.14.1), with key parameters including: centWave m / z = 10 ppm, peakwidth = c(10, 60), prefilter = c(10, 100);

[0044] c) Grouping the screened peaks, key parameters include: bw = 5, mzwid = 0.025, minfrac = 0.5;

[0045] d) annotating isotopes and adducts using CAMERA software (version 3.6), retaining extracted ion features with more than 50% non-zero values ​​in all samples;

[0046] e) Identification of metabolite species was performed by comparing the m / z values ​​(<10 ppm) with the mass spectra in the database (from Sino-Science New Life Science Company);

[0047] f) The raw intensity values ​​will be divided by the median value of all metabolite abundance values ​​in each sample to obtain the metabolite abundance dataset.

[0048] In some embodiments, before step (3), the following steps are also included: performing correlation analysis between the four candidate data sets obtained in steps (1) and (2) and age to screen for age-related features.

[0049] In some embodiments, the four candidate data sets obtained in steps (1) and (2) are subjected to Pearson correlation analysis with age, with BMI used as a covariate, the P value is corrected by the BH method, and the results with P<0.05 are selected as age-related features.

[0050] In some embodiments, the results are standardized (eg, by Z-Score transformation), and K-Means clustering is performed using the ComplexHeatmap R package to screen features that stably change with age as the age-related features.

[0051] In some embodiments, the age-related characteristic is selected from:

[0052] Clinical phenotypic measurement data, including body composition test-right upper extremity (ECW / TBW), alkaline phosphatase, lactate dehydrogenase, systolic blood pressure, globulin, glycated hemoglobin, cholesterol, anti-Mullerian hormone, lung function-VC MAX, lung function-FVC, and pulse;

[0053] Gene expression data, including KRT7, LGALS1, TIMP1, U91328.20, CD70, SPSB1, SLA2, LMNA, CTB-37A13.1, AC099552.2, LYPD2, NR4A1, CEACAM3, TESC, YPEL4, UOX, BPIFB4, MTHFD2P1, PAX9, IGFBP3, AC006126.4, SLC1A3, PTK7, DLL1, ZNF521, ARPP21, HHIP-AS1 , SERPINF1, PHEX, CUX2, LINC00865, AR, PLEKHA7, TRBV6-4, REG4, RP11-395N3.1, RP11-318K12.1, HPGD, RP11-930P14. 2. FAM171A2, SPTSSB, RP1-266L20.2, HDAC9, CCDC110, CYGB, KPNA7, SLC2A7, TANC1, AC128709.3, GNAI1, ST13P4 and SRPX;

[0054] Protein abundance data, including NEO1, LUM, COMP, THBS4, CRTAC1, LRP1, PCOLCE, MRC1, SPTB, IGDCC4, CILP2, and VTN; and,

[0055] Metabolite abundance data, including 1-octadecanoyl-2-octadecenoyl-sn-glycero-3-phosphocholine, 1,2-arachidonyl-sn-glycero-3-phosphocholine, prolineamide, 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, urea, Pc 40:7, 1-palmitoyl-2-stearoyl-sn-glycero-3-phosphocholine, pimelic acid-1-(2-cyclopentylamino)hydrazide, 2,4,6-tri-tert-butylaniline, phosphocholine, 4-pregnen-17α,20β-diol-3-one-20-sulfate, trans-dehydroandrostenone, and ramipril.

[0056] In some embodiments, in step (3), before merging the four candidate data sets, a standardization step is also included.

[0057] In some embodiments, the normalization formula is:

[0058]

[0059] The Matrix is ​​all the values ​​of the four candidate data sets, and missing values ​​will not be included in the calculation of the maximum / minimum values.

[0060] In some embodiments, the dataset is split into a training set and a test set, for example, in a 1:1 ratio.

[0061] In some embodiments, the age-related features (i.e., the age-related features obtained by screening the correlation analysis between the 4 candidate data sets mentioned above and age) are annotated with biological functions, and classified based on the annotation results to obtain three sub-candidate data sets of hormones, lipids, and immunity. In some embodiments, the above-mentioned age-related features are annotated with pathways using the MetaboAnalyst tool, divided into three pathways of hormones, lipids, and immunity, and the successfully annotated age-related features are classified according to the pathways to obtain the three sub-candidate data sets.

[0062] In some embodiments, the three candidate sub-data sets are subjected to the aforementioned standardization processing. For example, the standardization processing is performed according to the following formula:

[0063]

[0064] The Matrix is ​​all the values ​​of the three candidate sub-datasets, and missing values ​​will not be included in the calculation of the maximum / minimum values.

[0065] In some embodiments, the dataset is split into a training set and a test set, for example, in a 1:1 ratio.

[0066] In some embodiments, the age-related features in the hormone sub-dataset are selected from:

[0067] clinical phenotypic measurements, including anti-Mullerian hormone, aldosterone, prolactin, luteinizing hormone, and follicle-stimulating hormone;

[0068] Gene expression data, including ATP1A2, GNAI1, KCNMB4, PLCB4, CYP2J2, CYP24A1, LRP5, HTR7, KCNK10, and SCT;

[0069] Protein abundance data, including IGF1; and,

[0070] Metabolite abundance data, including 5α-pregnan-3α,17-diol-20-one 3-sulfate, estrone sulfate, 7α-hydroxy-3-oxo-4-cholesterol acid, and β-estradiol 17-valerate.

[0071] In some embodiments, the age-related features in the immune subset dataset are selected from:

[0072] Clinical phenotypic measurements, including globulin, albumin, and cytomegalovirus IgG;

[0073] Gene expression data, including VCAM1, ARG2, F2RL3, ITGA2B, ITGB5, KRT31, CCL24, CXCR1, PF4, CD70, and TNFSF9;

[0074] Protein abundance data, including CLU, PROS1, KNG1, C8G, C2, FGB, TFPI, NOTCH2, SPP1, THBS4, CTSS, and MRC1; and,

[0075] Metabolite abundance data including L-kynurenine and arachidonic acid (without peroxides).

[0076] In some embodiments, the age-related features in the lipid subset dataset are selected from:

[0077] Clinical phenotypic measurements, including cholesterol, low-density lipoprotein, and apolipoprotein A;

[0078] Gene expression data, including UGCG, GNAI1, PLCB4, and PTEN;

[0079] Protein abundance data, including ANGPTL3, APOA4, APOB, APOC3, LCAT, LDLR, and PCSK9; and,

[0080] Metabolite abundance data, including 1-(1Z-octadecenyl)-sn-glycero-3-phosphocholine, 1-docosahexaenoyl-2-hydroxy-sn-glycero-3-phosphocholine, 1-lignano-2-hydroxy-sn-glycero-3-phosphocholine, 1-palmitoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, 1-palmitoyl-2-stearoyl-sn-glycero-3-phosphocholine, 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, 1-stearoyl-sn-glycero-3-phosphocholine, 1,2-arachidonyl-sn-glycero-3-phosphocholine, 1,2-dioleoyl-sn-glycero-3-phosphate, 1,2-dioleoyl-sn-glycerol, Pc 38:4, Pc 40:7, phosphorylcholine, N-lauroyl-d-erythrylphosphatidylcholine, and Pro-Trp.

