Model construction method, biological age prediction method and device, and storage medium

Abstract

A model construction method, a biological age prediction method and apparatus, and a storage medium, the model construction method comprising: generating a first data set, the first data set comprising gene expression level and methylation level data corresponding to a plurality of genes, selecting a plurality of candidate genes from the first data set, the gene expression level corresponding to each candidate gene is in significant negative correlation with the methylation level; a training data set and a verification data set are generated, the training data set comprises methylation level data of the multiple first genes, and the verification data set comprises methylation level data of the multiple second genes; taking an intersection of the candidate gene, the first gene and the second gene to obtain an intersection gene; and constructing a model by using methylation level data corresponding to a plurality of intersection genes in the training data set and the verification data set as input features. According to the method, the gene with biological significance is selected as the input feature of the model, so that a genetic and biological basis is added for the model, and the robustness of the model is enhanced.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to, but are not limited to, the field of biotechnology, and in particular to a model building method, a biological age prediction method and device, and a storage medium. Background Art

[0002] Breast cancer (BC) is the most common female malignancy, and its overall incidence has been increasing over the past 20 years, especially in the younger patient population under 45 years old. Compared with older patients, younger patients are usually more seriously ill at diagnosis, with more aggressive pathological types and a higher risk of recurrence and death. However, all breast cancer patients currently receive the same treatment regimen, which often leads to under- or over-treatment of young patients, with associated side effects and unnecessary economic burden. More importantly, the etiology and underlying biological mechanisms of breast cancer in young patients remain poorly understood. Breast cancer is a heterogeneous disease with different molecular subtypes, including luminal A, luminal B, basal-like breast cancer (BLBC), triple-negative breast cancer (TNBC), and human epidermal growth factor receptor 2 (HER2) overexpression, which are closely related to treatment response and prognosis. Among them, the more aggressive TNBC and HER2-positive breast cancer patients are more common in the younger patient population than in the overall population, and the luminal type seems to have a worse prognosis in younger female patients. Somatic mutations in the Breast Cancer Susceptibility Gene (BRCA) can explain the association between molecular subtypes and clinical features of breast cancer in some young women. However, hereditary breast cancer accounts for only a small proportion of young-onset patients (about 12%) and therefore cannot fully explain its rapidly increasing incidence, more aggressive disease, and relatively poor prognosis. Summary of the invention

[0003] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0004] The present disclosure provides a model building method, including: Generate a first data set, the first data set includes gene expression level and methylation level data corresponding to a plurality of genes, select a plurality of candidate genes from the first data set, and there is a significant negative correlation between the gene expression level and the methylation level corresponding to each of the candidate genes; Generate a training data set and a validation data set, wherein the training data set includes methylation level data corresponding to a plurality of first genes, and the validation data set includes methylation level data corresponding to a plurality of second genes; Taking the intersection of the plurality of candidate genes, the plurality of the first genes and the plurality of the second genes to obtain a plurality of intersection genes; The model is constructed using the methylation level data corresponding to the plurality of intersection genes in the training data set and the validation data set as input features.

[0005] An embodiment of the present disclosure also provides a model building device, including a memory; and a processor connected to the memory, wherein the memory is used to store instructions, and the processor is configured to execute the steps of the model building method described in any embodiment of the present disclosure based on the instructions stored in the memory.

[0006] The embodiments of the present disclosure also provide a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the model building method described in any embodiment of the present disclosure is implemented.

[0007] An embodiment of the present disclosure further provides a program product, comprising instructions. When the computer program product is executed by a computer, the instructions execute the model building method as described in any embodiment of the present disclosure.

[0008] The present disclosure also provides a biological age prediction method, including: Obtaining a DNA methylation aging clock model, wherein the DNA methylation aging clock model is constructed by the model construction method described in any embodiment of the present disclosure; Acquiring methylation data of the sample to be tested, and obtaining the methylation level of the sample to be tested on the intersection gene according to the methylation data of the sample to be tested; The methylation level of the sample to be tested on the intersection gene is input into the DNA methylation aging clock model to obtain the biological predicted age of the sample to be tested.

[0009] An embodiment of the present disclosure also provides a biological age prediction method device, including a memory; and a processor connected to the memory, the memory is used to store instructions, and the processor is configured to execute the steps of the biological age prediction method described in any embodiment of the present disclosure based on the instructions stored in the memory.

[0010] The embodiments of the present disclosure also provide a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the biological age prediction method described in any embodiment of the present disclosure is implemented.

[0011] An embodiment of the present disclosure also provides a program product, including instructions. When the computer program product is executed by a computer, the instructions execute the biological age prediction method as described in any embodiment of the present disclosure.

[0012] The model building method, biological age prediction method and device, and storage medium of the embodiments of the present disclosure generate a first data set, the first data set includes gene expression levels and methylation level data corresponding to multiple genes, and multiple candidate genes are selected from the first data set, and the gene expression level corresponding to each candidate gene is significantly negatively correlated with the methylation level; generate a training data set and a verification data set, the training data set includes methylation level data of multiple first genes, and the verification data set includes methylation level data of multiple second genes; take the intersection of multiple candidate genes, multiple first genes, and multiple second genes to obtain intersection genes; use the methylation level data corresponding to the intersection genes in the training data set and the verification data set as input features to build a model, make full use of multiple functional genes related to aging, and add genetic and biological foundations to the model by selecting genes with biological significance as input features of the model, thereby further enhancing the overall robustness of the model.

[0013] Other features and advantages of the present disclosure will be described in the following description, and partly become apparent from the description, or be understood by implementing the present disclosure. Other advantages of the present disclosure can be realized and obtained by the schemes described in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings are used to provide an understanding of the technical solution of the present disclosure and constitute a part of the specification. Together with the embodiments of the present disclosure, they are used to explain the technical solution of the present disclosure and do not constitute a limitation on the technical solution of the present disclosure.

