Method for generating biological mathematical models and biological simulation method

A biomathematical model using in vivo and in vitro gene expression data and signal transduction genes FMOD and THBS1 accurately predicts skin aging and enhances drug screening by simulating gene expression changes, addressing the limitations of previous methods.

JP7595868B2Active Publication Date: 2024-12-09ROHTO PHARM CO LTD +1
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
JP2020152856
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Filing Date
2020-09-11
Publication Date
2024-12-09
Estimated Expiration
2040-09-11

AI Technical Summary

Technical Problem

Existing methods for identifying gene groups related to skin aging are inaccurate due to the lack of temporal gene expression data and the absence of a biomathematical model based on signal transduction, leading to suboptimal screening of plant extracts and drugs for anti-aging effects.

Method used

A method for generating a biomathematical model that incorporates in vivo and in vitro gene expression data, utilizing genes related to signal transduction, particularly FMOD and THBS1, to create a model that accurately reflects aging processes, allowing for the construction of a mathematical model using ordinary differential equations (ODEs) and verification through simulation.

Benefits of technology

The biomathematical model achieves higher accuracy in predicting skin aging and enables precise drug and external factor screening by simulating gene expression changes over time, facilitating targeted interventions to slow down aging.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007595868000005
    Figure 0007595868000005
  • Figure 0007595868000006
    Figure 0007595868000006
  • Figure 0007595868000007
    Figure 0007595868000007
Patent Text Reader

Abstract

To provide a method for generating a bio-mathematical model with which it is possible to verify a bio-mathematical model itself and, hence, to obtain a bio-mathematical model with higher accuracy that is based on signal transduction regarding aging.SOLUTION: Provided is the method for generating a bio-mathematical model that pertains to aging. The method includes: a step for selecting a first gene set which is common to a first gene group obtained from first gene expression data in "in vivo" of a population that includes different ages and the expression level of which changes with aging, and a second gene group obtained from second gene expression data in "in vitro" obtained by successive culture of cell strains and the expression level of which changes with population doubling number; a step for selecting a second gene set from the first gene set, on the basis of information that includes a correlation between the expression level and aging or population doubling number of gene expression and the gene expression level in the first or second gene expression data; and a step for generating a bio-mathematical model that includes the expression level of the gene included in the second gene set as a variable.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a method for generating a biomathematical model relating to aging, and a biological simulation method and a screening method using the biomathematical model. [Background technology]

[0002] The general concept that potential drug targets could be identified by mapping connections in databases of gene expression profiles for drug-treated cells was initiated in 2000 with the publication of a seminal paper by TR Hughes et al. (Non-Patent Document 1), followed shortly thereafter by the initiation of the Connectivity map (C-map) (Map Project by Justin Lamb and researchers at MIT) (Non-Patent Document 2).

[0003] In 2006, Lamb's group began publishing a detailed overview of the mechanistic structure of the C-map, the introduction of the reference collection of gene expression profiles used to create the first generation C-map, and the initiation of the currently ongoing large-scale community C-map project, which is available at the "supporting materials" hyperlink at http: / / www.sciencemag.org / content / 313 / 5795 / 1929 / suppl / DC1.

[0004] The invention described in Patent Document 1 is based on the above concept and attempts to create a relationship map useful for selecting cosmetic active agents, etc., by selecting cell types and generating a reference collection of gene expression profiles for known cosmetic active agents, in order to provide a method useful for generating new active agents for the treatment of aged skin.

[0005] Specifically, a method for constructing a data architecture is disclosed in which a group of related genes is identified from changes in the gene expression profile of dermal fibroblasts due to the presence or absence of exposure to a perturbation factor (such as an active agent), a group of related genes is identified from changes in the gene expression profile of keratinocyte cells due to the presence or absence of exposure to a perturbation factor, and a gene expression signature of a fibroblast instance is obtained by comparing the related genes of both. In addition, a method is disclosed in which, when using this to identify an active agent effective for treating aged skin, an aging gene expression signature is obtained from a comparison of gene expression profiles of elderly and young people, and the aging gene expression signature is compared with the fibroblast instance to identify an effective active agent. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2016-105296 A [Non-patent literature]

[0007] [Non-Patent Document 1] "Functional discovery via a compendium of expression profiles" Cell 102,109-126(2000) [Non-Patent Document 2] “Connectivity Map: Gene Expression Signatures to Connect Small Molecules, Genes, and Disease”, Science, Vol 313,2006 Summary of the Invention [Problem to be solved by the invention]

[0008] However, the method disclosed in Patent Document 1 uses only in vivo gene expression data when identifying a group of genes related to skin aging, and does not obtain gene expression data over time associated with aging. In addition, it has only been found that the gene expression profile of dermal fibroblasts and the like is useful for screening plant extracts, and a biological mathematical model based on signal transduction related to aging has not yet been constructed. For this reason, the plant extract is only screened by comparing the gene expression signature obtained from the change in the gene expression profile of dermal fibroblasts and the like when exposed to the plant extract with the aging gene expression signature, and it is difficult to say that high-precision screening has been achieved.

[0009] Therefore, an object of the present invention is to provide a method for generating a bio-mathematical model, which can obtain a bio-mathematical model based on aging-related signal transduction with higher accuracy.

[0010] Another object of the present invention is to provide a biological simulation method capable of calculating the degree of aging of an individual's skin with higher accuracy by utilizing a biological mathematical model.

[0011] A further object of the present invention is to provide a screening method capable of screening drugs and the like with higher accuracy by utilizing a biological mathematical model. [Means for solving the problem]

[0012] The above object can be achieved by the present invention, which has the following advantageous effects. That is, the present invention includes the following aspects.

[0013] [1] A method for generating a biomathematical model relating to aging, comprising: selecting a first gene set common to a first group of genes whose expression levels change with age, obtained from in vivo first gene expression data of populations including different ages, and a second group of genes whose expression levels change with population doubling number, obtained from in vitro second gene expression data obtained by subculture of a cell line; selecting a second gene set from the first gene set based on information including an expression level of a gene in the first gene expression data or the second gene expression data and a correlation between aging or a population doubling number and the gene expression level; generating a biological mathematical model including, as variables, expression levels of genes included in the second gene set; A method for generating a biological mathematical model comprising:

[0014] According to this embodiment, in the step of selecting the first gene set, the second gene expression data in vitro is used, so that the time-dependent change in the expression level of a specific gene calculated using the generated biomathematical model can be compared with the change in the expression level of a specific gene in vitro obtained by subculturing a cell line due to population doubling number, thereby verifying the generated biomathematical model, so that a biomathematical model based on signal transduction related to aging can be obtained with higher accuracy. At that time, the second gene set is selected from the first gene set based on information including the expression level of gene expression and the correlation between aging or population doubling number and the gene expression level, so that the experimental accuracy in the verification is higher, and since the time-dependent changes can be compared, the fitting accuracy over time can also be higher.

[0015] [2] The method for generating a biomathematical model according to [1], wherein in the step of selecting the first gene set, genes related to signal transduction involved in fibrosis are selected.

[0016] According to this embodiment, it is possible to select, with a higher probability, genes related to signal transduction involved in skin aging in particular, including genes having a high correlation between the expression level of gene expression and the correlation between aging or population doubling number and the gene expression level.

[0017] [3] A method for generating a biomathematical model according to [1] or [2], wherein in the step of selecting the second gene set, the second gene set includes FMOD and THBS1.

[0018] According to this embodiment, when generating a biological mathematical model including the expression levels of those genes as variables, a biological mathematical model based on signal transduction related to aging can be obtained with higher accuracy. That is, both FMOD and THBS1 have a high correlation with the expression levels of gene expression, and with aging or population doubling number and the gene expression levels, and since it has been reported that fibrosis is induced by a signal transduction pathway, it is easy to explain it as a model of time change. And, as shown in Figures 8 to 9, simulations using a biological mathematical model including the expression levels of FMOD and THBS1 as variables can reproduce in vitro experimental data very well.

[0019] [4] A method for generating a biological mathematical model according to any one of [1] to [3], wherein the step of generating the biological mathematical model includes a step of constructing a mathematical model using ordinary differential equations (ODEs) according to an assumed signal transduction pathway.

[0020] According to this embodiment, modeling is performed assuming a specific signal transduction pathway, thereby reducing the labor of repeated trial and error modeling. In addition, when verifying the biological mathematical model generated by the mathematical model based on ordinary differential equations (ODE), the time-dependent changes by simulation can be compared with the experimental results, so that the accuracy of the mathematical model can be improved and verified.

[0021] [5] The method for generating a biomathematical model described in any one of [1] to [4] further comprises a step of verifying the generated biomathematical model by comparing the change over time in the expression level of a specific gene calculated using the generated biomathematical model with the change in the expression level of the specific gene in vitro due to population doubling number obtained by subculturing a cell line.

[0022] According to this embodiment, the generated biological mathematical model can be verified by comparing the changes over time in the simulation with the experimental results, thereby making it possible to further improve the accuracy of the mathematical model.

[0023] [6] A biological simulation method for calculating a degree of skin aging from input information of an individual using a biological mathematical model, comprising: A biological simulation method, wherein the biological mathematical model includes the expression levels of FMOD and THBS1 as variables.

[0024] According to this embodiment, since the biomathematical model includes the expression levels of FMOD and THBS1 as variables, the biomathematical model based on signal transduction related to aging exhibits higher accuracy, and therefore the degree of skin aging can be calculated with high accuracy from an individual's input information.

[0025] [7] A screening method for screening drugs based on the results of calculations made by inputting information about a drug into a biomathematical model that calculates the degree of skin aging, comprising: A screening method, wherein the biomathematical model includes the expression levels of FMOD and THBS1 as variables.

[0026] According to this embodiment, since the biomathematical model includes the expression amount of FMOD and the expression amount of THBS1 as variables, the biomathematical model based on the signal transduction related to aging shows higher accuracy, and therefore, drugs can be screened with higher accuracy from the information on the drug. In addition, more accurate screening based on quantitative results is possible.