[0081] In some embodiments, the subject is an Asian (eg, East Asian or Southeast Asian) person, such as a Chinese. In some embodiments, the subject is ≥100 people, for example, ≥500 people, ≥1000 people, or ≥10000 people.

[0082] In some embodiments, the subject is over 18 years old, preferably 18-70 years old. In some embodiments, the subject age distribution is balanced. In some embodiments, the difference in the number of subjects in each age group (for example, divided into 18-44 years old age group, 45-59 years old age group and 60 years old age group) is less than 50%, less than 40%, less than 30%, less than 20%, less than 10%, less than 5% or less than 2%.

[0083] In a second aspect, the present invention provides a method for training an age estimation model, comprising:

[0084] (1) establishing a data set according to any method described in the first aspect;

[0085] (2) calculating the estimated age of the subject corresponding to the data using the age estimation model according to the data set;

[0086] (3) Training the age estimation model based on the difference between the estimation result and the age label of the data.

[0087] In some embodiments, the training is performed using the glmnet package, version 4.1.4. In some embodiments, the alpha value in the glmnet package is selected from 0.1-0.9, for example, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8 or 0.9.

[0088] In some embodiments, the estimated age y of the subject can be calculated by the following formula:

[0089] A x represents the age-related features in the dataset, a x Represents the corresponding A x The coefficient of , b represents a constant (i.e., the intercept mentioned in this article);

[0090] Through the training, each age-related feature A can be determined x The coefficient a x and the constant b.

[0091] In some embodiments, the age-related characteristic is selected from:

[0092] Clinical phenotypic measurement data, including body composition test-right upper extremity (ECW / TBW), alkaline phosphatase, lactate dehydrogenase, systolic blood pressure, globulin, glycated hemoglobin, cholesterol, anti-Mullerian hormone, lung function-VC MAX, lung function-FVC, and pulse;

[0093] Gene expression data, including KRT7, LGALS1, TIMP1, U91328.20, CD70, SPSB1, SLA2, LMNA, CTB-37A13.1, AC099552.2, LYPD2, NR4A1, CEACAM3, TESC, YPEL4, UOX, BPIFB4, MTHFD2P1, PAX9, IGFBP3, AC006126.4, SLC1A3, PTK7, DLL1, ZNF521, ARPP21, HHIP-AS1 , SERPINF1, PHEX, CUX2, LINC00865, AR, PLEKHA7, TRBV6-4, REG4, RP11-395N3.1, RP11-318K12.1, HPGD, RP11-930P14. 2. FAM171A2, SPTSSB, RP1-266L20.2, HDAC9, CCDC110, CYGB, KPNA7, SLC2A7, TANC1, AC128709.3, GNAI1, ST13P4 and SRPX;

[0094] Protein abundance data, including NEO1, LUM, COMP, THBS4, CRTAC1, LRP1, PCOLCE, MRC1, SPTB, IGDCC4, CILP2, and VTN; and,

[0095] Metabolite abundance data, including 1-octadecanoyl-2-octadecenoyl-sn-glycero-3-phosphocholine, 1,2-arachidonyl-sn-glycero-3-phosphocholine, prolineamide, 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, urea, Pc 40:7, 1-palmitoyl-2-stearoyl-sn-glycero-3-phosphocholine, pimelic acid-1-(2-cyclopentylamino)hydrazide, 2,4,6-tri-tert-butylaniline, phosphocholine, 4-pregnen-17α,20β-diol-3-one-20-sulfate, trans-dehydroandrostenone, and ramipril.

[0096] In some embodiments, the coefficient a and the constant b of each age-related feature in the age estimation model are:

[0097]

[0098]

[0099]

[0100]

[0101]

[0102]

[0103] In some embodiments, the dataset is the hormone sub-dataset described above, and the age-related feature is selected from:

[0104] clinical phenotypic measurements, including anti-Mullerian hormone, aldosterone, prolactin, luteinizing hormone, and follicle-stimulating hormone;

[0105] Gene expression data, including ATP1A2, GNAI1, KCNMB4, PLCB4, CYP2J2, CYP24A1, LRP5, HTR7, KCNK10, and SCT;

[0106] Protein abundance data, including IGF1; and,

[0107] Metabolite abundance data, including 5α-pregnan-3α,17-diol-20-one 3-sulfate, estrone sulfate, 7α-hydroxy-3-oxo-4-cholesterol acid, and β-estradiol 17-valerate.

[0108] In some embodiments, the coefficient a and the constant b of each age-related feature in the age estimation model are:

[0109]

[0110]

[0111] In some embodiments, the dataset is the immune subset dataset described above, and the age-related feature is selected from:

[0112] Clinical phenotypic measurements, including globulin, albumin, and cytomegalovirus IgG;

[0113] Gene expression data, including VCAM1, ARG2, F2RL3, ITGA2B, ITGB5, KRT31, CCL24, CXCR1, PF4, CD70, and TNFSF9;

[0114] Protein abundance data, including CLU, PROS1, KNG1, C8G, C2, FGB, TFPI, NOTCH2, SPP1, THBS4, CTSS, and MRC1; and,

[0115] Metabolite abundance data including L-kynurenine and arachidonic acid (without peroxides).

[0116] In some embodiments, the coefficient a and the constant b of each age-related feature in the age estimation model are:

[0117]

[0118]

[0119] In some embodiments, the dataset is the lipid subset dataset described above, and the age-related feature is selected from:

[0120] Clinical phenotypic measurements, including cholesterol, low-density lipoprotein, and apolipoprotein A;

[0121] Gene expression data, including UGCG, GNAI1, PLCB4, and PTEN;

[0122] Protein abundance data, including ANGPTL3, APOA4, APOB, APOC3, LCAT, LDLR, and PCSK9; and,

[0123] Metabolite abundance data, including 1-(1Z-octadecenyl)-sn-glycero-3-phosphocholine, 1-docosahexaenoyl-2-hydroxy-sn-glycero-3-phosphocholine, 1-lignano-2-hydroxy-sn-glycero-3-phosphocholine, 1-palmitoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, 1-palmitoyl-2-stearoyl-sn-glycero-3-phosphocholine, 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, 1-stearoyl-sn-glycero-3-phosphocholine, 1,2-arachidonyl-sn-glycero-3-phosphocholine, 1,2-dioleoyl-sn-glycero-3-phosphate, 1,2-dioleoyl-sn-glycerol, Pc 38:4, Pc 40:7, phosphorylcholine, N-lauroyl-d-erythrylphosphatidylcholine, and Pro-Trp.

[0124] In some embodiments, the coefficient a and the constant b of each age-related feature in the age estimation model are:

[0125]

[0126]

[0127] In three aspects, the present invention provides an age estimation model, which is trained by the age estimation model training method described above.

[0128] In four aspects, the present invention provides an age estimation device, comprising:

[0129] An input unit, used for inputting age-related characteristics of the subject, wherein the age-related characteristics are as defined in any of the above items;

[0130] an estimating unit, configured to calculate the estimated age of the subject according to the input age-related features using the age estimation model;

[0131] and an optional output unit for outputting the estimated age of the subject.

[0132] In a fifth aspect, the present invention provides an age estimation method, comprising:

[0133] Providing a subject with age-related characteristics, wherein the age-related characteristics are as defined in any of the above items;

[0134] The age-related features are input into the age estimation model described in any one of the third aspect or the age estimation device described in any one of the fourth aspect to estimate the age of the subject.