[0015] Figure 1 A flowchart of a model building method provided for an exemplary embodiment of the present disclosure; Figure 2 A schematic diagram of KEGG pathway enrichment analysis of multiple intersection genes according to an exemplary embodiment of the present disclosure; FIG. 3A to FIG. 3C A schematic diagram of the aging rate analysis results of paired matched tumor samples and adjacent normal tissue samples in TCGA-BRCA and Sun Yat-sen University (SYSU) Cancer Center breast cancer data cohorts using the DNA methylation aging clock model constructed by the present disclosure; FIG. 4A to FIG. 4D A schematic diagram showing the comparison of prediction results on the same data set using the DNA methylation aging clock model constructed by the present invention and three other publicly published aging clock models; Figure 5 A schematic flow chart of a biological age prediction method provided for an exemplary embodiment of the present disclosure; Figure 6 A schematic diagram of the structure of a model building device provided for an exemplary embodiment of the present disclosure; Figure 7 A schematic diagram of the structure of a biological age prediction device provided by an exemplary embodiment of the present disclosure; Figure 8 A schematic structural diagram of another model building device provided for an exemplary embodiment of the present disclosure; Fig. 9 A schematic structural diagram of another biological age prediction device provided by an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0016] The present disclosure describes multiple embodiments, but the description is exemplary rather than restrictive, and it is apparent to those skilled in the art that there may be more embodiments and implementations within the scope of the embodiments described in the present disclosure. Although many possible feature combinations are shown in the drawings and discussed in the specific embodiments, many other combinations of the disclosed features are possible. Unless specifically limited, any feature or element of any embodiment may be used in combination with any other feature or element in any other embodiment, or may replace any other feature or element in any other embodiment.

[0017] The present disclosure includes and contemplates combinations of features and elements known to those of ordinary skill in the art. The embodiments, features, and elements disclosed in the present disclosure may also be combined with any conventional features or elements to form a unique invention scheme defined by the claims. Any features or elements of any embodiment may also be combined with features or elements from other invention schemes to form another unique invention scheme defined by the claims. Therefore, it should be understood that any feature shown and / or discussed in the present disclosure may be implemented individually or in any appropriate combination. Therefore, except for the limitations made according to the attached claims and their equivalents, the embodiments are not subject to other limitations. In addition, various modifications and changes may be made within the scope of protection of the attached claims.

[0018] In addition, when describing representative embodiments, the specification may have presented the method and / or process as a specific sequence of steps. However, to the extent that the method or process does not rely on the specific order of the steps described herein, the method or process should not be limited to the steps in the specific order described. As will be appreciated by those of ordinary skill in the art, other orders of steps are also possible. Therefore, the specific order of the steps set forth in the specification should not be interpreted as a limitation to the claims. In addition, the claims for the method and / or process should not be limited to the steps performed in the order written, and those skilled in the art can easily understand that these orders can be changed and still remain within the spirit and scope of the disclosed embodiments.

[0019] From the perspective of aging, cancer can be considered an age-related disease. With the increase of age, the accumulation of cell damage is often considered to be the reason why the elderly population is more susceptible to cancer. In fact, many biological processes associated with cancer characteristics are also closely related to the characteristics of the aging process, such as epigenetic changes, increased cell aging, and abnormal protein homeostasis. DNA methylation is an important epigenetic regulatory mechanism that plays a key role in the occurrence and development of tumors. Studies have shown that significant changes in DNA methylation levels in breast cancer tissues are closely related to the molecular typing, invasiveness, and prognosis of tumors. Therefore, DNA methylation detection is expected to become a new biomarker for predicting the prognosis of breast cancer patients. In recent years, it has been found that cell senescence is also an important factor affecting tumor prognosis. Cell senescence is a stable state of cell growth arrest, which is usually accompanied by biological changes such as DNA damage, chromatin remodeling, and metabolic abnormalities. More and more evidence shows that the degree of senescence of tumor cells is closely related to their proliferation, invasion, and metastasis capabilities. Therefore, establishing an "aging clock" detection method that can accurately assess the level of cellular senescence is of great significance for predicting tumor prognosis. In particular, age-related DNA methylation biomarkers have been used to construct an "epigenetic clock", a statistical model that can predict an individual's actual age. Interestingly, recent studies have found that the clock-predicted age of tumor samples is significantly greater than the actual age compared with normal samples, indicating that accelerated aging exists in tumors. Although some new insights into cancer have been gained from the perspective of aging, the heterogeneity of accelerated aging in cancer remains to be further explored.

[0020] like Figure 1 As shown, the embodiment of the present disclosure provides a model building method, including: Step 101: Generate a first data set, the first data set includes gene expression levels and methylation level data corresponding to a plurality of genes, select a plurality of candidate genes from the first data set, and the gene expression level corresponding to each candidate gene is significantly negatively correlated with the methylation level; Step 102: Generate a training data set and a validation data set, wherein the training data set includes methylation level data corresponding to a plurality of first genes, and the validation data set includes methylation level data corresponding to a plurality of second genes; Step 103, taking intersections of the plurality of candidate genes, the plurality of first genes and the plurality of second genes to obtain a plurality of intersection genes; Step 104: Use the methylation level data corresponding to multiple intersection genes in the training data set and the validation data set as input features to build a model.

[0021] The model building method provided by the embodiment of the present disclosure generates a first data set, wherein the first data set includes gene expression levels and methylation level data corresponding to multiple genes, selects multiple candidate genes from the first data set, and the gene expression level corresponding to each candidate gene is significantly negatively correlated with the methylation level; generates a training data set and a validation data set, wherein the training data set includes methylation level data corresponding to multiple first genes, and the validation data set includes methylation level data corresponding to multiple second genes; takes the intersection of multiple candidate genes, multiple first genes, and multiple second genes to obtain intersection genes; uses the methylation level data corresponding to multiple intersection genes in the training data set and the validation data set as input features to build a model, making full use of multiple functional genes related to aging, and by selecting genes with biological significance as input features of the model, adding genetic and biological foundations to the model, further enhancing the overall robustness of the model.

[0022] In some exemplary embodiments, the methylation level corresponding to the candidate gene is the average of the methylation levels of multiple methylation sites included in the candidate gene; the methylation level corresponding to the first gene is the average of the methylation levels of multiple methylation sites included in the first gene; and the methylation level corresponding to the second gene is the average of the methylation levels of multiple methylation sites included in the second gene.

[0023] In the disclosed embodiments, by using the methylation level data at the gene level (i.e., the average value of the methylation levels of multiple methylation sites included in the gene) to construct a model, it is possible to better reflect the overall state of gene epigenetic regulation, thereby better resisting the impact of the lack of DNA methylation probes, and thus showing stable and reliable prediction performance in various experimental platforms and sample types.

[0024] In some exemplary embodiments, the training target of the model may be the actual age of the samples in the training data set, however, the present disclosure is not limited in this regard.