[0027] [8] A screening method for screening external factors based on the results of calculations made by inputting information obtained in a test in which external factors are applied into a biomathematical model that calculates the degree of skin aging, comprising: A screening method, wherein the biomathematical model includes the expression levels of FMOD and THBS1 as variables.

[0028] According to this embodiment, since the biomathematical model includes the expression level of FMOD and the expression level of THBS1 as variables, the biomathematical model based on the signal transduction associated with aging shows higher accuracy, and therefore, the external factors can be screened with higher accuracy from the information obtained in the test in which the external factors are applied. In addition, more accurate screening based on quantitative results is possible.

[0029] [9] The screening method described in [7] or [8], wherein the biomathematical model includes a rate equation for each elementary reaction according to a signal transduction model, and includes a step of pre-analyzing which elementary reaction has a greater effect on the expression level of FMOD or the expression level of THBS1.

[0030] According to this embodiment, the elementary reactions that have a greater impact on skin aging are revealed based on the analysis results, making it possible to preliminarily screen drugs and external factors that affect these elementary reactions, thereby enabling more efficient screening of targeted drugs and external factors.

[0031]

[10] A screening method for screening a drug or an external factor using a biomathematical model for calculating a degree of skin aging, comprising: The biomathematical model includes the expression amount of FMOD and the expression amount of THBS1 as variables, and includes a rate equation for each elementary reaction according to a signal transduction model; Analyzing which elementary reaction has a large effect on the expression level of the FMOD or the expression level of the THBS1; A step of selecting a drug or an external factor that has an effect on a large elementary reaction based on the result of the analysis; A screening method comprising:

[0032] According to this embodiment, the analysis step identifies elementary reactions that have a significant effect on skin aging, and the step of selecting drugs or external factors that affect these elementary reactions makes it possible to screen for targeted drugs or external factors. Effect of the Invention

[0033] According to the method for generating a biomathematical model of the present invention, since the biomathematical model itself can be verified, a biomathematical model based on signal transduction related to aging can be obtained with higher accuracy. In addition, according to the biological simulation method of the present invention, the degree of skin aging of an individual can be calculated with higher accuracy by using the biomathematical model. Furthermore, according to the screening method of the present invention, drugs and the like can be screened with higher accuracy by using the biomathematical model. [Brief description of the drawings]

[0034] [Figure 1] FIG. 1 is a schematic diagram showing the flow of large-scale in vivo gene expression analysis. [Diagram 2] FIG. 1 is a schematic diagram showing the flow of in vitro gene expression information analysis. [Diagram 3] This figure shows the workflow for performing a series of analyses using Sequence Read Archives (SRA) files. [Figure 4] This figure shows the results of KEGG analysis using ClusterProfiler. The figure on the left shows the overlapping relationship between genes whose expression is decreased or increased in vivo and in vitro, and the figure on the right shows a mapping of the expression ratio of related genes according to the type of signal transduction pathway. [Diagram 5] FIG. 1 shows changes in gene expression levels with cell passage for a group of fibrosis-related genes. [Figure 6A] The horizontal axis represents the correlation coefficient between the expression of each gene and age, and the vertical axis represents the logarithmic value of TPM. The top 5 genes are color-coded as TRUE, and the rest are color-coded as FALSE. [Figure 6B] FIG. 1 shows the expression distribution of THBS1, FMOD, and DCN, correlation with age, and correlation between the genes. [Figure 7] FIG. 1 shows the characteristics of the new model for TGFβ signaling in comparison with existing models. [Figure 8] 1 is a graph showing the simulation results of the expression levels of THBS1 in Example 1 and Comparative Examples 1 to 12 together with experimental data. [Figure 9] 1 is a graph showing the simulation results of the expression levels of CDKN2B in Example 2 and Comparative Examples 13 to 24 together with the experimental data.

[0035] [Figure 10] FIG. 1 shows an overview of an integrated biological model for TGFβ signaling and Wnt.

[0036] [Figure 11] 1 is a graph showing the simulation results of the expression levels of THBS1, CDKN2B, FMOD and CCND1 in Examples 3, 4, 5 and 6, together with experimental data.

[0037] [Figure 12] FIG. 1 illustrates a workflow for calculating the degree of skin aging of an individual.

[0038] [Figure 13] 1 is a graph showing the results of calculating the degree of skin aging of an individual.

[0039] [Figure 14] FIG. 1 shows the results of sensitivity analysis on the gene expression level of THBS1.

[0040] [Figure 15] FIG. 1 shows the results of sensitivity analysis of gene expression levels of FMOD.

[0041] [Figure 16] 1 is a graph showing the results of a simulation of the effect of drugs on the gene expression level of THBS1.

[0042] [Figure 17]1 is a graph showing the results of a simulation of the effect of drugs on the gene expression level of FMOD.

[0043] [Figure 18] 1 is a graph showing the results of a simulation of the effect of drug combinations on the gene expression level of THBS1. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0044] Hereinafter, an embodiment of the present invention will be described in detail.

[0045] [Terminology definition] The terms used in this specification are defined as follows.

[0046] A "mathematical model" refers to a mathematical description, such as a differential equation, that mimics the movement of measurable indicators of a time-varying phenomenon, and a "biomathematical model" refers to a mathematical model of a phenomenon that occurs in a living organism.

[0047] When we simply say "model," we mean the conceptual description that is the premise of the mathematical description, and a "signal model" refers to a model regarding signal transmission.

[0048] "In vivo" refers to an environment inside a living body or an environment similar to that inside a living body, and "in vitro" refers to an artificial environment outside a living body.

[0049] [How to generate a biological mathematical model] The method for generating a biomathematical model of the present invention is a method for generating a biomathematical model related to aging, which includes natural aging due to aging as well as aging due to external factors such as ultraviolet rays, gamma rays, oxygen concentration, chemical substances, drugs, heat, osmotic stress, pH, microorganisms, viruses, and small interfering RNA.

[0050] The biomathematical model includes all mathematical descriptions that mimic the behavior of measurable indicators of time-varying phenomena in a living organism, such as a signal transduction pathway involving gene expression in a living organism.

[0051] Examples of the site of the living body include the skin, eye, lung, liver, kidney, heart, intestine, adipose tissue, skeletal muscle, uterus, etc. However, the skin, eye, lung, kidney, heart, etc. are preferred from the viewpoints of easy availability of cell lines, reproducibility, and social necessity.

[0052] Mathematical descriptions include ordinary differential equations (ODE), data-driven models, Bayesian estimation, Boolean networks, modular response analysis, etc. However, ordinary differential equations (ODE) are preferably used because they allow comparison of changes over time when verifying the generated biomathematical model and allow creation of a mathematical model based on biosignals already reported from molecular biological knowledge.

[0053] The method for generating a biological mathematical model of the present invention includes a step of selecting a first gene set, a step of selecting a second gene set from the first gene set, and a step of generating a biological mathematical model. Each step will be described below in order.

[0054] [Step of selecting the first gene set] This step is a step of selecting a first gene set common to a first group of genes whose expression levels change with age obtained from in vivo first gene expression data of a population including different ages, and a second group of genes whose expression levels change with population doubling number obtained from in vitro second gene expression data obtained by subculturing a cell line. As for the first group of genes and the second group of genes, either genes whose expression levels increase or decrease may be selected as genes whose expression levels change, but since many of the genes whose expression levels decrease are genes related to the arrest of the cell cycle, which is a general characteristic of aging, it is preferable to select genes whose expression levels increase, which allows for more confirmation of the characteristics caused by aging.

[0055] Although it is possible to newly obtain experimental data as the first gene expression data, it is preferable to use open data. Examples of such open data include those described in Reference 1 (Genome Biol. 2018 Dec 20;19(1):221.(GSE113957)).

[0056] In Reference 1, gene expression was comprehensively analyzed by RNA-Seq analysis for human skin-derived fibroblasts obtained as biopsies from 85 healthy Caucasian men and women in their teens to 90s, as shown in Figure 1, and the gene expression data is published. In addition, gene expression data can be obtained from Gene Expression Omnibus (GEO), a public database operated by the National Center for Biotechnology Information (NCBI).

[0057] The first group of genes whose expression levels change with age is obtained from the first gene expression data, and the first group of genes whose expression levels change with age can be selected through the workflow shown in Fig. 3, for example, as in Test Example 1. The workflow shown in Fig. 3 includes file conversion, filtering processing, mapping processing, data counting processing, normalization conversion processing, robust principal component analysis, hierarchical clustering processing, expression variation analysis processing, and the like.

[0058] Although it is possible to obtain new experimental data as the second gene expression data, it is preferable to use open data. Examples of such open data include those described in Reference 2 (PLoS One. 2016 May 3;11(5):e0154531.(GSE63577)).

[0059] Reference 2 uses human neonatal foreskin fibroblasts (HFF cells) established by the American Type Culture Collection (ATCC), which are subcultured to induce cellular senescence, as shown in Figure 2. The progression of senescence is controlled by the population doubling (PD) level, which is an index of the doubling of the number of cells in culture, and gene expression data is published that comprehensively analyzed gene expression by RNA-Seq analysis. This gene expression data is available from GEO, a public database operated by NCBI.

[0060] The second group of genes whose expression levels change with the population doubling number is obtained from the second gene expression data, and the second group of genes whose expression levels change with the population doubling number can be selected through the workflow shown in Fig. 3, for example, as in Test Example 1. The workflow shown in Fig. 3 includes file conversion, filtering processing, mapping processing, data counting processing, normalization conversion processing, expression variation analysis processing, and the like.