[0135] The age-related characteristics include the following four characteristics: clinical phenotype measurement data, gene expression data, protein abundance data and metabolite abundance data. For each clinical phenotype measurement data, the units can refer to those listed in Example Table 2. For the gene expression data, protein abundance data and metabolite abundance data, refer to the method for establishing the gene expression data set, protein abundance data set and metabolite abundance data set described in the first aspect above. Before performing the age estimation, it also includes the step of standardizing the above four aspects of data, such as performing TPM conversion on the gene expression data set, median standardization on the protein abundance data set and the metabolite abundance data set, and processing the four types of data through the following formula to unify the value ranges of different types of data:

[0136]

[0137] The Matrix is ​​all the values ​​of the four candidate data sets, and missing values ​​will not be included in the calculation of the maximum / minimum values.

[0138] In a sixth aspect, the present invention provides an electronic device, comprising:

[0139] Memory; and

[0140] A processor coupled to the memory, the processor being configured to execute, based on instructions stored in the memory, the training method for the age estimation model described in any one of the second aspects, or the age estimation method described in any one of the fifth aspects.

[0141] In a seventh aspect, the present invention provides a non-volatile computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the training method for the age estimation model described in any one of the second aspect, or the age estimation method described in any one of the fifth aspect.

[0142] In an eighth aspect, the present invention provides a method for establishing a data set according to any one of the first aspect, a training method for an age estimation model according to any one of the second aspect, an age estimation model according to any one of the third aspect, an age estimation device according to any one of the fourth aspect, an age estimation method according to any one of the fifth aspect, an electronic device according to any one of the sixth aspect, or a non-volatile computer-readable storage medium according to any one of the seventh aspect, in the preparation of a kit, wherein the kit is used for one or more of the following:

[0143] 1) Estimate the age of the subjects;

[0144] 2) Evaluate or assist in evaluating the aging rate or degree of subjects;

[0145] 3) Evaluate or assist in evaluating the effectiveness of aging interventions;

[0146] 4) Evaluate or assist in evaluating the risk of aging-related diseases (such as cardiovascular diseases, neurodegenerative diseases, diabetes or chronic obstructive pulmonary disease);

[0147] 5) Evaluate or assist in evaluating the therapeutic effects of aging-related diseases;

[0148] 6) Evaluate or assist in evaluating the rate or extent of hormonal aging;

[0149] 7) Evaluate or assist in evaluating the rate or extent of lipid aging;

[0150] 8) Evaluate or assist in evaluating the rate or extent of immune senescence.

[0151] Definition of terms

[0152] In the present invention, unless otherwise specified, the scientific and technical terms used herein have the meanings commonly understood by those skilled in the art. In addition, the molecular genetics, nucleic acid chemistry, immunology, and bioinformatics laboratory operation steps used herein are conventional steps widely used in the corresponding fields. At the same time, in order to better understand the present invention, the definitions and explanations of the relevant terms are provided below.

[0153] As used herein, the term "aging" has the meaning commonly understood by those skilled in the art, and is a common biological phenomenon in the growth and development of the body. Aging can be considered as a process of body function decline accompanied by aging, which is usually accompanied by a continuous decline in physical health and an increase in the risk of aging-related diseases, functional disorders and death.

[0154] As used herein, the term "aging-related diseases" refers to aging-related diseases whose risk of occurrence increases significantly with aging of the body. In certain embodiments, the aging-related diseases include, but are not limited to, neurodegenerative diseases (such as Alzheimer's disease), diabetes, heart disease, cardiovascular and cerebrovascular diseases, hypertension, hyperlipidemia, osteoporosis, osteoarthritis, chronic kidney disease, cancer, and vision loss. In certain embodiments, the aging-related diseases are cardiovascular diseases, neurodegenerative diseases, diabetes, or chronic obstructive pulmonary disease.

[0155] The term "aging clock" as used in this article refers to a predictive model that can estimate the age of a subject based on multidimensional data using a machine learning model.

[0156] The term "hormonal aging" as used herein refers to the progressive degenerative changes in multidimensional indicators related to hormones in function with aging.

[0157] The term "lipid aging" as used herein refers to the progressive degenerative changes in multidimensional indicators related to lipids in function with aging.

[0158] The term "immunosenescence" as used herein refers to the progressive degenerative changes in multidimensional indicators related to immunity with aging.

[0159] The age-related diseases described in the present invention are selected from cardiovascular diseases (such as hypertension, coronary heart disease, heart failure or stroke), neurodegenerative diseases (such as Alzheimer's disease or Parkinson's disease), diabetes or chronic obstructive pulmonary disease.

[0160] As used herein, the term "gene" refers to a fragment of DNA associated with the production of a polypeptide chain. It may include regions before and after the coding region (leader and non-transcribed tail), as well as intervening sequences (introns) between individual coding segments (exons). The term gene is used interchangeably with nucleic acids, cDNA, mRNA, small non-coding RNA, microRNA (miRNA), Piwi-interacting RNA, and short hairpin RNA (shRNA) encoded by a gene or locus. The terms "nucleic acid" and "polynucleotide" refer to deoxyribonucleic acid (DNA) or ribonucleic acid "RNA" and polymers thereof in single-stranded or double-stranded form. Unless otherwise specified, the term includes nucleic acids containing known analogs of natural nucleotides, which have similar binding properties to the reference nucleic acid and are metabolized in a manner similar to naturally occurring nucleotides. Unless otherwise specified, a specific nucleic acid sequence also implicitly includes conservatively modified variants thereof (such as degenerate codon substitutions), alleles, orthologs, SNPs, and complementary sequences as well as sequences explicitly indicated. Specifically, the replacement of degenerate codons can be achieved by generating a sequence in which the third position of one or more selected (or all) codons is replaced by mixed bases and / or deoxyinosine residues (Batzer et al., Nucleic Acid Res. 19:5081 (1991); Ohtsuka et al., J. Biol. Chem. 260:2605-2608 (1985); and Rossolini et al., Mol. Cell. Probes 8:91-98 (1994)). "Genome" in the present invention refers to the sum of all DNA of an organism.

[0161] As used herein, the term "protein" or "protein" is a biological molecule composed of amino acids, which is essential for life and has many roles such as structure, metabolism, transport, immunity, signaling and regulation. The term "proteome" is a general term that refers to all proteins that can be expressed by an organism, a tissue or a cell.

[0162] As used herein, the term "metabolite" refers to a substance produced or consumed by a metabolic process. Small molecule compounds generated or transformed by enzymes during metabolism are also called metabolites. "Metabolome" refers to the collection of all metabolites in an organism, a tissue or a cell.

[0163] As used herein, compound "Pc 38:4", i.e., phosphatidylcholine 38:4 (lecithin 38:4, ChEBI: 64526), ​​has the meaning generally understood by those skilled in the art. Generally speaking, it refers to a 1,2-diacyl-sn-glycero-3-phosphocholine, wherein the acyl groups at R1 and R2 contain a total of 38 carbons and 4 double bonds. In certain embodiments, the structural formula of the compound Pc 38:4 is as follows:

[0164]

[0165] Similarly, Pc 40:7, phosphatidylcholine 40:7 (lecithin 40:7, ChEBI: 64521), has a total of 40 carbons and 7 double bonds in the acyl groups at R1 and R2.