[0025] In the embodiment of the present disclosure, the actual age of the sample may be expressed in years, however, the present disclosure is not limited to this.

[0026] In some exemplary embodiments, the training target of the model can also be set by the following method: Get the actual age of the samples in the training data set and transform the actual age of the samples as follows: When the actual age of the sample is less than or equal to the preset age threshold a, the training target of the model is set to: F(age)=log(age+1)-log(m+1); When the actual age of the sample is greater than the preset age threshold a, the training target of the model is set to: F(age)=(age-m) / (m+1), where a is a real number greater than 15.

[0027] The disclosed embodiment can make the age distribution more reasonable through this age conversion method, so that the age prediction target is closer to the normal distribution during prediction.

[0028] Exemplarily, the preset age threshold a may be 18 or 20 years old, however, the present disclosure is not limited thereto.

[0029] Taking the preset age threshold a of 20 years old as an example, in order to improve the model prediction effect, in the embodiment of the present disclosure, we transform the sample age as follows: If age≤20, then F(age)=log(age+1)-log(20+1); If age>20, then F(age)=(age-20) / (20+1).

[0030] The embodiment of the present disclosure first generates a first data set, the first data set includes gene expression levels and methylation level data corresponding to multiple genes, selects multiple candidate genes from the first data set, and the gene expression level corresponding to each candidate gene is significantly negatively correlated with the methylation level, then takes the intersection of the multiple candidate genes, the multiple first genes and the multiple second genes to obtain the intersection genes, and performs KEGG pathway enrichment analysis on the characteristic genes (intersection genes) used in the model. The characteristic genes used in the present disclosure are significantly enriched in pathways closely related to aging and cancer progression.

[0031] In some exemplary embodiments, generating a first data set includes: Downloading gene expression level data and methylation data of multiple disease patient samples from a first database; Preprocess the methylation data of samples from patients with multiple diseases to obtain methylation level data of multiple sites; The methylation level data of multiple sites belonging to the same gene were averaged to obtain the methylation level data corresponding to each gene.

[0032] Exemplarily, the first database may be the Cancer Genome Atlas (TCGA) database, but the present disclosure is not limited thereto.

[0033] The following is an introduction using the first database, which is the BRCA sub-database in the TCGA database. TCGA-BRCA is a sub-project in TCGA that is dedicated to understanding and studying the genomic characteristics of breast cancer. However, the present disclosure is not limited to this, and the model construction method of the present disclosure can also be used to predict the biological age of patients with any other disease type.

[0034] In some exemplary embodiments, the methylation data downloaded from the first database may be DNA methylation sequencing data or microarray data.

[0035] In some exemplary embodiments, the methylation data is preprocessed, including: Perform quality control and filtering on methylation data; The methylation data after quality control and filtering were compared to the human reference genome to obtain methylation level data for multiple sites.

[0036] In some exemplary embodiments, considering that the methylation level data of sites located on sex chromosomes may be affected by gender factors, which is not conducive to establishing a universal prediction model, before averaging the methylation level data of multiple sites belonging to the same gene, the method also includes: eliminating the methylation level data of sites located on sex chromosomes.

[0037] In the disclosed embodiment, taking the methylation data of multiple disease patient samples downloaded from the TCGA database (first database) as an example, before building the model, the CpG probes are first screened and processed. Specifically, all CpG probes located on the sex chromosomes are first eliminated, because these probes may be affected by gender factors, which is not conducive to establishing a universal prediction model. For multiple CpG probes corresponding to the same gene, the average value (beta value) of the probe methylation level is used as a representative indicator of the gene. The purpose of this is to eliminate measurement errors or fluctuations between probes to the greatest extent possible, and to more accurately reflect the overall methylation status of the gene in tumor samples.

[0038] In some exemplary embodiments, selecting candidate genes from the first data set includes: For each gene in the first data set, the Pearson correlation coefficient and the significant P value between the gene expression level and the methylation level corresponding to the gene are calculated; Genes whose Pearson correlation coefficient is less than the preset correlation coefficient threshold and whose significant P value is less than the preset significance threshold are selected as candidate genes.

[0039] In some exemplary embodiments, the preset correlation coefficient threshold may be between -1 and 0, and the preset significance threshold is less than or equal to 0.05 (for example, it may be 0.05, 0.01, 0.001, etc.). Optionally, the preset correlation coefficient threshold is 0, and the preset significance threshold is 0.05.

[0040] For example, in order to screen candidate genes with biological significance, we collected gene expression profiles and DNA methylation profiles of the TCGA-BRCA cohort. Specifically, we first selected gene symbols whose gene expression profiles showed a significant negative correlation with their DNA methylation levels (Pearson correlation coefficient <0 and P<0.05). The basis of this screening strategy is that during the development of tumors, high DNA methylation in the promoter region of genes usually leads to transcriptional inhibition of the gene, thereby reducing its expression level. Therefore, genes whose expression profiles and methylation levels show a significant negative correlation are more likely to be involved in the pathogenesis of breast cancer.

[0041] In some exemplary embodiments, in step 101, generating a training data set includes: Downloading methylation data of multiple healthy human samples from a second database; Preprocessing the methylation data of multiple healthy human samples to obtain methylation level data of multiple sites of the multiple healthy human samples; The methylation level data of multiple sites belonging to the same gene in multiple healthy human samples are averaged to obtain the methylation level data corresponding to multiple first genes.

[0042] Exemplarily, the second database may be a Gene Expression Omnibus (GEO) database, however, the present disclosure is not limited thereto.

[0043] In the disclosed embodiment, taking the methylation data of multiple healthy human samples downloaded from the GEO database (second database) as an example, before building the model, the CpG probes are first screened and processed. Specifically, all CpG probes located on the sex chromosomes are first eliminated, because these probes may be affected by gender factors, which is not conducive to establishing a universal prediction model. For multiple CpG probes corresponding to the same gene, the average value (beta value) of the probe methylation level is used as a representative indicator of the gene. The purpose of this is to eliminate the measurement errors or fluctuations between probes to the greatest extent, and to more accurately reflect the overall methylation status of the gene in the tumor sample.

[0044] In some exemplary embodiments, the methylation data downloaded from the second database may be DNA methylation sequencing data or microarray data.