[0061] When selecting the first gene set common to the first gene group and the second gene group, it is possible to select all common genes, but it is preferable to select genes related to specific signal transduction. For example, by selecting genes related to signal transduction involved in fibrosis, it is possible to select genes related to signal transduction particularly involved in skin aging, which include genes with high correlation between gene expression level and aging or population doubling number and gene expression level, with a high probability.

[0062] [Step of selecting the second gene set] This step is a step of selecting a second gene set from the first gene set based on information including gene expression levels in the first gene expression data or the second gene expression data, and correlations between aging or population doubling number and gene expression levels. When the first gene expression data is used, information including gene expression levels and correlations between aging and gene expression levels can be used, and when the second gene expression data is used, information including gene expression levels and correlations between population doubling number and gene expression levels can be used.

[0063] For example, when using the first gene expression data, as shown in Table 1 of Test Example 2, for the signal transduction-related genes involved in fibrosis (16 genes), genes with high correlation with age and Transcripts Per Million (hereinafter referred to as TPM) values ​​in the in vivo gene expression data can be ranked, and the genes ranked at the top can be selected as the second gene set.

[0064] When the second gene expression data is used, the genes having a high correlation with the population doubling number and the high TPM value can be ranked for the in vitro second gene expression data obtained by subculturing the cell line, and the genes ranked at the top can be selected as the second gene set. Of course, it is also possible to perform a comprehensive ranking using the first gene expression data and the second gene expression data, and select the genes ranked at the top as the second gene set.

[0065] When selecting the second gene set from the first gene set, it is preferable to select genes for which molecular biological knowledge regarding their relationship with signal transduction involved in aging has been obtained to a certain extent, which makes it easier to model a specific signal transduction pathway and reduces the effort required for repeated trial and error modeling.

[0066] For example, in the step of selecting the second gene set, if the second gene set includes FMOD and THBS1, it is easy to model the signaling pathway because multiple findings related to fibrosis have been reported for these. Furthermore, by modeling TGFβ signaling, as shown in Figures 8 to 9, simulations using the generated biomathematical model can reproduce in vitro experimental data very well.

[0067] Regarding DCN (Decolin), it shows the highest ranking in Table 1, is the same proteoglycan as FMOD, and has been reported in reference 7 to inhibit the binding of TGFβ to its receptor TGFR, just like FMOD. Therefore, it can also be selected in the second gene set as a substitute for FMOD.

[0068] [Process of generating a biomathematical model] This step is a step of generating a biological mathematical model including the expression levels of the genes included in the second gene set as variables. Although the specific method is not particularly limited, the general procedure is to first determine a predicted signal transduction pathway, create a signal model corresponding to the signal transduction pathway, and then construct a mathematical model based on the signal model.

[0069] To determine the expected signal transduction pathway, the signal transduction pathway involving the genes included in the selected second gene set may be searched for using literature information, public databases, etc., and the signal transduction pathway may be determined from among them. In this case, when the multiple genes included in the second gene set have mutual correlation in gene expression level, it is preferable to select the signal transduction pathway involving the multiple genes. In Test Example 3, the TGFβ signal transduction pathway was determined, but it is also possible to select the Wnt signal transduction pathway, the NF-κB signal transduction pathway, etc., which have been reported for fibroblast fibrosis.

[0070] In some cases, the determined signaling pathway can be used as a signaling model as is, but in order to speed up the calculation time of the mathematical model, a more simplified signaling model can also be created. In Test Example 3, the existing signaling model (left side of Figure 7) was modified to create a new signaling model as shown in the right side of Figure 7. It is also considered desirable to create a signaling model by integrating Wnt signaling and TGFβ signaling, and a simulation was performed using the integrated model in Test Example 4.

[0071] The construction of a mathematical model based on a signal model can be carried out, for example, according to the method described in reference 8 (Lucarelli et al., 2018, Cell Systems 6, 75-89).

[0072] In Test Example 3, a mathematical model is constructed using the rate equations of each elementary reaction, the initial values ​​of the concentrations of each substance (Species), the rate equations of the mass balance (ordinary differential equations), and the parameters that are the coefficients of these equations, according to a signal model created from the TGFβ signal transduction pathway. As a solution to this ordinary differential equation, a simulation result such as that shown in FIG. 8 or FIG. 9 can be obtained. This simulation was performed using BioMASS (software provided free of charge by https: / / github.com / okadalabipr / biomass), but it is also possible to perform the simulation using various other simulation software (e.g., COPASI, BioNetGen, Matlab Simbiology, etc.). However, BioMASS (Biological Modeling and Analysis of Signaling Systems) has the following characteristics, and can therefore be used preferably. The environment is easy to set up, making it easy to use even for biology researchers who are not familiar with programming. - It is easier to share information, execute, and find errors with more people, and computing resources can be used more effectively. Suitable for building medium to large scale models.

[0073] [Verifying the generated biomathematical model] In the present invention, it is preferable that the method further comprises a step of verifying the generated biomathematical model by comparing the time-dependent change in the expression level of a specific gene calculated using the generated biomathematical model with the in vitro change due to population doubling number in the expression level of the specific gene obtained by subculturing a cell line.

[0074] The time course of the expression level can be obtained by calculating (simulating) the expression level of a specific gene over time using the generated biological mathematical model. The change in expression level due to the population doubling number can be obtained by investigating the expression level of a specific gene according to the population doubling number in vitro by subculturing a cell line, although it is also possible to use experimental values ​​included in open data.

[0075] In Test Example 3, the simulation results are verified using experimental values ​​as shown in Figures 8 and 9. If the verification results show insufficient consistency with the experimental values, the accuracy of the biomathematical model can be improved by improving the signal model or changing the initial values ​​of the concentrations of each substance (species).

[0076] [Applications of biomathematical models] A biomathematical model of aging can be used to quantify and visualize the degree of aging. For example, the degree of aging in the living body of an individual or group can be simulated and the result can be compiled into a database. In the case of aging, for example, in the case of skin aging, the relationship between wrinkles, sagging, elasticity, dullness, rough skin, loss of transparency, etc. can be clarified by simulation. The same can be said about the relationship between aging and glycation at the cellular level. In addition, in the case of eye aging, the relationship between presbyopia, cataracts, age-related macular degeneration, glaucoma, dry eye, etc. can be clarified by simulation.

[0077] It can also be used to screen external factors related to aging, and to screen drugs that suppress the progression of aging. It can also be used to select drugs that suppress the action of external factors related to aging. It can also be used to confirm how the variables and coefficients themselves included in the biomathematical model affect the degree of aging, and to search for factors that control the variables and coefficients.

[0078] The above uses will make it possible to suppress (prevent) the progression of aging, as well as predict the future progression of aging and healthy life expectancy.

[0079] [Bio-simulation method] The biological simulation method of the present invention is a biological simulation method for calculating the degree of skin aging from individual input information using a biological mathematical model, the biological mathematical model including the expression amount of FMOD and the expression amount of THBS1 as variables. As the biological mathematical model, for example, the biological mathematical model generated by the present invention described above can be used.

[0080] Examples of input information for individuals include gene expression levels and concentrations of signal transduction substances related to the signal transduction pathways assumed by the biomathematical model.

[0081] The degree of skin aging can be calculated (quantified, etc.) by simulating variables that change with aging in the signal transduction pathway assumed by the biomathematical model. In Test Example 5, a biosimulation was performed using the biomathematical model to calculate the degree of skin aging from individual input information, and the results shown in FIG. 13 were obtained.

[0082] [Drug screening methods] The drug screening method of the present invention is a screening method for screening drugs based on the results of calculations made by inputting drug information into a biomathematical model that calculates the degree of skin aging, and the biomathematical model includes the expression levels of FMOD and THBS1 as variables. As the biomathematical model, for example, the biomathematical model generated by the present invention described above can be used.

[0083] The information on drugs includes information obtained from drug administration tests, database information showing the association with drugs, information obtained by simulating reactivity with drugs, etc. Drugs for suppressing aging can be screened in silico by using database information showing the association with drugs.

[0084] The information obtained from drug administration tests includes information related to the signal transduction pathways assumed by the biomathematical model, such as gene expression levels and signal transduction substance concentrations, which change depending on the presence or absence of a drug and its administration concentration.

[0085] Examples of database information showing associations with drugs include information showing the effects of drugs on gene expression levels, signal transduction substance concentrations, and the like, which are related to the signal transduction pathways assumed by the biomathematical model.

[0086] Information that can be obtained by simulating reactivity with drugs includes the results of simulating the reactivity of drugs with gene transcription products, signal transduction substances, etc. that are related to the signal transduction pathways assumed by the biomathematical model.

[0087] The results of calculations using biomathematical models include simulations of variables that change with aging, and based on these results, drugs that suppress the progression of aging can be screened. In addition, screening using models allows quantitative evaluation of not only one type of drug, but also combinations of two or more types of drugs.

[0088] [Screening methods for external factors] The method for screening for external factors of the present invention is a method for screening for external factors based on the results of calculations made by inputting information obtained in a test in which an external factor is applied into a biomathematical model for calculating the degree of skin aging, and the biomathematical model includes the expression levels of FMOD and THBS1 as variables. For example, the biomathematical model generated by the present invention described above can be used as the biomathematical model.

[0089] Examples of external factors that are conventionally known to be involved in aging include ultraviolet rays, gamma rays, oxygen concentration, chemical substances, drugs, heat, osmotic stress, pH, microorganisms, viruses, small interfering RNA, etc. The present invention enables more accurate screening of these external factors based on quantitative results, which is significant in that more quantitative screening can be performed.

[0090] It is also possible to screen for various external factors that have not been previously known.

[0091] Results calculated using biomathematical models include simulations of variables that change with aging, and based on these results, it is possible to screen for external factors that inhibit the progression of aging.