[0166] Each gene, protein and metabolite mentioned in this article is known in the art and has the meaning generally understood in the art. Specifically, some gene and protein information involved in this article is as follows:

[0167]

[0168]

[0169]

[0170]

[0171] Advantageous Effects of the Invention

[0172] The present invention provides an age estimation method based on multidimensional data. The method of the present invention has at least one of the following effects:

[0173] 1) Improved data quality: Strict cohort establishment standards have eliminated the influence of factors other than age on experimental and analysis results to the greatest extent;

[0174] 2) Measuring aging in Asian populations: The technical solution of the present invention is based on data from the Chinese population, and the resulting aging measurement system is suitable for Asian populations, especially East Asian or Southeast Asian populations, such as the Chinese population;

[0175] 3) Increased information dimension: The technical solution of the present invention collects multi-dimensional omics information including clinical test information of the population, blood cell transcriptome, plasma proteome, and plasma metabolome, expanding the information dimension of the aging measurement system from the source and providing more extensive aging-related information;

[0176] 4) Enhance the readability of the aging measurement model: In addition to analyzing and modeling a single-omics level to obtain a single-level indicator for evaluating aging, the present invention also integrates multi-level omics information, establishes a composite age prediction model for the first time, and further establishes a composite age model for hormone, immune, and lipid aging based on the prior knowledge graph, which enhances the readability of the aging clock and is conducive to specifically judging the heterogeneity of aging in the subjects in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0177] Figure 1 Shows the basic technical route for establishing an aging evaluation system.

[0178] Figure 2 Shows the technical route for establishing a multi-dimensional composite aging clock.

[0179] Figure 3 The results show that K-means clustering can be used to screen the multi-level data for stable changes with age, among which Group 1 and Group 2 are stable change groups.

[0180] Figure 4 show Figure 3 The trend graph of the characteristics of different groups changing with age also clearly shows that the characteristics in Group 1 and Group 2 change steadily with age.

[0181] Figure 5 Functional annotation results showing age-related molecular signatures. The pathways circled by different lines are three main types.

[0182] Figure 6 The results of age prediction using the multidimensional composite aging clock are shown. R is the Pearson correlation coefficient and MAE is the mean absolute error. DETAILED DESCRIPTION

[0183] The embodiments of the present invention will be described in detail below in conjunction with the examples. It is believed that even without these descriptions, those skilled in the art can make the most of the present invention based on the foregoing content. Therefore, the following examples are merely illustrative and do not limit the present invention in any way. If specific conditions are not indicated in the examples, they are carried out according to normal conditions or the conditions recommended by the manufacturer. If the manufacturer is not indicated for the reagents or instruments used, they are all conventional products that can be obtained commercially.

[0184] The technical route of the present invention mainly involves seven steps: cohort establishment, data collection, data processing, age-related analysis, aging clock establishment, and aging clock evaluation and application. The simple process is as follows: Figure 2 shown.

[0185] 1. Volunteer recruitment and sample collection

[0186] The "Preliminary Screening Questionnaire" was published through the WeChat Mini Program to collect basic information of potential volunteers who are interested in joining this study based on the inclusion criteria, and to screen volunteers who meet the inclusion criteria; the inclusion criteria are as follows:

[0187] 1.1 Enrollment criteria

[0188] a) Aged 18 or above, with no major bad habits;

[0189] b) The applicant is in good health and has not undergone any major surgery within the past six months. Routine conditions related to the normal aging process are permitted, including:

[0190] i. Hypertension, well controlled (taking no more than 3 medications);

[0191] ii. Osteoporosis, osteopenia and / or osteoarthritis;

[0192] iii. Benign prostatic hypertrophy;

[0193] iv. Cataract, glaucoma, macular degeneration;

[0194] v.Dyslipidemia;

[0195] vi. Hypothyroidism;

[0196] vii. Prediabetes or abnormal fasting blood sugar (fasting blood sugar 100-126 mg / dL).

[0197] c) The subjects can communicate well with the researchers and cooperate with all tasks during the study;

[0198] i. Subjects must voluntarily sign a written informed consent.

[0199] 1.2 Exclusion criteria

[0200] a) Age does not meet the inclusion criteria;

[0201] b) Participants are currently undergoing other clinical trials or are receiving any research drugs or devices for treatment within 30 days prior to enrollment;

[0202] c) The investigator reports a history or current diagnosis of significant chronic medical conditions, including:

[0203] i. Any cancer (except polycythemia vera, basal cell or squamous cell skin cancer);

[0204] ii. Coronary artery disease / myocardial infarction / clinically significant congestive heart failure;

[0205] iii. Stroke / transient ischemic attack;

[0206] iv. Deep vein thrombosis / pulmonary embolism;

[0207] v. Serum creatinine>1.5mg / dl (male);

[0208] vi. poorly controlled hypertension (significant hypertension (systolic blood pressure > 160 mmHg, or diastolic blood pressure > 100 mmHg) despite treatment);

[0209] vii. History of active liver disease or metabolic acidosis;

[0210] viii. Chronic kidney disease / hemodialysis treatment, history of severe renal damage and / or eGFR≤45ml / min / 1.73m2;

[0211] ix. Severe autoimmune / inflammatory diseases, such as rheumatoid arthritis, lupus, Crohn's disease, etc.;

[0212] x. Neurological diseases such as dementia, such as Alzheimer's / Parkinson's, etc.;

[0213] xi. Diabetes (hemoglobin A1C>6.5% or fasting blood glucose>126 mg / dL or taking diabetes medication or insulin treatment);

[0214] xii. Recent (within 3 months) cardiovascular disease events (myocardial infarction, coronary artery intervention, coronary artery bypass grafting);

[0215] xiii. Infectious diseases such as HIV, hepatitis, tuberculosis, etc.

[0216] d) Currently taking the following medications regularly:

[0217] i. Chemotherapy drugs (such as tamoxifen, doxorubicin, mitoxantrone, bleomycin);

[0218] ii. Antiplatelet drugs (e.g. clopidogrel / Plavix, dipyridamole / Angeleno, ticlopidine / Ticlopidine, excluding aspirin);

[0219] iii. Cholinesterase inhibitors for Alzheimer's disease (donepezil / alicept).

[0220] e) Continuous alcohol or drug abuse;

[0221] f) Unable to provide informed consent;

[0222] g) The researcher believes that the physical factors of the participants may have an adverse effect on the research process or results.

[0223] 1.3 Termination criteria

[0224] a) Failure to follow the diet, work and rest schedule and physical examination requirements in accordance with the research content;

[0225] b) Experiencing abnormal physical conditions such as colds or fever during the study;

[0226] c) Accidental injury or physical discomfort during the research process;

[0227] d) Comply with the subject’s subjective wishes to terminate the program.

[0228] All volunteers who meet the inclusion criteria and are confirmed to be included in the group signed an informed consent form and joined the volunteer WeChat group for management and information communication. Within one month before the start of the project, volunteers need to maintain a normal schedule, eat a proper diet, avoid strenuous exercise, and refrain from excessive smoking or drinking. The day before the project, volunteers received a unified standardized diet (menu see Table 1) at the implementation hospital and were no longer allowed to take other foods.