[0045] In some exemplary embodiments, in step 101, generating a verification data set includes: Downloading methylation data of samples of patients with multiple diseases from a third database; Preprocessing the methylation data of multiple disease patient samples to obtain methylation level data of multiple sites of multiple disease patient samples; The methylation level data of multiple sites belonging to the same gene in multiple disease patient samples are averaged to obtain the methylation level data corresponding to multiple second genes.

[0046] Exemplarily, the third database may be a self-test database and / or the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) database, however, the present disclosure is not limited thereto.

[0047] In some exemplary embodiments, the methylation data downloaded from the third database (self-test database) may be DNA methylation sequencing data or microarray data.

[0048] In other examples, the methylation data downloaded from the third database (METABRIC) may be a methylation expression profile matrix based on the gene level.

[0049] Exemplarily, taking the whole genome bisulfite sequencing data downloaded from the third database (self-test database) as an example, when pre-processing the data, the sequencing data can be first subjected to quality control filtering, and the filtered sequencing data can be compared to obtain a BAM file, remove the repeated sequences (reads) in the BAM file, extract the methylation level data of all sites of all samples, and merge the methylation level data of all sites of all samples into a unified data matrix. Then calculate the methylation level of 2kb upstream and 0.5kb downstream of the gene transcription start site, and the specific calculation method is to divide the number of methylated sequences measured at all CpG sites in the promoter region of the gene by the total number of sequences. If 20% of the samples are missing values ​​or have low coverage in the promoter region, the data in the promoter region is not considered; finally, the methylation data matrix of the self-test data is obtained, and the methylation data matrix includes multiple promoter regions and methylation level data corresponding to the multiple promoter regions, and the value range of the methylation level data is between 0 and 1.

[0050] After obtaining the first data set for selecting candidate genes, the training data set for training, and the validation data set for validation, in order to better construct a DNA methylation aging prediction model in the future, genes detected in different data sets were further screened out, that is, multiple intersection genes existing in all data sets were obtained.