[0092] [Analysis process] In the above-mentioned drug screening method or external factor screening method, the biomathematical model preferably includes a rate equation for each elementary reaction according to a signal transduction model, and includes a step of analyzing in advance which elementary reaction has a large effect on the expression level of FMOD or the expression level of THBS1. By performing such an analysis in advance, the elementary reaction that has a large effect on skin aging is clarified based on the analysis results, and it becomes possible to preliminarily screen drugs or external factors that affect this elementary reaction. As a result, it is possible to more efficiently screen targeted drugs or external factors.

[0093] For example, in Test Example 6, the signal transduction model shown in FIG. 10 is assumed, and the biomathematical model includes the rate equations v[1] to v

[49] , which are the rate equations of each elementary reaction (numbers 1 to 49 in FIG. 10) according to the signal transduction model. Then, which elementary reaction has a large effect on the expression level of FMOD or the expression level of THBS1 is analyzed (so-called sensitivity analysis) based on the magnitude of change in the expression level when the parameters of the rate equations v[1] to v

[49] are changed. Such sensitivity analysis can also be performed using BioMASS. As a result, reactions 22, 32, 35, and 12 are identified as those that suppress the expression of THBS1 and promote the expression of FMOD.

[0094] In Test Example 6, the parameters included in the rate equation for each elementary reaction were changed to perform the sensitivity analysis. Alternatively, a similar sensitivity analysis can be performed by adding coefficients to the rate equations for v[1] to v

[49] in the material balance rate equation and changing these coefficients to values ​​other than 1.

[0095] Then, when the elementary reactions that have a large effect on the expression level of FMOD or the expression level of THBS1 are clarified by sensitivity analysis, it is possible to select a drug that affects the elementary reaction to some extent from the information on the drug. Then, by determining the parameters in the biomathematical model for each drug, it is possible to simulate the effect on aging using the biomathematical model, so that a more desirable drug can be screened from among several candidate drugs. When screening for external factors that affect aging, screening can be done in the same way as for drugs.

[0096] [Analysis-based screening method] The screening method using the above analysis includes a screening method for screening drugs or external factors using a biomathematical model that calculates the degree of skin aging, the biomathematical model includes the expression level of FMOD and the expression level of THBS1 as variables, and includes a rate equation for each elementary reaction according to a signal transduction model, and includes a step of analyzing which elementary reaction has a large effect on the expression level of FMOD or the expression level of THBS1, and a step of selecting a drug or external factor that affects the elementary reaction with a large effect based on the results of the analysis. Note that, as the biomathematical model, for example, the biomathematical model generated by the present invention described above can be used.

[0097] That is, in the screening method described above, drugs and the like are screened based on the results of calculations made by inputting drug-related information and the like into a biomathematical model that calculates the degree of skin aging, but a method of selecting drugs or external factors that affect elementary reactions with large influences based on analysis results without inputting drug-related information and obtaining calculation results is also possible. When screening for more desirable drugs from the candidate drugs selected in this way, drugs and the like can be screened using a normal evaluation system for skin aging. EXAMPLES

[0098] More specific test examples of the present invention will be shown below, but the present invention is not limited to these specific examples.

[0099] Test Example 1. (Selection of aging-related genes (first gene set)) By analyzing large-scale gene expression information from clinically obtained human dermal fibroblasts classified by age (in vivo, see Figure 1) and human dermal fibroblasts established as cell lines (in vitro, see Figure 2), we succeeded in identifying important genetic groups that correlate with human age. Specifically, the following expression data were obtained from public databases and analyzed.

[0100] The in vivo analysis was performed using data (first gene expression data) from reference 1 (Genome Biol. 2018 Dec 20;19(1):221.(GSE113957)). The human skin-derived fibroblasts used to obtain the in vivo data were obtained as biopsies from 85 healthy Caucasian men and women in their teens to 90s, as shown in Figure 1. Gene expression data obtained by comprehensively analyzing gene expression using these fibroblasts was used.

[0101] The in vitro analysis was performed using data from Reference 2 (PLoS One. 2016 May 3;11(5):e0154531.(GSE63577)). The HFF cells used to obtain the in vitro data are human neonatal foreskin-derived fibroblasts established by the American Type Culture Collection (ATCC). As shown in Figure 2, HFF cells were subcultured to induce cellular senescence, and the progression of senescence was measured using data managed by the population doubling (PD) level, which is an index of the doubling of the number of cells in culture.

[0102] Each data set was obtained from the Gene Expression Omnibus (GEO), a public database operated by the National Center for Biotechnology Information (NCBI), following the workflow in Figure 3, and then groups of genes with common expression changes were obtained by analysis.

[0103] Specifically, as shown in Figure 3, we obtained Sequence Read Archives (SRA) files from the GEO database, and then used sra-tools (version: 2.8.0) (http: / / ncbi.github.io / sra-tools / ) to convert the SRA files to FASTQ files. The obtained FASTQ files were then used trim-galore (version: 0.6.5) (https: / / github.com / FelixKrueger / TrimGalore) to remove unnecessary adapter sequences. Next, the genome mapping tool HISAT2 (version: 2.1.0) (reference: Nat Methods. 2015 Apr;12(4):357-60.) was used to map the reads to the reference genome (version: GRCh38.p13) and obtain SAM format data. The SAM format was converted to BAM format using SAMtools (version: 1.9) (Reference: Bioinformatics. 2009 Aug 15;25(16):2078-9.), and then the BAM format file representing the mapping results was used as input using subread (version: 2.0.0) (Reference: Nucleic Acids Res. 2013 May 1;41(10):e108.) to obtain count data of the number of reads for each gene region.

[0104] The obtained count data was normalized to TPM, and then low-expressing genes with an average TPM value of 5 or less were removed.

[0105] For in vivo data, data from the 10th to 70th decade were used in the expression change analysis to extract genes related to relatively early aging in the 10th to 70th decade. In addition, to exclude outliers, robust principal component analysis was performed for each generation using rrcov (version: 1.5.2) (Reference: Journal of Statistical Software, 2009, 32(3), 1-47). To take into account inter-individual variability, each sample was classified into three clusters by hierarchical clustering according to the gene expression data in each generation, and gene expression changes were confirmed between each cluster. For in vitro data, gene expression changes between each PD were confirmed.

[0106] Gene expression changes were confirmed using DESeq2 (version: 1.26.0) (Reference: Genome Biol. 2014;15(12):550.). Specifically, the count data before normalization, in which low-expressing genes were removed according to the TPM value, was used as input to confirm differentially expressed genes. KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis was performed on the common gene groups in the genes whose expression changes were changed in vivo and in vitro using ClusterProfiler (version: 3.14.3) (Reference: OMICS. 2012 May;16(5):284-7.). The results are shown in Figure 4.

[0107] As a result, the gene group whose expression was decreased was related to the cell cycle, and the gene group whose expression was increased was related to fibrosis, as shown in Figure 4. Fibrosis is known to be a phenomenon that occurs with aging, and this test identified a particularly important gene group (16 genes).

[0108] Figure 5 shows the change in gene expression levels with cell passage for fibrosis-related genes. This data shows the passage change in the values ​​obtained by adding 1 to the normalized TPM value of the in vitro data, logarithmizing it into Log2 and then converting it into Zscore. As shown in Figure 5, it can be confirmed that the expression levels of the fibrosis-related genes increase with cell passage.

[0109] Test Example 2. (Selection of aging-related genes (second gene set)) Next, based on the expression levels of genes in vivo and the correlation between age and gene expression levels, we found that FMOD (Fibromodulin) and THBS1 (Thrombospondin-1) are important factors in controlling aging.

[0110] Specifically, the correlation coefficient between each gene expression and age was calculated using Spearman's rank correlation coefficient, and ranking was performed according to the correlation coefficient. Next, ranking was performed according to the average TPM calculated previously. To visualize these, a scatter plot was created with the correlation coefficient with age on the horizontal axis and the logarithm of TPM on the vertical axis. The results are shown in Figure 6A.

[0111] Among the genes in FIG. 6A, THBS1 and FMOD were ranked highly, and since multiple findings related to fibrosis had been reported (list of papers below), both genes were selected as important genes.

[0112] THBS1 literature Document 3: Matrix Biol. 2018 Aug;68-69:28-43 Document 4: Nat Commun. 2019 Mar 8;10(1):1146. FMOD Literature Document 5:Front Pharmacol. 2020 Jan 28;10:1649. Document 6:Front Pharmacol. 2019 Nov 25;10:1308. DCN Literature Reference 7:J Histochem Cytochem. 2012 Apr;60(4):262-8. FIG. 6B shows the expression distribution of THBS1, FMOD, and DCN, the correlation with age, and the correlation between each gene.

[0113] As is clear from Figure 6B, the expression of both THBS1 and FMOD genes increases with age, and the correlation coefficient of their expression levels is high (Corr: 0.655), indicating that the two genes are correlated and are important genes that are correlated with aging.

[0114] In literature 3, it has been reported that THBS1 activates TGF-β1, a ligand of the TGF-β signal, which is the most important signal pathway in fibrosis, and in literature 4, it has been reported that the transcription of THBS1 is controlled by the TGF-β signal. In literature 5, it has been reported that FMOD inhibits the binding of TGFβ to its receptor TGFR, and in literature 6, it has been reported that the transcription of FMOD is controlled by the Wnt signal, which is also involved in fibrosis.

[0115] Regarding DCN (Decolin), which showed the highest ranking in Table 1, it is the same proteoglycan as FMOD, and it has been reported in reference 7 that, like FMOD, it inhibits the binding of TGFβ to its receptor TGFR, so it can also be used as a substitute gene for FMOD.

[0116] Test Example 3. (Construction of a mathematical model of skin aging) A literature survey of the aging-related genes FMOD and THBS1 in relation to fibroblast fibrosis revealed that both genes are involved in the TGFβ signaling pathway, the Wnt signaling pathway, and the NF-κB signaling pathway (hereafter, this series of signaling pathways will be abbreviated as aging signals). In addition, published literature has revealed that in the aging signaling pathway, FMOD suppresses signals, i.e., acts as a positive parameter for aging, while THBS1 promotes signals, i.e., acts as a negative parameter for aging. Therefore, we constructed a mathematical model of skin aging using FMOD and THBS1 as factors. Since it has been reported that TGFβ signaling is a core mechanism in fibrosis in the series of aging signals (Reference 3), we first constructed a mathematical model of TGFβ signaling.