[0229] On the day of sample collection and physical examination, volunteers are required to wear clothes and shoes that are convenient for movement, are not allowed to wear makeup, and must bring identification documents (such as ID cards, etc.) to the hospital for physical examination and biological sample collection.

[0230] Table 1 Unified standardized diet menu

[0231]

[0232] 2. Data Collection

[0233] 2.1 Clinical phenotype detection

[0234] The physical examination included general examination, internal medicine examination, surgical examination, B-ultrasound examination, body composition analysis, blood routine test, glycosylated hemoglobin, six items of sex hormones, urine routine test, stool routine test, lung function test, bone density test, electrocardiogram and biochemical analysis. The specific indicators are shown in Table 2.

[0235] Table 2 Detailed indicators of physical examination

[0236]

[0237]

[0238]

[0239]

[0240]

[0241]

[0242]

[0243]

[0244]

[0245]

[0246]

[0247]

[0248]

[0249]

[0250] In addition, the clinical measurement indicators also include 5 mobility tests, as shown in Table 3.

[0251] Table 3 Action ability test detection method

[0252]

[0253]

[0254] The measurement data in Tables 2 and 3 were collected to obtain a clinical measurement value matrix.

[0255] 2.2 Biological sample processing

[0256] Blood samples were obtained by venous blood sampling after the volunteers fasted overnight. The collected blood was placed in two anticoagulant tubes covered with EDTA (for the separation of plasma and peripheral mononuclear cells, respectively), and immediately transported to the laboratory's 4°C refrigerator for subsequent processing. One tube of blood sample (approximately 5 ml) was centrifuged (400g, 15min, 4°C) to separate the plasma sample, which was stored in a cryopreservation tube and placed in a -80°C refrigerator for long-term storage. The steps for separating peripheral mononuclear cells are as follows:

[0257] a) Dilute EDTA anticoagulated peripheral blood with PBS in a 1:1 ratio in a 15 mL centrifuge tube and invert 4 to 5 times to mix thoroughly;

[0258] b) Take another 15mL centrifuge tube, add 1 / 3 volume of lymphocyte separation solution, and carefully add the diluted peripheral blood vertically and slowly along the tube wall onto the lymphocyte separation solution, keeping the liquid surface clear; (It is recommended to use a 200μL pipette to add blood samples at the beginning, and use a 1mL pipette after a certain thickness is formed);

[0259] c) Horizontal centrifugation, 400g, 30min, 4°C, the centrifuge lifting speed needs to be adjusted to the lowest, the lifting speed is 1, the lowering speed is 0, and the total centrifugation time is about 1h;

[0260] d) After centrifugation, it is divided into three layers: the upper layer is plasma and PBS, the middle layer is lymphocyte separation fluid, and the lower layer is mainly red blood cells and granulocytes. There is a narrow white cloudy band at the interface between the upper and middle layers, which is PBMC. First, use a 1mL pipette to discard most of the upper yellow liquid, and the remaining 2mL to the PBMC should be carefully transferred to a 15mL centrifuge tube with a 1mL or 200μL pipette;

[0261] e) Add 10 mL of PBS, mix gently, centrifuge (1500 rpm, 15 min), and remove the supernatant.

[0262] f) Resuspend in 6 mL PBS, take 1 mL aliquots into individual cryovials, centrifuge at 400 g, 15 min, 4°C, and discard the supernatant;

[0263] g) Resuspend in 1 mL Trizol and mix by gently pipetting;

[0264] h) Store in a -80℃ freezer.

[0265] 2.3 Peripheral monocyte transcriptome sequencing

[0266] a) For PBMC samples treated with Trizol, thaw on ice and add 0.2 ml chloroform per ml sample, shake and place at room temperature for 3-5 minutes, centrifuge at 4°C for 15 minutes (10000g), transfer the colorless aqueous phase to a clean centrifuge tube, add 0.5 ml isopropanol and place at room temperature for 10 minutes. Centrifuge at 4°C for 10 minutes (10000g), discard the supernatant and add 1 ml pre-cooled 75% alcohol, resuspend and centrifuge at 4°C (8000g) for 5 minutes, discard the supernatant, dry at room temperature, dissolve the sample in 20 μl DEPC water and store at -80°C for later use.

[0267] b) After dissolving the RNA sample in the above step, dilute according to the concentration and take 1-3 μg RNA sample for use in VAHTSUniversal V6 RNA-seq Library Prep Kit (Illumina, NR604-01 / 02) to construct RNA-seq sequencing library;

[0268] c) The obtained RNA-seq library was sequenced with double ends using the Illumina Hisses X-Ten sequencing platform to obtain a sequencing file with a read length of 150 bp.

[0269] 2.4 Plasma proteome quantification

[0270] a) Protein extraction: The sample was taken out from -80℃ and centrifuged at 12000g for 10 minutes at 4℃ to remove cell debris. The supernatant was transferred to a new centrifuge tube and centrifuged using Pierce TM Top 14Abundant Protein Depletion Spin Columns Kit (Thermo Scientific)Instructions Remove high-abundance proteins. Protein concentration was determined using the BCA kit.

[0271] b) Trypsin hydrolysis: Take an equal amount of each sample protein for hydrolysis, add an appropriate amount of standard protein, adjust the volume to the same with lysis buffer, and then add dithiothreitol (DTT) to a final concentration of 5mM, and reduce at 56°C for 30min. Then add iodoacetamide (IAA) to a final concentration of 11mM, and incubate at room temperature in the dark for 15min. Transfer the alkylated sample to an ultrafiltration tube, centrifuge at 12000g for 20min at room temperature, replace with 8M urea 3 times, and then replace urea with replacement buffer 3 times, add trypsin at a ratio of 1:50 (protease: protein, m / m), and hydrolyze overnight. Centrifuge at 12000g for 10min at room temperature to recover the peptides, and then use ultrapure water to recover the peptides once, and combine the two peptide solutions.

[0272] c) HPLC fractionation: Agilent 300Extend C18 column was used to separate the samples. The column was 4.6×250 in size and the wavelength was 214 nm. The column oven temperature was 35°C. The column was equilibrated with 95% buffer A for 30 min. After the baseline was leveled, the graded gradient method was used. The peptide samples were loaded onto the HPLC and the method was started to separate the samples. The samples were collected at 1 min / tube. Tubes 11 to 46 were collected, a total of 36 tubes, which were combined into 12 components and dried under vacuum.

[0273] d) Liquid chromatography-mass spectrometry analysis:

[0274] Construction of spectral library: The peptides were dissolved in liquid chromatography mobile phase A and separated using the EASY-nLC 1200 ultra-high performance liquid phase system. Mobile phase A was an aqueous solution containing 0.1% formic acid and 2% acetonitrile; mobile phase B was an acetonitrile solution containing 0.1% formic acid. Liquid phase gradient settings: 0-55min, 4%~20%B; 55-67min, 20%~32%B; 67-71min, 32%~80%B; 71-75min, 80%B, and the flow rate was maintained at 500nL / min. After the peptides were separated by the ultra-high performance liquid phase system, they were injected into the Capillary ion source for ionization and then into the Orbitrap Exploris TM 480 mass spectrometer for data collection. DIA data collection: Liquid phase parameters are consistent with those used in library construction. After peptides are separated by ultra-high performance liquid phase system, OrbitrapExploris TM The ion source voltage was set to 1.6 kV, and the peptide parent ion and its secondary fragments were detected and analyzed using TOF. The data acquisition mode used the data-independent parallel accumulation serial fragmentation (dia-PASEF) mode, and the primary mass spectrometry scanning range was set to 390-810 m / z. After one primary mass spectrum was acquired, 10 PASEF mode acquisitions were performed, and the secondary mass spectrometry scan was in the range of 100-1700 m / z, with every 25 m / z as a window.