[0051] In one example, when screening candidate genes, 9174 candidate genes were screened from the TCGA database. Then, genes that intersect with the aforementioned 9174 candidate genes were selected from the self-test database, the METABRIC database, and the GEO database, and 656 intersection genes were obtained as feature genes used to construct the model, which were identified by Symbol names as follows: CDK10, CD2BP2, MXD4, SND1, TMEM43, RFX5, MXRA7, SH2B3, WWP2, RRS1, GDF7, ARPC2, S100A13, ARRDC4, DCAF4L2, SPPL2B, DPH2, ANAPC11, PLEC, CEP68, RNF111, ZFP41, PLA2G15, SNN, ELOVL5, NGFR, NR2F2, EPB41L3, THBS1, RBM8A, FAM126A, TSPYL5, MGRN1, RUNX3, ENOPH1, MYCL, AP1S1, CAPS2, CBX2, STAG3 , FOXP1, ATP1B2, C1orf220, HOXC13, FYTTD1, KDM1A, BANP, RHOQ, HERC1, CLCN3, ARAP3, NUDT1, UBTD2, POMP, SOX11, MORF4L1, IFFO1, MAP3K5, SMPD4, ABHD 2, MEF2D, ARHGEF2, GABBR2, FAM91A1, SPG21, MCM7, TRAPPC8, KAZN, ZBTB14, CD46, KDM1B, HOTAIRM1, GNAI3, MAP3K6, NFATC2IP, MRPL22, RAB27A, NUTF2, ARG2, HGS, SLC38A1, NFE2L2, SMAD7, HIVEP3, SPSB1, ZNF517, ANXA2, RASGRP2, DHODH, ESPL1, OSGIN2, MCCC1, TTL, CALR3, ENAH, TMEM184B, AMOTL2, SERP1 , RHOA, MLLT11, MT1E, TRIB1, PREP, ITPR2, ACAD9, WDFY2, SCAP, TNS3, LINC00472, SOHLH2, LRPAP1, BIN2, GABRA2, MIRLET7BHG, MGAT1, ZNF646, BNC1, AU NIP, EPHB3, C12orf60, KIAA0319L, PNKD, GAPDH, WLS, IRX4, DNM3, LMO7, ERI2, PCNP, CSNK1E, PTK7, ZFAND3, NEIL1, SIRT7, CCDC77, EFHD1, AP3B1, RAD9B,<h2 style=";text-align:left;direction:ltr">ARL10,RNF32,RNF139,POU2F1,ICA1L,MTERF3,NSMCE2,ZNF274,TTC12,INP P5B,SLC35C1,CNST,WASF2,EXO1,RCAN3,DUSP4,CBX3,CD81,ELAVL4,EIF4E ,ETV1,C3orf52,PTCH2,MEGF8,CSRNP1,PDE8A,C19orf48,SNAP29,PRKD2,C DK9,ZC3H18,GRIA4,CTDP1,PRMT1,ACTR3,MGST3,BLOC1S6,PAX8,ACSS3,CHM P1A,NLRP14,HSPG2,BOLL,EVC,SRPK2,NTN4,HS3ST1,RARB,MOB3A,OTX1,AP 1S3,AKT1S1,LTV1,C18orf54,GLUL,SRF,MAPK8IP3,EAF2,ACTR6,CCDC82,P TPRJ,TLE3,MX1,LYNX1,CPT1C,SMAD4,FPGT,YWHAZ,SMO,COQ7,BCAT2,CASC 3,CCND2,RASSF3,AHCYL2,IGFBP3,PHTF1,CHERP,MYH10,SDC1,ISYNA1,CAMK 2N1,MT1H,PHLDA1,NFATC4,SYNM,TNFAIP1,LMNB2,PENK,ZNF420,MMS22L,D APK3,COLGALT1,TUBA4A,ATN1,SLAIN1,KBTBD8,ST14,RAF1,SMG5,LPXN,RB M42,ZNF667,CCDC130,KRI1,TUBA1C,CPNE5,BCL2L13,RPL8,LIMK2,BMP4,A CTN4,UPP1,CYGB,ZNF256,HMGA1,CBX7,FLVCR2,MDFI,FAM216A,PPAN,CYP1B 1,NELFA,CRACR2A,RPL18A,SDHC,ABCC5,CNNM3,TBX20,PRKCSH,NDUFB2,PO LR2D,AGBL5,FBXO45,EXOSC3,RUNX2,GAS1,PITPNA,B4GALT2,BOC,PIK3CA,D YNC1H1,LEP,CTDSP2,MRC2,CFLAR,CRIP2,AHNAK,SLC39A1,HOXA11,EIF5,R AD51B,CYHR1,DYNLRB2,FHL3,CUL4A,GFOD1,TAS1R1,FOXA1,HADHB,GPRC5B,<h2 style=";text-align:left;direction:ltr">RPL28,CIART,RWDD2A,PDSS2,ERRFI1,AKAP8,G2E3,BCAP29,TPRG1L,FKBP4 ,DUSP10,GART,MOK,IL6R,ECSIT,DAAM1,PEX14,MYCN,PIGZ,MED22,DTD2,MA N1A1,ZNF584,SOX7,KLC1,SKP1,RPL36,ARAP2,RALBP1,COPS7B,EXOC2,ANK MY2,LMO2,FABP5,VAT1,PTPN7,GALNT7,SLC25A29,CDX1,AMIGO2,AK9,STPG1 ,PTGR2,SNIP1,NPY2R,ADAM15,SGK1,ZNF785,CGGBP1,MEN1,COX10,CFAP36 ,SYT2,PTGER4,GALE,PAK4,GPX7,ANKRD35,YWHAH,FBXO32,PIK3R3,ECI1,GA LNT13,PKP1,TP53BP2,TEAD4,SLC11A2,ITGA6,MANF,NDUFA7,TMC7,NTHL1, ARNTL,NR1D1,THEM5,SZT2,RBP1,DSP,TAF1D,FOXB1,ZNF655,PPP1R3B,IMP4 ,BUD31,TCF7,MTHFR,FUZ,ZNF660,ZNF354C,SLC30A1,NSUN7,HSP90AA1,RN F144A,KCNJ11,SLC40A1,BTG3,ZSCAN2,CHST6,EOMES,BHLHE40,ZNF496,ZNF 197,MCHR2,ROCK1,UBE2I,GOLIM4,FTH1,WWP1,TPBG,FANCC,RABGEF1,S100 A11,SSRP1,GPR135,KLHDC2,POFUT2,PC,LYSMD1,NOTCH3,C21orf91,IPO13, ZNF324,FOXC2,VKORC1,ZNF615,SPRYD3,MRPL36,ZNF57,BLM,ACVR2B,HNRN PUL1,DIS3L,PTTG1IP,SLC39A8,CNP,NEK11,NOL7,ZNF227,SMC4,CNPPD1,HD AC11, RIC8B, MYO9A, SLC9B2, LOX, TXNDC9, CEBPA, SDHD, RASSF5, TRIM65, TMEM92, TSKU, ATF7, BCAT1, SLC39A9, CNKSR3, CCDC51, SPHK2, TIMM13, IRF2BPL<h2 style=";text-align:left;direction:ltr">TSPAN1,SLC25A22,ZNF721,GSTM4,PEX16,ALDH1L1,MEX3B,TP53I11,RHPN2 ,RELA,CALM2,GUK1,PCYT2,RPS5,BMPR2,HSPH1,SIPA1,CDC20,UBE2V2,TRMT 112,CYP39A1,LETM1,SMAD6,IRX5,NDUFA4L2,ATP6V0B,ADCY3,EMB,AP1AR, RAP2B,CDKN1B,TRIM14,ADRB2,TMCO1,CDK5R1,SLC35B3,SMG7,MEGF10,GADD 45A,DPAGT1,EDEM1,EIF4G1,NMBR,RAD21,TPT1,IZUMO2,SUPT6H,XRCC1,SO CS2,ZC3H7A,TMEM41A,EZH2,ACAD11,TUBA8,USP49,BRINP2,MIIP,STAT5B,S TIP1,FADS2,ETV3,EIF5A2,NOVA1,PLEKHJ1,PTRH1,TAX1BP3,FOXO3,HSPA4 ,CNIH1,ING2,ANKRD13A,GANAB,TACR1,UBAC2,MYO1B,S1PR3,HOXB4,MEF2C, C12orf57, TMTC4, LYPD6, FBXO27, TTC9C, STEAP2, HERC5, TFG, RAB35, STX4, SUP5H, ALG3, RAB33B, TFCP2, GNMT, UCHL1, USP13, CA3, SNRPA, FAR1, ALKBH4 ,CDC42EP3,SMARCA4,AKAP1,ZYX,ABT1,ADAMTS15,PRKAR1A,TLX3,ACSF2,R PS10,MIEF1,MAU2,CWF19L2,CDC42EP1,TMEM107,SWI5,MCM2,S1PR1,ZAR1,S LC35G2,MXD3,MBOAT2,GRK4,RELB,HINFP,QTRT1,MATK,ZNF408,SOD3,DPEP 3,EDEM3,REV3L,FLCN,LIN7C,TUT1,EFNA1,DSC3,CPEB4,CA12,SYT12,MSANT D2, SNX24, OSTC, N4BP2L2, MSC, PTPN18, CHMP2B, RPS19, RWDD2B, RPS7, NETO1, GK5, C1orf43, FBXL12, MAPK7, BACH1, PXDC1, PTRH2, PSMC3, SFT2D2, SDHBE2F3, CEP85, SIRT3, RIOK1, RPL7, STOM, TRIM60, COL4A4, MAP4K2, TCP11L1, UBN1, ZNF639, ANPEP, MXD1, TOB1, COLCA1, ZNF177, RGS2, TPT E, LMAN1, CCPG1, RPL12, BRWD1, TLE4, WEE1, SLC13A5, EDN1, PLEKHG6, ALDH1A3, MOCOS, SPARC, TYRO3, OSBPL8, GTF2I, PLD3, ODF2, CHRDL2. ,

[0052] We performed KEGG pathway enrichment analysis on these characteristic genes used in model construction to observe whether the genes used were significantly enriched in pathways related to aging and cancer progression. Figure 2 As shown, the results of enrichment analysis showed that these genes were significantly enriched in multiple biological pathways closely related to the aging process. This result fully verifies that the DNA methylation aging clock model we established has a solid biological basis. Specifically, the pathways enriched by these key genes include cell cycle regulation (Cell cycle), neurotrophic factor (TGF)-β signaling pathway, circadian rhythm signaling pathway, etc. These pathways play a key role in the aging process of the body, and their abnormal functions can lead to typical aging manifestations such as uncontrolled cell proliferation and unbalanced cell apoptosis. In short, the results of KEGG enrichment analysis fully reveal the biological functions and molecular mechanisms of the key genes we screened out related to aging, which provides strong support for the biological significance of the DNA methylation aging clock model.