[0117] Based on an existing mathematical model of TGFβ signaling (Reference 8; Lucarelli et al., 2018, Cell Systems 6, 75-89), we constructed a mathematical model using ordinary differential equations (ODEs) with FMOD and THBS1 as inputs (details will be described later). Figure 7 shows the characteristics of the new signaling model for TGFβ signaling compared with the existing model.

[0118] Compared with existing signaling models, the new signaling model is unique in that it reproduces the promotion and inactivation of TGFβ signaling and shows positive feedback as the signal progresses.

[0119] Specifically, TGFβ1 is known to be translated in an inactivated state (hereafter referred to as inact TGFβ1), while THBS1 is a factor responsible for activation, and is known to change TGFβ1 to an activated state (hereafter referred to as act TGFβ1) (Reference 3). On the other hand, FMOD is known to incorporate act TGFβ1 into its structure, forming a complex and inhibiting TGFβ1 binding to the receptor (Reference 5). The transcription of THBS1 itself is controlled by the TGFβ signal (Reference 4), and the expression level of THBS1 increases as the signal progresses. This is a signal model in which the TGFβ signal itself is activated by the increase in THBS1, and the signal progresses further. Parameter estimation and simulation in this signal model were performed using BioMASS (https: / / github.com / okadalabipr / biomass).

[0120] In Example 1, a simulation of the expression level of THBS1 was performed using parameter values ​​calculated by parameter estimation, and in Example 2, a simulation of the expression level of CDKN2B was performed using parameter values ​​calculated by parameter estimation. These data are shown in Figures 8 to 9 together with the experimental values ​​of THBS1 and CDKN2B in vitro.

[0121] For the comparative example, the following gene parameter set shown in Reference 8 was used. These genes are representative genes of each cluster when expression data of microarray of mouse hepatocellular carcinoma Hepa1-6 cells was clustered by k-means method. These results are also shown in Figures 8 to 9. Comparative examples 1 and 13: Ski Comparative Examples 2 and 14: Skil Comparative Examples 3 and 15: Dnmt3a Comparative Example 4, 16: Sox4 Comparative example 5, 17:Jun Comparative Examples 6 and 18: Smad7 Comparative Examples 7 and 19: Klf10 Comparative Examples 8 and 20: Bmp4 Comparative Examples 9, 21: Cxcl15 Comparative Examples 10 and 22: Dusp5 Comparative Examples 11 and 23: Tgfa Comparative Examples 12 and 24: Pdk4 The triangle values ​​in Figures 8 and 9 indicate experimental data (mean ± SD, n = 3), and the solid lines indicate the simulation results when each parameter was used. The horizontal axis indicates the degree of cell senescence by PD, and the vertical axis indicates the gene expression level normalized by TPM, and then added 1 to each TPM to convert it to logarithm, and then converted to Log2.

[0122] The experimental values ​​in Figures 8 to 9 were obtained as follows. The experimental values ​​were obtained from in vitro analysis results, i.e., data from literature 2 (PLoS One. 2016 May 3;11(5):e0154531.(GSE63577)). The analysis results are shown in Table 1 below.

[0123] [Table 1]

[0124] Regarding the expression level of THBS1, as shown in Figure 8, the original parameters published in the paper were unable to reproduce the THBS1 expression data (Comparative Examples 1 to 12). However, by constructing parameters based on a new signal model (Example 1), we succeeded in constructing a mathematical model that reproduced the experimental data.

[0125] Furthermore, with regard to CDKN2B, whose transcription is controlled by TGF-β signaling, as shown in Figure 9, the comparative examples (Comparative Examples 13 to 24) were unable to reproduce the experimental data, whereas Example 2 reproduced the in vitro experimental data very well.

[0126] The details of the ODE model (biomathematical model) used in the simulation are shown below. The ODE model is composed of the rate equations of each elementary reaction according to the signal transduction model, the initial values ​​of the concentrations of each substance (species), the rate equations of the mass balance (ordinary differential equations), and the parameters that serve as the coefficients of these equations. As the solution to this ordinary differential equation, the simulation results shown in Figure 8 or Figure 9 can be obtained. BioMASS (https: / / github.com / okadalabipr / biomass) was used for the simulation.

[0127] First, the following values ​​were used as parameters. These parameters were estimated using BioMASS by the least squares method between the experimental and simulation values ​​so that they were closest to each other. The number after "e" means a multiplier of 10 and indicates the product with the number before "e".

[0128] (Common parameters) Rec_act=6.00e-3 S2tot=1.42e2 S3tot=1.42e1 S4tot=6.94e1 S_dephos=2.89e-1 S_dephosphos=4.95e-2 S_phos=3.80e-1 k_on_223=0.00e0 k_on_224=0.00e0 k_on_233=1.47e-1 k_on_234=4.03e-4 k_on_244=8.63e-6 k_on_334=0.00e0 k_on_344=0.00e0 k_on_222=0.00e0 k_on_333=0.00e0 k_on_444=0.00e0 kdiss_SS=0.00e0 pRec_degind=4.01e-2 kf=1.798e+01 Kmf=2.299e-01 k_on_FMOD=1.402e-04 k_off_FMOD=7.138e-02 (Parameters for Example 1) THBS1_act3=5.027e+00 THBS1_act2=4.942e-02 THBS1_act1=5.353e+02 THBS1_inh3=4.597e-04 THBS1_inh2=1.914e-01 THBS1_inh1=1.554e-01 THBS1_turn=1.415e-01 (Example 2 parameters) CDKN2B_act3=9.789e+00 CDKN2B_act2=4.419e+02 CDKN2B_act1=9.184e-04 CDKN2B_inh3=4.308e-05 CDKN2B_inh2=1.000e+03 CDKN2B_inh1=3.112e-01 CDKN2B_turn=1.156e-04

[0129] The parameters used in the estimation of the comparative example are shown below, where GENE indicates THBS1 or CDKN2B.

[0130] (Comparative Examples 1 and 13: Ski Parameter) GENE_act3=1.41e-2 GENE_act2=8.06e-4 GENE_act1=0.00e0 GENE_inh3=0.00e0 GENE_inh2=4.66e-2 GENE_inh1=2.68e-2 GENE_turn=4.56e-3 (Comparative Examples 2 and 14: Skil parameters) GENE_act3=4.63e-1 GENE_act2=7.45e-2 GENE_act1=3.07e-2 GENE_inh3=0.00e0 GENE_inh2=7.53e-1 GENE_inh1=4.10e-1 GENE_turn=1.12e-2 (Comparative Examples 3 and 15: Dnmt3a parameters) GENE_act3=7.15e-3 GENE_act2=0.00e0 GENE_act1=0.00e0 GENE_inh3=0.00e0 GENE_inh2=1.93e-2 GENE_inh1=3.62e-2 GENE_turn=3.81e-3 (Comparative Examples 4 and 16: Sox4 parameters) GENE_act3=1.12e-2 GENE_act2=2.72e-4 GENE_act1=0.00e0 GENE_inh3=0.00e0 GENE_inh2=8.73e-2 GENE_inh1=1.03e-1 GENE_turn=8.02e-4 (Comparative Examples 5 and 17: Jun parameters) GENE_act3=0.00e0 GENE_act2=1.02e0 GENE_act1=5.99e0 GENE_inh3=8.20e0 GENE_inh2=1.46e0 GENE_inh1=9.76e0 GENE_turn=1.21e-1 (Comparative Examples 6 and 18: Smad7 parameters) GENE_act3=9.98e2 GENE_act2=2.12e-1 GENE_act1=2.09e1 GENE_inh3=3.67e0 GENE_inh2=0.00e0 GENE_inh1=0.00e0 GENE_turn=3.64e1 (Comparative Examples 7 and 19: Klf10 parameters) GENE_act3=1.00e3 GENE_act2=9.05e1 GENE_act1=0.00e0 GENE_inh3=0.00e0 GENE_inh2=1.79e-2 GENE_inh1=7.59e-4 GENE_turn=4.74e2 (Comparative Examples 8 and 20: Bmp4 parameters) GENE_act3=0.00e0 GENE_act2=9.04e1 GENE_act1=1.00e3 GENE_inh3=1.00e3 GENE_inh2=1.94e1 GENE_inh1=2.18e2 GENE_turn=9.91e-1 (Comparative Examples 9 and 21: Cxcl15 parameters) GENE_act3=9.24e0 GENE_act2=0.00e0 GENE_act1=0.00e0 GENE_inh3=0.00e0 GENE_inh2=3.20e2 GENE_inh1=1.00e3 GENE_turn=3.28e-3 (Comparative Examples 10 and 22: Dusp5 parameters) GENE_act3=0.00e0 GENE_act2=0.00e0 GENE_act1=0.00e0 GENE_inh3=4.72e-1 GENE_inh2=1.15e-2 GENE_inh1=1.09e-2 GENE_turn=1.41e-2 (Comparative Examples 11 and 23: Tgfa parameters) GENE_act3=1.28e-2 GENE_act2=2.93e-4 GENE_act1=0.00e0 GENE_inh3=0.00e0 GENE_inh2=1.39e0 GENE_inh1=2.49e0 GENE_turn=3.60e-3 (Comparative Examples 12 and 24: Pdk4 parameters) GENE_act3=0.00e0 GENE_act2=4.86e-4 GENE_act1=0.00e0 GENE_inh3=1.29e0 GENE_inh2=6.43e-2 GENE_inh1=1.40e-1 GENE_turn=1.50e-2

[0131] The following formulas were used as the rate equations for each elementary reaction. Note that "**" is synonymous with "^" and means exponentiation.