[0275] 2.5 Plasma metabolome analysis

[0276] Plasma samples were thawed at 4°C, and 100 μL of the sample was mixed with 400 μL of ice-cold methanol / acetonitrile (volume ratio 1:1) to remove proteins. The mixture was centrifuged at 14,000 g for 15 min at 4°C, and the supernatant was removed. For LC-MS analysis, the sample was reconstituted in 100 μL of acetonitrile / water (volume ratio 1:1) solvent. The analysis was performed using high-performance liquid chromatography (UHPLC) (1290 Infinity LC from Agilent Technologies) coupled to a quadrupole time-of-flight mass spectrometer (AB Sciex TripleTOF 6600). For hydrophilic interaction liquid chromatography (HILIC) separation, the samples were analyzed using a 2.1 mm × 100 mm ACQUIY UPLC BEH 1.7 μm column (Waters, Ireland). For the first MS / MS acquisition, the instrument was set to acquire over the m / z range of 60–1000 Da, and the product ion scan time was set to 0.2 s / spectrum. The second MS / MS acquisition was performed in the m / z range of 25-1000Da with a time of 0.05 sec / spectrum. Product ion scans were performed using information-dependent acquisition (IDA) with high sensitivity mode selected. Briefly, the dissociation potential (DP) was fixed at ±60V and isotopes within the 4Da range were excluded. The collision energy (CE) was fixed at 35V, ±15eV. Ten candidate ions were monitored per cycle.

[0277] 3. Data Preprocessing

[0278] 3.1 Peripheral monocyte transcriptome data

[0279] a) Use TrimGlalore software (version 0.4.5) to remove the adapter from the original fastq file of sequencing;

[0280] b) Use HISAT2 software to align the processed fastq file to the hg19 human reference genome to obtain a bam format alignment result file;

[0281] c) Use the samtools tool to convert the bam file into a sam file;

[0282] d) The alignment results with alignment scores > 20 were used to quantify gene expression using HTSeq software (version 0.11.0);

[0283] e) The obtained quantitative matrix is ​​processed according to the following formula to obtain the TPM values ​​of all genes in each sample and obtain the TPM gene expression matrix:

[0284]

[0285] Here, n is the number of all genes.

[0286] 3.2 Plasma proteome data

[0287] a) The raw data of mass spectrometry were first imported into MaxQuant software (version 1.6.15.0) and then searched based on the SwissProt human database. The key parameters included:

[0288] i. Lysing enzyme is Trypsin / P

[0289] ii. A maximum of 2 missing cleavages are allowed

[0290] iii. For the precursor ion, the mass tolerance is 20ppm and 5ppm in the first and main searches respectively.

[0291] iv. For fragment ions, the mass tolerance is 0.02Da

[0292] v. Fixed modification is carbamidomethylation on Cys

[0293] vi. Variable modifications include acetylation of the protein N-terminus and oxidation of Met

[0294] b) The screening criteria for search results was FDR < 1%;

[0295] c) The raw intensity values ​​will be divided by the median value of each sample to obtain a normalized protein abundance matrix.

[0296] 3.3 Plasma metabolome data

[0297] a) First, the mass spectrometry raw data was converted into an MzXML file using ProteoWizard MSConvert (version 3.0.6428);

[0298] b) Then, peak screening was performed using XCMS software (version 3.14.1), with key parameters including: centWave m / z = 10 ppm, peakwidth = c (10, 60), prefilter = c (10, 100);

[0299] c) Then the screened peaks are grouped, and the key parameters include: bw = 5, mzwid = 0.025, minfrac = 0.5;

[0300] d) Then, the isotopes and adducts were annotated using CAMERA software (version 3.6), and only the extracted ion features with non-zero values ​​exceeding 50% in all samples were retained;

[0301] e) Identification of metabolite species was performed by comparing the m / z values ​​(<10 ppm) with the mass spectra in the database (from Sino-Science New Life Science Company);

[0302] f) The raw intensity values ​​will be divided by the median value of each sample to obtain a normalized metabolite abundance matrix.

[0303] 4. Age-related feature screening

[0304] a) First, perform age correlation analysis on the clinical measurement value matrix obtained in the "2.1 Clinical phenotype detection" section and the TPM gene expression matrix, protein abundance matrix, and metabolite abundance matrix obtained in the "3. Data preprocessing" section, that is, perform Pearson correlation analysis on the value of each clinical measurement value / gene / protein / metabolite and the individual's age, and use BMI as a covariate;

[0305] b) Filter the results and select only those with a BH-corrected P value less than 0.05 as clinical measurements / genes / proteins / metabolites related to age;

[0306] c) Merge the matrices of age-related features obtained in the previous step together (features as rows, individuals as columns), and perform Z-Score transformation on the rows;

[0307] d) The transformed feature-individual matrix will be further clustered using the ComplexHeatmap R package using k-means to obtain six feature groups with different abundance patterns in the age trajectory ( Figure 3 );

[0308] e) Select the two feature groups (group 1 and group 2) that are most stable with age as the final age-related features ( Figure 3-4 ).

[0309] 5. Biological function annotation of age-related traits

[0310] a) First, the age-related genes / proteins / metabolites obtained in the "4. Age-related feature screening" section were annotated using the MetaboAnalyst tool (version 5.0);

[0311] b) Pathways with similar functions were manually merged into one major category, and finally the three major categories of hormones, lipids, and immunity were determined ( Figure 5 );

[0312] c) Classify age-related clinical indicators into hormone, lipid, and immune categories, and annotate clinical indicators including:

[0313] i. Lipids: cholesterol, low-density lipoprotein, apolipoprotein A, triglycerides

[0314] ii. Immunity: globulin, albumin, cytomegalovirus IgG

[0315] iii. Hormones: anti-Mullerian hormone, dehydroepiandrosterone sulfate, aldosterone, prolactin, luteinizing hormone, follicle-stimulating hormone

[0316] 6. Establishment of a multidimensional composite aging clock

[0317] a) First, the clinical measurement value matrix and the TPM gene expression matrix, protein abundance matrix, and metabolite abundance matrix obtained in the "3. Data Preprocessing" section are standardized to 0-1, that is, the following processing is performed:

[0318]

[0319] Matrix is ​​all the values ​​of the four matrices, and missing values ​​will not be included in the calculation of the maximum / minimum values.