[0053] In some exemplary embodiments, the model may be an elastic network prediction model. However, the present disclosure is not limited thereto.

[0054] In some exemplary embodiments, the method further comprises: The methylation level data in the training dataset and the validation dataset were de-batched.

[0055] In the disclosed embodiment, the methylation level data in the training data set and the validation data set are de-batched to minimize the non-biological differences caused by the sequencing platform. Then, the methylation level data of the selected intersection genes are used to construct an elastic network prediction model.

[0056] In one example, the R package glmnet was used to select the best model hyperparameter lambda with hyperparameter alpha=0.5, nfolds=10, and family='gaussian'. Then, the DNA methylation aging clock model was constructed using the hyperparameter lambda of the best model selected. At the same time, the intercept term of the regression model and the model coefficient of each gene were obtained using the final model obtained by training.

[0057] In summary, in one example, the model building method of the embodiment of the present disclosure includes: obtaining the TCGA transcriptome and DNA methylation CpG sequencing data of the public data set of breast cancer, preprocessing the sequencing data, and obtaining the TCGA breast cancer methylation expression spectrum at the gene level matching the transcriptome sample. After that, if the expression level of a gene is significantly negatively correlated with the methylation level matched with it, it is selected as a candidate gene; then the selected candidate genes are intersected with the sequencing genes of the breast cancer data set involved in modeling and verification to ensure the consistency of features between different samples and platforms; then the obtained intersection feature genes are used to perform 10-fold cross validation based on the elastic network algorithm on public healthy human blood / tissue samples to determine the optimal model parameters; finally, the determined optimal parameters are used, and the selected optimal features are used to train the methylation aging clock model based on the elastic network algorithm on the public data set.

[0058] In the disclosed embodiments, in order to further evaluate the prediction ability of the model, we used the Pearson correlation coefficient to determine the accuracy of the predicted age and the actual age. At the same time, we selected matching tumor and adjacent normal data from the TCGA-BRCA database and the self-test database, and further evaluated the robustness of the model by comparing the differences in methylation age between matching samples.

[0059] To test the reliability of the established DNA methylation aging clock model, we applied it to paired matched tumor samples and adjacent normal tissue samples in TCGA-BRCA and the self-test database (breast cancer data cohort of the Cancer Center of Sun Yat-sen University (SYSU)), and calculated the ratio of the predicted biological age to the actual age of these samples, the so-called "aging rate". Figure 3A As shown in Figure 2, the aging rates of these tumors and their adjacent normal samples were significantly higher than those of healthy control samples. Figure 3B and Figure 3CAs shown in the figure, the aging rate of tumor samples in the TCGA-BRCA and SYSU Cancer Center breast cancer data cohorts was significantly increased compared with their adjacent normal tissues. This result fully verifies that the DNA methylation aging clock model we constructed has good reliability and discrimination ability. The accelerated aging of tumor tissues may reflect that during the development of tumors, tumor cells generally experience more serious epigenetic abnormalities and physiological dysfunctions, thus presenting a faster cell aging phenotype.

[0060] Given that gene-level DNA methylation measurements can resist the impact of single probe loss by integrating the average methylation levels of multiple probes, we used other publicly published aging clock models and compared them with the gene-level DNA methylation aging clock model disclosed in this paper on the same dataset to further verify the robustness of the gene-level DNA methylation aging clock model we constructed.

[0061] Specifically, we applied our model together with three other publicly published aging clock models (Horvath S2018, McEwen L2019, and Shireby G2020) to a DNA methylation dataset GSE41037 based on the 27K microarray platform. FIG. 4A to FIG. 4D As shown in the figure, the results show that our model achieves higher correlation and lower mean absolute error between actual age and predicted DNA methylation age compared with other models. This result strongly confirms that the gene-level DNA methylation aging clock model we developed has excellent robustness and versatility. Compared with the previous aging clock based on a single CpG site, our model can better resist the influence of DNA methylation probe loss, thus showing stable and reliable prediction performance in various experimental platforms and sample types. This excellent anti-interference ability can be partly attributed to the fact that we use methylation data at the gene level rather than a single CpG site as model input. The average methylation value at the gene level can better reflect the overall state of gene epigenetic regulation, thereby reducing the impact of single probe loss on the prediction results. In addition, we adopted a more comprehensive feature selection strategy, making full use of multiple functional genes related to aging, further enhancing the overall robustness of the model. In summary, our research results show that the gene-level DNA methylation aging clock model we established has excellent versatility and adaptability, not only showing excellent performance on different experimental platforms, but also being able to resist common DNA methylation measurement noise, laying a solid foundation for further application in this field.

[0062] In the disclosed embodiments, the aging rate of breast cancer pathological samples (DNA methylation aging predicted age / patient's actual age) can be calculated and associated with the clinical information of the samples in the cohort (such as overall survival, recurrence-free survival, whether the cancer has metastasized, whether it has recurred, and pathological stage) to perform correlation analysis, and the clinical guidance significance of the DNA methylation aging prediction model can be identified through hypergeometric test, Fisher test, and t-test.

[0063] like Figure 5 As shown, the embodiment of the present disclosure also provides a biological age prediction method, including: Step 501: construct a DNA methylation aging clock model, wherein the DNA methylation aging clock model is constructed by the model construction method described in any embodiment of the present disclosure; Step 502: Acquire methylation data of the sample to be tested, and obtain the methylation level of the sample to be tested on the intersection gene according to the methylation data of the sample to be tested; Step 503: input the methylation level of the sample to be tested on the intersection gene into the DNA methylation aging clock model to obtain the biological predicted age of the sample to be tested.

[0064] The DNA methylation aging clock model constructed by the model construction method disclosed in the present invention can be used to predict the actual biological age of breast cancer patients and evaluate their prognosis and ability to predict pathological progression, thereby further providing new technical support for the clinical diagnosis and individualized treatment of breast cancer. The present invention constructs a DNA methylation clock to assist in the precise treatment of breast cancer, thereby making the calculation of the biological age of breast cancer patients easier to promote.