[0132] v[1]=[Rec]*Rec_act*[TGFb_act] v[2]=[TGFb_pRec]*pRec_degind v[3]=k_on_222*([ppS2]**3) v[4]=3*S_dephosphos*[ppS2_ppS2_ppS2] v[5]=k_on_333*[ppS3]**3 v[6]=3*S_dephosphos*[ppS3_ppS3_ppS3] v[7]=([S4]**3)*k_on_444 v[8]=[S4_S4_S4]*kdiss_SS v[9]=[S2]*S_phos*[TGFb_pRec] v

[10] =S_dephosphos*[ppS2] v

[11] =S_dephos*[pS2] v

[12] =[S3]*S_phos*[TGFb_pRec] v

[13] =S_dephosphos*[ppS3] v

[14] =S_dephos*[pS3] v

[15] =k_on_223*([ppS2]**2)*[ppS3] v

[16] =2*S_dephosphos*[ppS2_ppS2_ppS3] v

[17] =S_dephosphos*[ppS2_ppS2_ppS3] v

[18] =[S4]*k_on_224*([ppS2]**2) v

[19] =2*S_dephosphos*[ppS2_ppS2_S4] v

[20] =k_on_233*[ppS2]*([ppS3]**2) v

[21] =S_dephosphos]*[ppS2_ppS3_ppS3] v

[22] =2*S_dephosphos]*[ppS2_ppS3_ppS3] v

[23] =[S4]*k_on_334]*([ppS3]**2) v

[24] =2*S_dephosphos*[ppS3_ppS3_S4] v

[25] =([S4]**2)*k_on_244*[ppS2] v

[26] =S_dephosphos*[ppS2_S4_S4] v

[27] =([S4]**2)*k_on_344*[ppS3] v

[28] =S_dephosphos]*[ppS3_S4_S4] v

[29] =[S4]*k_on_234*[ppS2]*[ppS3] v

[30] =S_dephosphos*[ppS2_ppS3_S4] v

[31] =S_dephosphos*[ppS2_ppS3_S4] v

[32] =(THBS1_turn+THBS1_act1*[pS2_ppS3_ppS3]+THBS1_act2*[ppS2_S4_S4]+THBS1_act3*[ppS2_pp S3_S4]) / (THBS1_inh1*[ppS2_ppS3_ppS3]+THBS1_inh2*[ppS2_S4_S4]+THBS1_inh3*[ppS2_ppS3_S4]+1) v

[33] =(CDKN2B_turn+CDKN2B_act1*[pS2_ppS3_ppS3]+CDKN2B_act2*[ppS2_S4_S4]+CDKN2B_act3*[ppS2_pp S3_S4]) / (CDKN2B_inh1*[ppS2_ppS3_ppS3]+CDKN2B_inh2*[ppS2_S4_S4]+CDKN2B_inh3*[ppS2_ppS3_S4]+1) v

[34] =[THBS1]*THBS1_turn v

[35] =[CDKN2B]*CDKN2B_turn] v

[36] =kf*[THBS1]*[TGFb_inact] / (Kmf+[TGFb_inact]) v

[37] =k_on_FMOD*[FMOD]*[TGFb_act]-k_off_FMOD*[FMOD_complex]

[0133] The following values ​​and formulas were used for the initial concentration of each substance (species) and the mass balance rate equation (ordinary differential equation). Note that for each substance, "y" means the concentration of that substance, and "dy / dt" indicates the rate of change in concentration of that substance. TGFb_act=0 :dy / dt=-v[1]+v

[36] -v

[37] Rec=1.84 :dy / dt=-v[1] TGFb_pRec=0 :dy / dt=+v[1]-v[2] S2=1.43e2 :dy / dt=-v[9]+v

[11] S3=1.63e1 :dy / dt=-v

[12] +v

[14] S4=6.71e1 :dy / dt=-3*v[7]+3*v[8]-v

[18] +v

[19] -v

[23] +v

[24] -2*v

[25] +2*v

[26] -2*v

[27] +2*v

[28] -v

[29] +v

[30] +v

[31] ppS2_ppS2_ppS2=0 :dy / dt=+v[3]-v[4] ppS3_ppS3_ppS3=0 :dy / dt=+v[5]-v[6] S4_S4_S4=0 :dy / dt=+v[7]-v[8] pS2=0 :dy / dt=+v[4]+v

[10] -v

[11] +v

[16] +v

[19] +v

[21] +v

[26] +v

[30] pS3=0 :dy / dt=+v[6]+v

[13] -v

[14] +v

[17] +v

[22] +v

[24] +v

[28] +v

[31] ppS2=0 :dy / dt=-3*v[3]+2*v[4]+v[9]-v

[10] -2*v

[15] +v

[16] +2*v

[17] -2*v

[18] +v

[19] -v

[20] +v

[22] -v

[25] -v

[29] +v

[31] ppS3=0 :dy / dt=-3*v[5]+2*v[6]+v

[12] -v

[13] -v

[15] +v

[16] -2*v

[20] +2*v

[21] +v

[22] -2*v

[23] +v

[24] -v

[27] -v

[29] +v

[30] ppS2_ppS2_S4=0 :dy / dt=+v

[18] -v

[19] ppS2_ppS2_ppS3=0 :dy / dt=+v

[15] -v

[16] -v

[17] ppS2_ppS3_ppS3=0 :dy / dt=+v

[20] -v

[21] -v

[22] ppS3_ppS3_S4=0 :dy / dt=+v

[23] -v

[24] ppS2_ppS3_S4=0 :dy / dt=+v

[29] -v

[30] -v

[31] ppS3_S4_S4=0 :dy / dt=+v

[27] -v

[28] ppS2_S4_S4=0 :dy / dt=+v

[25] -v

[26] TGFb_inact=1 :dy / dt=-v

[36] THBS1=1 :dy / dt=+v

[32] -v

[34] CDKN2B=1 :dy / dt=+v

[33] -v

[35] FMOD=1 :dy / dt=-v

[37] FMOD_complex=0 :dy / dt=+v

[37]

[0134] Test Example 4. (Construction of an extended mathematical model of skin aging) In Test Example 3, it was found that the mathematical model of TGFβ signaling is a model capable of simulating experimental values. Meanwhile, FMOD and THBS1 are also involved in Wnt signaling, and a mathematical model regarding TGFβ-Wnt signaling was constructed.

[0135] This model is based on a mathematical model of TGFβ signaling (Test Example 3), and integrates a mathematical model of Wnt signaling (Reference 9; Kim et al. (2007) Oncogene. Jul. 5; 26(31): 4571-9.), and a mathematical model was constructed using ordinary differential equations (ODEs) with FMOD and THBS1 as inputs (details will be described later). An overview of the new signaling model is shown in Figure 10. In Figure 10, the numbers of each reaction correspond to the rate equation of each elementary reaction. Other meanings are described in Figure 10.

[0136] In addition to the characteristics of Test Example 3, the novel signal model is characterized in that TGFβ signaling activates Wnt signaling via PI3K-Akt signaling, as shown in References 10 and 11. Reference 10.Acta Biochim Pol.2018;65(3):341-349. Reference 11.EBioMedicine.2017 Apr;18:179-187.

[0137] Specifically, it has been shown in References 10 and 11 that TGFβ signaling phosphorylates GSK3β via PI3K-Akt signaling. Parameter estimation and simulation in this signaling model were performed using BioMASS.

[0138] In Examples 3, 4, 5, and 6, simulations of gene expression levels were carried out using parameter values ​​calculated by parameter estimation for THBS1, CDKN2B, FMOD, and CCND1, respectively. These data are shown in FIG. 11 together with the in vitro experimental values ​​of THBS1, CDKN2B, FMOD, and CCND1.

[0139] The triangle values ​​in Figure 11 indicate experimental data (mean ± SD, n = 3), and the solid lines indicate the simulation results when each parameter was used. The horizontal axis indicates the degree of cell senescence by PD, and the vertical axis indicates the gene expression level normalized by TPM, and then added 1 to each TPM to convert it to logarithm, and then converted to Log2.

[0140] The experimental values ​​in FIG. 11 were obtained as follows. The experimental values ​​were obtained from in vitro analysis results, i.e., data from Reference 2 (PLoS One. 2016 May 3;11(5):e0154531.(GSE63577)). The analysis results are shown in Table 2 below. The experimental values ​​for THBS1 and CDKN2B used the values ​​shown in Table 1.

[0141] [Table 2]

[0142] Regarding the expression levels of THBS1 and CDKN2B, whose transcription is controlled by TGF-β signaling, parameters were constructed using a new signaling model (Examples 3 and 4) in which Wnt signaling was added to Test Example 3. As a result, as shown in Figure 11, we succeeded in constructing a mathematical model that reproduced the experimental data, and it was found that simulation was possible even if the model was expanded.

[0143] Furthermore, with regard to FMOD and CCND1, whose transcription is controlled by Wnt signals, it can be seen that, as shown in FIG. 11, Examples 5 and 6 well reproduce the in vitro experimental data.

[0144] Details of the ODE model (biomathematical model) used in the simulation are given below. The ODE model is composed of the rate equations for each elementary reaction according to the signal transduction model, the initial values ​​of the concentrations of each substance (species) and the rate equations for the mass balance (ordinary differential equations), and the parameters that serve as the coefficients of these equations. As a solution to this ordinary differential equation, the simulation results shown in Figure 11 can be obtained. BioMASS was used for the simulation.

[0145] First, the following values ​​were used as parameters. These parameters were estimated using BioMASS by the least squares method between the experimental and simulation values ​​so that they were closest to each other. The number after "e" means a multiplier of 10 and indicates the product with the number before "e".