[0320] b) Merge the four standardized matrices and retain only the age-related features screened in the "4. Age-related feature screening" section to obtain the screened full matrix;

[0321] c) Perform subsequent operations using the full matrix (corresponding to the total composite clock) and the three sub-matrices (corresponding to the hormone, lipid, and immune composite clocks) consisting of the features included in the three major categories obtained in the "5. Biological function annotation of age-related features" section;

[0322] d) Filter out samples with missing values ​​greater than or equal to 10;

[0323] e) Split the matrix into 1:1 training and test sets according to the random assignment in advance;

[0324] f) Age was used as the training label and the glmnet package (version 4.1.4) was used to construct the model. Nine alpha values ​​between 0.1 and 0.9 were used for modeling and tested in the test set. Finally, the alpha with the highest accuracy was selected as the parameter of the final model. The final alpha values ​​were 0.3 (total composite clock), 0.9 (hormone composite clock), 0.9 (lipid composite clock), and 0.7 (immune composite clock). The final model is shown in Tables 4-5.

[0325] Table 4. Final model of total composite clock

[0326]

[0327]

[0328]

[0329]

[0330]

[0331]

[0332] Table 5. Functional annotation of the final model of the sub-complex clock

[0333]

[0334]

[0335]

[0336]

[0337]

[0338]

[0339]

[0340] 7. Evaluation of the Aging Clock

[0341] The four models obtained in Part 6 were applied to the test set to obtain the predicted age of the individuals in the test set and compared with their corresponding actual ages. The results showed that the total composite clock had the highest accuracy, with a correlation coefficient of 0.93 with age (Pearson correlation, with BMI as a covariate), and an average absolute error of 4 years ( Figure 6 ). Three composite clocks with biological function annotations achieved high accuracy predictions, among which the hormone composite clock performed best, with a correlation coefficient of 0.87 and an average absolute error of 4.83 years; the immune composite clock was second, with a correlation coefficient of 0.72 and an average absolute error of 7.04 years; the lipid composite clock had a correlation coefficient of 0.67 with actual age, and an average absolute error of 7.63 years ( Figure 6 ). The accuracy of different sub-complex clocks varies, which may reflect the different extent to which different functions change with age, that is, hormonal aging changes most significantly with age, while immune and lipid aging changes more slowly than hormonal aging.

[0342] In summary, the four multidimensional composite clocks all have strong age prediction capabilities, and after biological function annotation, they can further reflect the degree of aging in different aspects of the subjects.

[0343] Although the specific embodiments of the present invention have been described in detail, it will be understood by those skilled in the art that various modifications and substitutions may be made to those details based on all the teachings disclosed, and these changes are within the scope of protection of the present invention. The full scope of the present invention is given by the attached claims and any equivalents thereof.

Claims

1. A method for establishing a data set, comprising: (1) performing clinical phenotype testing on multiple subjects to obtain a clinical phenotype measurement data set; (2) collecting biological samples from the multiple subjects, performing the following analysis on the biological samples, and obtaining a gene expression dataset, a protein abundance dataset, and a metabolite abundance dataset based on the analysis results: 1) Transcriptome sequencing to obtain the expression level of each gene in the biological sample; 2) Proteomic quantitative analysis to obtain the abundance of each protein in the biological sample; 3) Quantitative metabolomics analysis to obtain the abundance of each metabolite in the biological sample; (3) The four candidate datasets in (1) and (2) are combined, and age is used as a training label to construct the dataset for training an age estimation model.

2. The method of claim 1, wherein the clinical phenotype detection includes a physical examination (for example, one or more of a general examination, an internal medicine examination, a surgical examination, an ultrasound examination, a body composition analysis, a blood routine test, glycosylated hemoglobin, six items of sex hormones, a urine routine test, a stool routine test, a lung function test, a bone density test, an electrocardiogram test, and a biochemical analysis), and / or a mobility test (for example, one or more of a single-leg stand, a grip strength test, a 30-second sit-and-stand test, a Purdue pegboard test, and a stepping test).

3. The method of claim 1 or 2, wherein the biological sample is selected from the group consisting of blood (eg, whole blood, plasma or serum), urine, saliva, cell and tissue extracts.

4. The method according to any one of claims 1 to 3, wherein the gene expression dataset is obtained by the following method: i) obtaining immune cells from the biological sample; ii) performing transcriptome sequencing on the immune cells; iii) obtaining the expression values ​​of all genes in the immune cells based on the transcriptome sequencing results; Preferably, the method further comprises normalizing the gene expression value, for example, using TPM, FPKM or RPKM conversion method to perform the normalization.

5. The method according to any one of claims 1 to 4, wherein the protein abundance dataset is obtained by the following method: i) isolating and quantifying proteins from the biological sample; ii) enzymatically hydrolyzing the protein to obtain a peptide sample; iii) analyzing the peptide sample by liquid chromatography-mass spectrometry and collecting mass spectrometry data; iv) performing qualitative analysis based on the mass spectrum data, and screening the mass spectrum data for successful qualitative analysis; Preferably, the method further comprises standardizing the successfully characterized mass spectrometry data, for example, dividing the original intensity value of the successfully characterized mass spectrometry data by the median value of all protein abundance values ​​in each biological sample.

6. The method according to any one of claims 1 to 5, wherein the metabolite abundance dataset is obtained by the following method: i) analyzing the biological sample by liquid chromatography-mass spectrometry and collecting mass spectrometry data; ii) performing qualitative analysis based on the mass spectrum data, and screening the mass spectrum data for successful qualitative analysis; Preferably, the method further comprises standardizing the successfully characterized mass spectral data, for example, dividing the original intensity value of the successfully characterized mass spectral data by the median value of all metabolite abundance values ​​in each biological sample.

7. The method according to any one of claims 1 to 6, further comprising the following steps before step (3): Perform correlation analysis between the four candidate data sets obtained in steps (1) and (2) and age to screen for age-related features; Preferably, the four candidate data sets obtained in steps (1) and (2) are subjected to Pearson correlation analysis with age, with BMI used as a covariate, the P value is corrected by the BH method, and the result with P<0.05 is selected as the feature related to age; Further preferably, the results are standardized (for example, by Z-Score transformation), and K-Means clustering is performed using the ComplexHeatmap R package to screen features that stably change with age as the age-related features.

8. The method according to any one of claims 1 to 7, further comprising a step of standardization before merging the four candidate data sets in step (3); Preferably, the standardization processing formula is: The Matrix is ​​all the values ​​of the four candidate data sets, and missing values ​​will not be included in the calculation of the maximum / minimum values.

9. The method according to any one of claims 1 to 8, wherein the data set is split into a training set and a test set, for example, in a 1:1 ratio.

10. The method according to any one of claims 7 to 9, wherein the age-related features are annotated with biological functions, and classified based on the annotation results to obtain three sub-candidate data sets of hormone, lipid and immune; Preferably, the three sub-datasets are processed using the method described in claim 8 or 9 to obtain three sub-datasets: a hormone sub-dataset, a lipid sub-dataset and an immune sub-dataset.

11. The method according to any one of claims 1 to 10, wherein the subject is an Asian (eg, East Asian or Southeast Asian) person, such as a Chinese person; preferably, the number of the subjects is ≥100, for example, ≥500, ≥1000, or ≥10000.

12. A method for training an age estimation model, comprising: (1) establishing a data set according to the method described in any one of claims 1 to 11; (2) calculating the estimated age of the subject corresponding to the data using the age estimation model according to the data set; (3) Training the age estimation model based on the difference between the estimation result and the age label of the data.