[0065] like Figure 6 As shown, the embodiment of the present disclosure also provides a model construction device, including a data set generation module 610, an intersection gene generation module 620 and a model construction module 630, wherein: The data set generation module 610 is configured to generate a first data set, the first data set includes gene expression levels and methylation level data corresponding to a plurality of genes, select a plurality of candidate genes from the first data set, and the gene expression level and methylation level corresponding to each of the candidate genes are significantly negatively correlated; generate a training data set and a validation data set, the training data set includes methylation level data corresponding to a plurality of first genes, and the validation data set includes methylation level data corresponding to a plurality of second genes; An intersection gene generation module 620 is configured to obtain a plurality of intersection genes by taking the intersection of the plurality of candidate genes, the plurality of the first genes and the plurality of the second genes; The model building module 630 is configured to use the methylation level data corresponding to the plurality of intersection genes in the training data set and the validation data set as input features to build a model.

[0066] In some exemplary embodiments, the methylation level corresponding to the candidate gene is the average methylation level of multiple methylation sites included in the candidate gene; the methylation level corresponding to the first gene is the average methylation level of multiple methylation sites included in the first gene; and the methylation level corresponding to the second gene is the average methylation level of multiple methylation sites included in the second gene.

[0067] In some exemplary embodiments, the data set generation module 610 generates a first data set, including: Download gene expression level data and methylation data of multiple disease patient samples from the first database, Preprocessing the methylation data of the multiple disease patient samples to obtain methylation level data of multiple sites; The methylation level data of multiple sites belonging to the same gene in the multiple disease patient samples are averaged to obtain the methylation level data corresponding to each gene.

[0068] In some exemplary embodiments, the dataset generation module 610 selects candidate genes from the first dataset, including: For each gene in the first data set, calculate the Pearson correlation coefficient and the significance P value between the gene expression level and the methylation level corresponding to the gene; The genes whose Pearson correlation coefficient is less than a preset correlation coefficient threshold and whose significance P value is less than a preset significance threshold are selected as the candidate genes.

[0069] In some exemplary embodiments, the dataset generation module 610 generates a training dataset, including: Downloading methylation data of multiple healthy human samples from a second database; Preprocessing the methylation data of the plurality of healthy human samples to obtain methylation level data of the plurality of sites of the plurality of healthy human samples; The methylation level data of multiple sites belonging to the same gene in the multiple healthy human samples are averaged to obtain the methylation level data corresponding to the multiple first genes.

[0070] In some exemplary embodiments, the model building module 630 sets the training target of the model by: The actual age of the samples in the training data set is obtained, and the actual age of the samples is converted as follows: When the actual age of the sample is less than or equal to the preset age threshold a, the training target of the model is set to: F(age)=log(age+1)-log(a+1); When the actual age of the sample is greater than a preset age threshold a, the training target of the model is set to: F(age)= (age-a) / (a+1), where a is a real number greater than 15.

[0071] like Figure 7 As shown, the embodiment of the present disclosure also provides a biological age prediction device, including a clock model construction module 710, a sample data processing module 720 and a prediction module 730, wherein: The clock model construction module 710 is configured to obtain a DNA methylation aging clock model, wherein the DNA methylation aging clock model is constructed by the model construction method described in any embodiment of the present disclosure; The test sample data processing module 720 is configured to obtain methylation data of the test sample, and obtain the methylation level of the test sample on the intersection gene according to the methylation data of the test sample; The prediction module 730 is configured to input the methylation level of the sample to be tested on the intersection gene into the DNA methylation aging clock model to obtain the biological predicted age of the sample to be tested.

[0072] An embodiment of the present disclosure also provides a model building device, including a memory; and a processor connected to the memory, wherein the memory is used to store instructions, and the processor is configured to execute the steps of the model building method described in any embodiment of the present disclosure based on the instructions stored in the memory.

[0073] like Figure 8As shown, in one example, the model building device may include: a first processor 810, a first memory 820, a first bus system 830 and a first transceiver 840, wherein the first processor 810, the first memory 820 and the first transceiver 840 are connected via the first bus system 830, the first memory 820 is used to store instructions, and the first processor 810 is used to execute the instructions stored in the first memory 820 to control the first transceiver 840 to send and receive signals. Specifically, the first transceiver 840 can obtain sequencing data under the control of the first processor 810, and the first processor 810 generates a first data set, wherein the first data set includes gene expression levels and methylation level data corresponding to multiple genes, and multiple candidate genes are selected from the first data set, and the gene expression level and methylation level corresponding to each of the candidate genes are significantly negatively correlated; a training data set and a verification data set are generated, wherein the training data set includes methylation level data of multiple first genes, and the verification data set includes methylation level data of multiple second genes; the intersection of the multiple candidate genes, the multiple first genes and the multiple second genes is taken to obtain multiple intersection genes; and the methylation level data corresponding to the multiple intersection genes in the training data set and the verification data set are used as input features to construct a model.

[0074] It should be understood that the first processor 810 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0075] The first memory 820 may include a read-only memory and a random access memory, and provide instructions and data to the first processor 810. A portion of the first memory 820 may also include a nonvolatile random access memory. For example, the first memory 820 may also store information on the device type.

[0076] The first bus system 830 may include a power bus, a control bus, a status signal bus, etc. in addition to the data bus. Figure 8 In the figure, various buses are labeled as a first bus system 830 .

[0077] In the implementation process, the processing performed by the processing device can be completed by the hardware integrated logic circuit in the first processor 810 or the instructions in the form of software. That is, the method steps of the embodiment of the present disclosure can be embodied as being executed by a hardware processor, or being executed by a combination of hardware and software modules in the processor. The software module can be located in a storage medium such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the first memory 820, and the first processor 810 reads the information in the first memory 820 and completes the steps of the above method in combination with its hardware. To avoid repetition, it will not be described in detail here.

[0078] The present disclosure also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the model building method described in any embodiment of the present disclosure is implemented. The model building method driven by executing executable instructions is basically the same as the model building method provided in the above embodiment of the present disclosure, and will not be described in detail here.

[0079] In some possible implementations, various aspects of the model building method provided by the present disclosure may also be implemented in the form of a program product, which includes a program code. When the program product is run on a computer device, the program code is used to enable the computer device to execute the steps of the model building method according to various exemplary embodiments of the present disclosure described above in this specification. For example, the computer device may execute the model building method recorded in the embodiments of the present disclosure.

[0080] An embodiment of the present disclosure also provides a biological age prediction device, comprising a memory; and a processor connected to the memory, wherein the memory is used to store instructions, and the processor is configured to execute the steps of the biological age prediction method as described in any embodiment of the present disclosure based on the instructions stored in the memory.