[0146] (Parameter) k1=1.820e-01 k2=1.820e-02 k3=5.000e-02 k4=2.670e-01 k5=1.330e-01 kf_6=9.100e-02 kr_6=9.090e-01 kf_7=1.000e+00 kr_7=5.000e+01 kf_8=1.000e+00 kr_8=1.200e+02 k9=2.060e+02 k10=2.060e+02 k11=4.170e-01 k13=2.570e-04 k14=8.220e-05 k15=1.670e-01 kf_16=1.000e+00 kr_16=3.000e+01 kf_17=1.000e+00 kr_17=1.200e+03 k18=2.152e+01 Km1=2.537e+01 k19=4.128e+01 Km2=1.500e-01 k20=1.000e-06 k_FMOD_turn=1.000e-02 k_CCND1_turn=1.695e-02 n1=1.219e+00 n2=1.120e+00 V12=4.230e-01 kf_21=5.072e+01 kr_21=2.304e+02 Kmf_3=5.513e+00 Kmr_3=7.549e+02 kf_22=4.580e+00 kr_22=1.065e+01 Kmf_4=1.567e+02 Kmr_4=9.224e-01 kf_23=1.547e-02 kr_23=9.643e-01 Kmf_5=2.132e+01 Kmr_5=4.864e+02 Rec_act=1.292e-03 pRec_degind=1.884e-02 S2tot=2.002e+03 S3tot=8.000e+02 S4tot=6.166e+00 S_dephos=1.321e-02 S_dephosphos=1.108e+00 S_phos=5.059e+00 k_on_233=1.470e-01 k_on_234=4.030e-04 k_on_244=8.630e-06 THBS1_act3=5.027e-02 THBS1_act2=1.004e-02 THBS1_act1=1.753e+01 THBS1_inh3=7.819e-06 THBS1_inh2=4.521e-02 THBS1_inh1=2.293e-02 THBS1_turn=3.305e-03 CDKN2B_act3=2.077e+00 CDKN2B_act2=3.875e+03 CDKN2B_act1=9.184e-06 CDKN2B_inh3=6.233e-06 CDKN2B_inh2=1.708e+01 CDKN2B_inh1=1.555e+01 CDKN2B_turn=5.657e-06 k24=7.814e-01 Km6=2.758e-03 kf_25=2.554e-05 kr_25=8.359e-03

[0147] The following formulas were used as the rate equations for each elementary reaction. Note that "**" is synonymous with "^" and means exponentiation. Note that some species are abbreviated as shown in Table 3 to avoid complexity in the formulas.

[0148] [Table 3]

[0149] v[1]=k1*[x1] v[2]=k2*[x2] v[3]=k3*[x2]*[x4] v[4]=k4*[x4] v[5]=k5*[x3] v[6]=kf_6*[x5]*[x6]-kr_6*[x4] v[7]=kf_7*[x7]*[x12]-kr_7*[x6] v[8]=kf_8*[x3]*[x11]-kr_8*[x8] v[9]=k9*[x8] v

[10] =k10*[x9] v

[11] =k11*[x10] v

[12] =V12 v

[13] =k13*[x11] v

[14] =k14+.k20*([x11]+[x14]) v

[15] =k15*[x12] v

[16] =kf_16*[x11]*[x13]-kr_16*[x14] v

[17] =kf_17*[x7]*[x11]-kr_17*[x15] v

[18] =k18*[x14]**n1 / (Km1**n1+[x14]**n1]) v

[19] =k19*[x14]**n2 / (Km2**n2+[x14]**n2]) v

[20] =k_FMOD_turn*[.FMOD] v

[21] =k_CCND1_turn*[.CCND1] v

[22] =Rec_act*[Rec]*[TGFb_act] v

[23] =pRec_degind*[TGFb_pRec] v

[24] =kf_21*[TGFb_pRec]*[PI3K_inact] / (Kmf_3+[PI3K_inact]) v

[25] =kr_21*[PI3K_act] / (Kmr_3+[PI3K_act]) v

[26] =kf_22*[PI3K_act]*[Akt_inact] / (Kmf_4+[.Akt_inact]) v

[27] =kr_22*[Akt_act] / (Kmr_4+[Akt_act]) v

[28] =kf_23*[Akt_act]*[x5] / (Kmf_5+[x5]) v

[29] =kr_23*[x16] / (Kmr_5+[x16]) v

[30] =k24*[THBS1]*[TGFb_inact] / (Km6+[TGFb_inact]) v

[31] =kf_25*[FMOD]*[TGFb_act]-kr_25*[FMOD_complex] v

[32] =[S2]*S_phos*[TGFb_pRec] v

[33] =S_dephosphos*[ppS2] v

[34] =S_dephos*[pS2] v

[35] =[S3]*S_phos*[TGFb_pRec] v

[36] =S_dephosphos*[.ppS3] v

[37] =S_dephos*[pS3] v

[38] =k_on_233*[ppS2]*([ppS3]**2) v

[39] =S_dephosphos*[ppS2_ppS3_ppS3] v

[40] =2*S_dephosphos*[ppS2_ppS3_ppS3] v

[41] =([S4]**2)*k_on_244*[.ppS2] v

[42] =S_dephosphos*[ppS2_S4_S4] v

[43] =[S4]*k_on_234*[ppS2]*[ppS3] v

[44] =S_dephosphos*[ppS2_ppS3_S4] v

[45] =S_dephosphos*[ppS2_ppS3_S4] v

[46] =(THBS1_turn+THBS1_act1*[ppS2_ppS3_ppS3]+THBS1_act2*[ppS2_S4_S4]+THBS1_act3*[ppS2_ppS3_S4]) / (THBS1_inh1*[ppS2_ppS3_ppS3]+THBS1_inh2*[.ppS2_S4_S4]+THBS1_inh3*[ppS2_ppS3_S4]+1) v

[47] =(CDKN2B_turn+CDKN2B_act1*[ppS2_ppS3_ppS3]+CDKN2B_act2*[ppS2_S4_S4]+CDKN2B_act3*[ppS2_ppS3_S4]) / (CDKN2B_inh1*[ppS2_ppS3_ppS3]+CDKN2B_inh2*[ppS2_S4_S4]+CDKN2B_inh3]*[ppS2_ppS3_S4]+1) v

[48] =[THBS1]*THBS1_turn v

[49] =[CDKN2B]*CDKN2B_turn

[0150] The following values ​​and formulas were used for the initial concentration of each substance (species) and the mass balance rate equation (ordinary differential equation). Note that for each substance, "y" means the concentration of that substance, and "dy / dt" indicates the rate of change in concentration of that substance.

[0151] x1=1.000e+02 :dy / dt=-v[1]+v[2] x2=0 :dy / dt=+v[1]-v[2] x3=1.530e-02 :dy / dt=+v[4]-v[5]-v[8]+v

[10] x4=7.600e-03 :dy / dt=-v[3]-v[4]+v[5]+v[6] x5=5.000e+01 :dy / dt=+v[3]-v[6]-v

[28] +v

[29] x6=1.500e-03 :dy / dt=+v[3]-v[6]+v[7] x7=9.660e+01 :dy / dt=-v[7]-v

[17] x8=2.000e-03 :dy / dt=+v[8]-v[9] x9=2.000e-03 :dy / dt=+v[9]-v

[10] x10=9.881e-01 :dy / dt=+v

[10] -v

[11] x11=4.272e+01 :dy / dt=-v[8]+v

[12] -v

[13] -v

[16] -v

[17] x12=8.000e-04 :dy / dt=-v[7]+v

[14] -v

[15] x13=6.188e+00 :dy / dt=-v

[16] x14=8.812e+00 :dy / dt=+v

[16] x15=3.439e+00 :dy / dt=+v

[17] x16=8.554e-01 :dy / dt=+v

[28] -v

[29] FMOD=1 :dy / dt=+v

[18] -v

[20] -v

[31] FMOD_complex=0 :dy / dt=+v

[31] CCND1=1 :dy / dt=+v

[19] -v

[21] THBS1=1 :dy / dt=+v

[46] -v

[48] CDKN2B=1 :dy / dt=+v

[47] -v

[49] TGFb_inact=1 :dy / dt=-v

[30] TGFb_act=0 :dy / dt=-v

[22] +v

[30] -v

[31] Rec=1.84 :dy / dt=-v

[22] TGFb_pRec=0 :dy / dt=+v

[22] -v

[23] PI3K_inact=2.961e+02 :dy / dt=-v

[24] +v

[25] PI3K_act=3.886e+00 :dy / dt=+v

[24] -v

[25] Act_inact=2.979e+02 :dy / dt=-v

[26] +v

[27] Act_act=2.110e+00 :dy / dt=+v

[26] -v

[27] S2=1.430e+02 :dy / dt=-v

[32] +v

[34] S3=1.630e+01 :dy / dt=-v

[35] +v

[37] S4=6.710e+01 :dy / dt=-2*v

[41] +2*v

[42] -v

[43] +v

[44] +v

[45] pS2=0 :dy / dt=+v

[33] -v

[34] +v

[39] +v

[42] +v

[44] pS3=0 :dy / dt=+v

[36] -v

[37] +v

[40] +v

[45] ppS2=0 :dy / dt=+v

[32] -v

[33] -v

[38] +v

[40] -v

[41] -v

[43] +v

[44] ppS3=0 :dy / dt=+v

[35] -v

[36] -2*v

[38] +2*v

[39] +v

[40] -v

[43] +v

[45] ppS2_ppS3_ppS3=0 :dy / dt=+v

[38] -v

[39] -v

[40] ppS3_ppS3_S4=0 :dy / dt=+v

[43] -v

[44] -v

[45] ppS2_ppS3_S4=0 :dy / dt=+v

[36] -v

[37] +v

[40] +v

[45] ppS2_S4_S4=0 :dy / dt=+v

[41] -v

[42]

[0152] Test example 5. (Biosimulation to calculate the degree of skin aging from individual input information using a biomathematical model) In Test Example 4, a mathematical model for TGFβ-Wnt signal transduction was constructed using the aging-related genes FMOD and THBS1 as input. In this test, a biological simulation was performed to calculate the degree of skin aging using in vivo input value information. That is, the gene expression levels of FMOD and THBS1 for each individual were input into the biological model constructed in Test Example 4, and the PD value was virtually calculated for the degree of aging of the individual. This makes it possible to calculate the degree of skin aging rather than the actual age.