13. The training method of claim 12, wherein the training is performed using a glmnet package.

14. The training method of claim 12 or 13, wherein the subject estimates his age A x represents the age-related features in the dataset, a x Represents the corresponding A x The coefficient of , b represents a constant; Through the training, each age-related feature A is determined x The coefficient a x and the constant b.

15. The training method according to any one of claims 12 to 14, wherein the age-related characteristic is selected from: Clinical phenotypic measurement data, including body composition test-right upper extremity (ECW / TBW), alkaline phosphatase, lactate dehydrogenase, systolic blood pressure, globulin, glycated hemoglobin, cholesterol, anti-Mullerian hormone, lung function-VC MAX, lung function-FVC, and pulse; Gene expression data, including KRT7, LGALS1, TIMP1, U91328.20, CD70, SPSB1, SLA2, LMNA, CTB-37A13.1, AC099552.2, LYPD2, NR4A1, CEACAM3, TESC, YPEL4, UOX, BPIFB4, MTHFD2P1, PAX9, IGFBP3, AC006126.4, SLC1A3, PTK7, DLL1, ZNF521, ARPP21, HHIP-AS1 , SERPINF1, PHEX, CUX2, LINC00865, AR, PLEKHA7, TRBV6-4, REG4, RP11-395N3.1, RP11-318K12.1, HPGD, RP11-930P14.

2. FAM171A2, SPTSSB, RP1-266L20.2, HDAC9, CCDC110, CYGB, KPNA7, SLC2A7, TANC1, AC128709.3, GNAI1, ST13P4 and SRPX; Protein abundance data, including NEO1, LUM, COMP, THBS4, CRTAC1, LRP1, PCOLCE, MRC1, SPTB, IGDCC4, CILP2, and VTN; and, Metabolite abundance data, including 1-octadecanoyl-2-octadecenoyl-sn-glycero-3-phosphocholine, 1,2-arachidonyl-sn-glycero-3-phosphocholine, prolineamide, 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, urea, Pc 40:7, 1-palmitoyl-2-stearoyl-sn-glycero-3-phosphocholine, pimelic acid-1-(2-cyclopentylamino)hydrazide, 2,4,6-tri-tert-butylaniline, phosphocholine, 4-pregnen-17α,20β-diol-3-one-20-sulfate, trans-dehydroandrostenone, and ramipril.

16. The training method according to claim 14 or 15, wherein the coefficient a and the constant b of each age-related feature in the age estimation model are:

17. The training method according to any one of claims 12 to 14, wherein the data set is a hormone sub-data set established according to the method of claim 10, and the age-related feature is selected from: clinical phenotypic measurements, including anti-Mullerian hormone, aldosterone, prolactin, luteinizing hormone, and follicle-stimulating hormone; Gene expression data, including ATP1A2, GNAI1, KCNMB4, PLCB4, CYP2J2, CYP24A1, LRP5, HTR7, KCNK10, and SCT; Protein abundance data, including IGF1; and, Metabolite abundance data, including 5α-pregnan-3α,17-diol-20-one 3-sulfate, estrone sulfate, 7α-hydroxy-3-oxo-4-cholesterol acid, and β-estradiol 17-valerate.

18. The training method of claim 17, wherein the coefficient a and the constant b of each age-related feature in the age estimation model are:

19. The training method according to any one of claims 12 to 14, wherein the data set is an immune sub-data set established according to the method of claim 10, and the age-related feature is selected from: Clinical phenotypic measurements, including globulin, albumin, and cytomegalovirus IgG; Gene expression data, including VCAM1, ARG2, F2RL3, ITGA2B, ITGB5, KRT31, CCL24, CXCR1, PF4, CD70, and TNFSF9; Protein abundance data, including CLU, PROS1, KNG1, C8G, C2, FGB, TFPI, NOTCH2, SPP1, THBS4, CTSS, and MRC1; and, Metabolite abundance data including L-kynurenine and arachidonic acid (without peroxides).

20. The training method of claim 19, wherein the coefficient a and the constant b of each age-related feature in the age estimation model are:

21. The training method according to any one of claims 12 to 14, wherein the data set is a lipid sub-data set established according to the method of claim 10, and the age-related feature is selected from: Clinical phenotypic measurements, including cholesterol, low-density lipoprotein, and apolipoprotein A; Gene expression data, including UGCG, GNAI1, PLCB4, and PTEN; Protein abundance data, including ANGPTL3, APOA4, APOB, APOC3, LCAT, LDLR, and PCSK9; and, Metabolite abundance data, including 1-(1Z-octadecenyl)-sn-glycero-3-phosphocholine, 1-docosahexaenoyl-2-hydroxy-sn-glycero-3-phosphocholine, 1-lignano-2-hydroxy-sn-glycero-3-phosphocholine, 1-palmitoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, 1-palmitoyl-2-stearoyl-sn-glycero-3-phosphocholine, 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine, 1-stearoyl-sn-glycero-3-phosphocholine, 1,2-arachidonyl-sn-glycero-3-phosphocholine, 1,2-dioleoyl-sn-glycero-3-phosphate, 1,2-dioleoyl-sn-glycerol, Pc 38:4, Pc 40:7, phosphorylcholine, N-lauroyl-d-erythrylphosphatidylcholine, and Pro-Trp.

22. The training method of claim 21, wherein the coefficient a and the constant b of each age-related feature in the age estimation model are:

23. An age estimation model, which is trained by the method according to any one of claims 12 to 22.

24. An age estimation device, comprising: An input unit for inputting age-related characteristics of a subject, wherein the age-related characteristics are as defined in claim 15, 17, 19 or 21; an estimation unit, configured to calculate the estimated age of the subject according to the input age-related features using the age estimation model according to claim 23; and an optional output unit for outputting the estimated age of the subject.

25. A method for estimating age, comprising: Providing a subject with an age-related characteristic, wherein the age-related characteristic is as defined in claim 15, 17, 19 or 21; The age-related features are input into the age estimation model described in claim 23 or the age estimation device described in claim 24 to estimate the age of the subject.

26. An electronic device comprising: Memory; and A processor coupled to the memory, the processor being configured to execute the training method of the age estimation model according to any one of claims 12 to 22, or the age estimation method according to claim 25, based on instructions stored in the memory.

27. A non-volatile computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the training method for the age estimation model described in any one of claims 12 to 22, or the age estimation method described in claim 25.

28. Use of the method for establishing a data set according to any one of claims 1 to 11, the method for training an age estimation model according to any one of claims 12 to 22, the age estimation model according to claim 23, the age estimation device according to claim 24, the age estimation method according to claim 25, the electronic device according to claim 26, or the non-volatile computer-readable storage medium according to claim 27 in preparing a kit, wherein the kit is used for one or more of the following: 1) Estimate the age of the subjects; 2) Evaluate or assist in evaluating the aging rate or degree of subjects; 3) Evaluate or assist in evaluating the effectiveness of aging interventions; 4) Evaluate or assist in evaluating the risk of aging-related diseases (such as cardiovascular diseases, neurodegenerative diseases, diabetes or chronic obstructive pulmonary disease); 5) Evaluate or assist in evaluating the therapeutic effects of aging-related diseases; 6) Evaluate or assist in evaluating the rate or extent of hormonal aging; 7) Evaluate or assist in evaluating the rate or extent of lipid aging; 8) Evaluate or assist in evaluating the rate or extent of immune senescence.

Images