[0081] like Fig. 9As shown, in one example, the biological age prediction device may include: a second processor 910, a second memory 920, a second bus system 930, and a second transceiver 940, wherein the second processor 910, the second memory 920, and the second transceiver 940 are connected via the second bus system 930, the second memory 920 is used to store instructions, and the second processor 910 is used to execute the instructions stored in the second memory 920 to control the second transceiver 940 to send and receive signals. Specifically, the second transceiver 940 can obtain the DNA methylation aging clock model and the methylation data of the sample to be tested under the control of the second processor 910, and the DNA methylation aging clock model is constructed by the model construction method as described in any embodiment of the present disclosure, and the second processor 910 obtains the methylation level of the sample to be tested on the intersection gene according to the methylation data of the sample to be tested; the methylation level of the sample to be tested on the intersection gene is input into the DNA methylation aging clock model to obtain the biological prediction age of the sample to be tested.

[0082] It should be understood that the second processor 910 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0083] The second memory 920 may include a read-only memory and a random access memory, and provide instructions and data to the second processor 910. A portion of the second memory 920 may also include a nonvolatile random access memory. For example, the second memory 920 may also store information on the device type.

[0084] In addition to the data bus, the second bus system 930 may also include a power bus, a control bus, and a status signal bus. Fig. 9 In the figure, various buses are labeled as a second bus system 930.

[0085] In the implementation process, the processing performed by the processing device can be completed by the hardware integrated logic circuit in the second processor 910 or the instructions in the form of software. That is, the method steps of the embodiment of the present disclosure can be embodied as being executed by a hardware processor, or being executed by a combination of hardware and software modules in the processor. The software module can be located in a storage medium such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the second memory 920, and the second processor 910 reads the information in the second memory 920 and completes the steps of the above method in combination with its hardware. To avoid repetition, it will not be described in detail here.

[0086] The embodiment of the present disclosure also provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the biological age prediction method as described in any embodiment of the present disclosure is implemented. The biological age prediction method driven by executing executable instructions is basically the same as the biological age prediction method provided in the above embodiment of the present disclosure, and will not be described in detail here.

[0087] In some possible embodiments, various aspects of the biological age prediction method provided by the present disclosure may also be implemented in the form of a program product, which includes a program code. When the program product is run on a computer device, the program code is used to enable the computer device to execute the steps of the biological age prediction method according to various exemplary embodiments of the present disclosure described above in this specification. For example, the computer device may execute the biological age prediction method recorded in the embodiments of the present disclosure.

[0088] The program product may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0089] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In hardware implementations, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0090] It should be noted that the above embodiments or implementations are merely illustrative and not restrictive. Therefore, the present disclosure is not limited to the contents specifically shown and described herein. Various modifications, substitutions or omissions may be made to the forms and details of implementation without departing from the scope of the present disclosure.

Claims

1. A model building method, characterized in that: include: Generate a first data set, the first data set includes gene expression level and methylation level data corresponding to a plurality of genes, select a plurality of candidate genes from the first data set, and there is a significant negative correlation between the gene expression level and the methylation level corresponding to each of the candidate genes; Generate a training data set and a validation data set, wherein the training data set includes methylation level data corresponding to a plurality of first genes, and the validation data set includes methylation level data corresponding to a plurality of second genes; Taking the intersection of the plurality of candidate genes, the plurality of the first genes and the plurality of the second genes to obtain a plurality of intersection genes; The model is constructed using the methylation level data corresponding to the plurality of intersection genes in the training data set and the validation data set as input features.

2. The method according to claim 1, characterized in that The methylation level corresponding to the candidate gene is the average methylation level of multiple methylation sites included in the candidate gene; the methylation level corresponding to the first gene is the average methylation level of multiple methylation sites included in the first gene; the methylation level corresponding to the second gene is the average methylation level of multiple methylation sites included in the second gene.

3. The method according to claim 2, characterized in that The generating of the first data set comprises: Download gene expression level data and methylation data of multiple disease patient samples from the first database, Preprocessing the methylation data of the multiple disease patient samples to obtain methylation level data of multiple methylation sites; The methylation level data of multiple sites belonging to the same gene in the multiple disease patient samples are averaged to obtain the methylation level data corresponding to each gene.

4. The method according to claim 1, characterized in that The selecting a plurality of candidate genes from the first data set comprises: For each gene in the first data set, calculate the Pearson correlation coefficient and the significance P value between the gene expression level and the methylation level corresponding to the gene; The genes whose Pearson correlation coefficient is less than a preset correlation coefficient threshold and whose significance P value is less than a preset significance threshold are selected as the candidate genes.

5. The method according to claim 1, characterized in that: The training objective of the model is set as follows: The actual age of the samples in the training data set is obtained, and the actual age of the samples is converted as follows: When the actual age of the sample is less than or equal to the preset age threshold a, the training target of the model is set to: F(age)=log(age+1)-log(a+1); When the actual age of the sample is greater than the preset age threshold a, the training target of the model is set to: F(age)= (age-a) / (a+1), where a is a real number greater than 15.

6. A model building device, characterized in that: The method comprises a memory; and a processor connected to the memory, wherein the memory is used to store instructions, and the processor is configured to execute the steps of the model building method according to any one of claims 1 to 5 based on the instructions stored in the memory.

7. A biological age prediction method, characterized in that: Acquire a DNA methylation aging clock model, wherein the DNA methylation aging clock model is constructed by the model construction method according to any one of claims 1 to 5; Acquiring methylation data of the sample to be tested, and obtaining the methylation level of the sample to be tested on the intersection gene according to the methylation data of the sample to be tested; The methylation level of the sample to be tested on the intersection gene is input into the DNA methylation aging clock model to obtain the biological predicted age of the sample to be tested.

8. A biological age prediction device, characterized in that: comprising a memory; and a processor connected to the memory, the memory being used to store instructions, the processor being configured to execute the steps of the biological age prediction method as claimed in claim 7 based on the instructions stored in the memory.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, which, when executed by a processor, implements the model building method as described in any one of claims 1 to 5, or implements the biological age prediction method as described in claim 7.

10. A computer program product, characterized in that The computer program product comprises instructions which, when executed by a computer, execute the model building method according to any one of claims 1 to 5, or the biological age prediction method according to claim 7.

Images