[0153] The in vivo input value information was the count data (first gene expression data) from reference 1 (Genome Biol. 2018 Dec 20;19(1):221.(GSE113957)). In order to remove the batch effect of the input value, the count data from in vitro reference 2 (PLoS One. 2016 May 3;11(5):e0154531.(GSE63577)) used in model construction and ComBat-seq from sva (sva: Surrogate Variable Analysis. R package version 3.35.2.) were used to remove the batch effect, and the TPM value was calculated as the input value. By inputting the TPM value into the biological model created in Test Example 4, the virtual PD, which is the degree of aging of an individual, was calculated using a mathematical model. For the calculation, the individual's data was input using the Python extension modules "NumPy" and "matplotlib.pyplot" for the simulation results output by BioMASS.

[0154] For the virtual PD, which is the degree of aging of an individual, the corresponding PD value can be obtained from the gene expression level in the simulation results, but the process shown in Figure 12 is performed for software calculations. Specifically, as shown in Figure 12, first, the logarithmic values ​​of the gene expression levels of each individual are subtracted from the gene expression simulation values ​​of FMOD and THBS1. The horizontal axis of Figure 12 indicates the degree of cellular aging by PD, and the vertical axis indicates the value obtained by normalizing the gene expression levels by TPM, adding 1 to each TPM to make it logarithmic, and then converting it to Log2. The gene expression levels of each individual are calculated by adding 1 to the TPM value from which the batch effect of the in vivo data has been removed, and then converting it to Log2.

[0155] Next, the values ​​below 0 were converted to absolute values, and the point at which the vertical axis was 0 was calculated to calculate the virtual PD for each FMOD and THBS1. Since the upper limit of gene expression reaches a steady state as PD progresses, the PD was set to a maximum value of 200 for the calculation. The virtual PD indicating the degree of aging of each individual was calculated by calculating the average value of the obtained virtual PD for each gene. The results are shown in FIG. 13. The horizontal axis of FIG. 13 shows the age of each individual, and the vertical axis shows the virtual PD value.

[0156] Looking at the whole data, it can be seen that the virtual PD value increases with age, while some individuals are more aged than others of the same age. This result is consistent with the fact that with age, even at the same age, the degree of aging of the skin, etc., tends to vary due to external factors. In other words, this method is a biological simulation method that uses a biological mathematical model to calculate the degree of skin aging from individual input information, and includes the expression levels of FMOD and THBS1 as variables, and can calculate the degree of skin aging of an individual with higher accuracy by using a biological mathematical model.

[0157] Test Example 6. (Drug screening simulation using a biological mathematical model) In this test, the parameters of the reaction promoted or inhibited by drug administration were changed to perform a biological simulation of the gene expression levels of FMOD and THBS1, and the effectiveness of the drug was simulated. That is, the parameters related to the reaction targeted by the drug were varied and input into the biological model constructed in Test Example 4 to perform a biological simulation of the gene expression levels of FMOD and THBS1. This makes it possible to screen drugs without actually conducting in vitro or in vivo experiments, and enables in silico screening of a wide variety of drugs.

[0158] First, in the biological model constructed in Test Example 4, a sensitivity analysis was performed to confirm which reaction formula is most related to the expression control of THBS1 and FMOD. The sensitivity analysis was performed by calculation using BioMASS. In the sensitivity analysis, when a specific parameter is changed, the tendency of various measured Species to change accordingly is analyzed. That is, by performing the sensitivity analysis, it is possible to visualize which reaction is most important in controlling the expression level of a certain gene. The results of the sensitivity analysis on the gene expression levels of FMOD and THBS1 are shown in Figures 14 and 15. The numbers on the horizontal axis in Figures 14 and 15 correspond to the reaction numbers shown in Figure 10, and the vertical axis is an index showing the magnitude of the influence of each reaction.

[0159] From the viewpoint of aging inhibition, the method of suppressing THBS1 and the method of promoting the expression of FMOD were confirmed from the sensitivity analysis results of Figures 14 and 15. As a result, it was found that binding to the TGF receptor abbreviated in reaction 22 or phosphorylation of SMAD abbreviated in reactions 32 and 35 is effective for suppressing the expression of THBS1, and that a method of promoting the synthesis of β-catenin abbreviated in reaction 12 is effective for promoting the expression of FMOD. Since the reaction that controls the binding of transcription factors has a large influence on the control of the expression of each gene, it is possible to control the gene expression levels of FMOD and THBS1 by promoting or suppressing these reactions.

[0160] The parameters used in the drug screening simulation are shown in Table 4. The other parameters used were the same as those used in Test Example 4. The drug screening simulation was calculated using BioMASS. The simulation results are shown in Figures 16, 17, and 18. The horizontal axis in Figures 16, 17, and 18 indicates the degree of cellular senescence by PD, and the vertical axis indicates the gene expression level normalized by TPM, and then added 1 to each TPM to convert to logarithm, and then converted to Log2.

[0161] [Table 4]

[0162] Figure 16 shows the results of simulating the gene expression level of THBS1 under various conditions. In Comparative Example 25, the parameter set adopted in Test Example 4 was used. In Examples 7, 8, and 9, simulations were performed in the case where the parameter Rec_act, which is related to reaction 22, i.e., ligand binding to the TGF receptor, was reduced. In Examples 10, 11, and 12, simulations were performed in the case where the parameter S_phos, which is related to reaction 32 and reaction 35, i.e., phosphorylation of SMAD2 / 3, was reduced.

[0163] In both reactions, the expression level of THBS1 was confirmed to decrease in a parameter-dependent manner by decreasing the reaction parameters. In other words, it was found that it is possible to effectively suppress the gene expression level of THBS1 by using TGF receptor antagonists (e.g., SB-431542, SB-525334, and LY-364947) and Smad2 / 3 phosphorylation inhibitors (e.g., sorafenib and SIS3). In addition, by determining the parameters in the biomathematical model from information on these drugs, it is possible to simulate the effects on aging using the biomathematical model, so that more desirable drugs can be screened from among several candidate drugs. When screening for external factors that affect aging, screening can be done in the same way as for drugs.

[0164] Figure 17 shows the results of simulating the gene expression level of FMOD under various conditions. In Comparative Example 26, the parameter set adopted in Test Example 4 was used. In Examples 13 and 14, simulations were performed in the case where the parameter V12 related to reaction 12, i.e., β-catenin synthesis, was increased. In Comparative Examples 27 and 28, simulations were performed in the case where the parameter k1 related to reaction 1, i.e., phosphorylation of Dvl, was increased.

[0165] In Examples 13 and 14, an increase in the gene expression level of FMOD was confirmed by increasing the reaction parameter V12. On the other hand, in Comparative Examples 27 and 28, which were determined to have little effect on the biological simulation by sensitivity analysis, the gene expression level of FMOD did not change even when the parameters were changed.

[0166] Screening using the model allows quantitative evaluation of not only one type of drug, but also combinations of two or more drugs. Figure 18 shows the results of a simulation of the gene expression level of THBS1 when a TGFR inhibitor and a SMAD2 / 3 phosphorylation inhibitor are combined.

[0167] In Example 15, a combination of parameter changes in Examples 7 and 10 was performed simultaneously, and in Example 16, a combination of parameter changes in Examples 8 and 11 was performed simultaneously. As a result, it was found that it is possible to suppress the expression level of the THBS1 gene more efficiently than by inhibiting each reaction alone.

[0168] Overall, as in the sensitivity analysis, it was found that THBS1 gene expression could be controlled by changing the parameters of TGFR binding and Smad2 / 3 phosphorylation, which were determined to have a large effect. Regarding FMOD gene expression control, it was found that although the gene expression level fluctuated by changing the parameters of β-catenin synthesis, the effect was not large. In other words, from the perspective of aging inhibition, the drug screening simulation revealed that it is possible to approach it more efficiently by suppressing THBS1 than by promoting the expression of FMOD.

Claims

1. A method for generating a biomathematical model relating to aging, comprising the steps of: selecting a first gene set common to a first group of genes whose expression levels change with age, obtained from in vivo first gene expression data of populations including different ages, and a second group of genes whose expression levels change with population doubling number, obtained from in vitro second gene expression data obtained by subculture of a cell line; selecting a second gene set from the first gene set based on information including an expression level of a gene in the first gene expression data or the second gene expression data, and a correlation between aging or a population doubling number and the gene expression level; generating a biological mathematical model including, as variables, expression levels of genes included in the second gene set; Including, The step of generating the biological mathematical model includes a step of constructing a mathematical model based on ordinary differential equations (ODE) according to an assumed signal transduction pathway. Methods for generating biomathematical models.

2. The method for generating a biomathematical model according to claim 1 , wherein in the step of selecting the first gene set, genes related to signal transduction involved in fibrosis are selected.

3. The method for generating a biomathematical model according to claim 1 or 2, wherein in the step of selecting the second gene set, the second gene set includes FMOD and THBS1.

4. The method for generating a biological mathematical model according to any one of claims 1 to 3, further comprising a step of verifying the generated biological mathematical model by comparing a change over time in the expression level of a specific gene calculated using the generated biological mathematical model with a change due to population doubling number in the expression level of the specific gene in vitro obtained by subculturing a cell line.

Citation Information

Patent Citations

  • Method of identifying cosmetic agents for producing skin care compositions

    JP2016105296A

  • Cell determination device, cell determination method and program

    JP2019150018A

  • Deep transcriptomic markers of human biological aging and methods of determining a biological aging clock

    US20190034581A1