Method and system for analyzing functional state of intestinal flora based on multi-omics joint analysis

Abstract

The application discloses a method and system for analyzing the functional state of intestinal flora based on multi-omics joint analysis, and relates to the technical field of data analysis and microorganism technology.The technical solution points are as follows: obtaining metagenomic sequencing data and metabolome detection data of host intestinal flora; constructing an interaction network atlas containing nodes and edges, wherein the nodes include metabolite nodes and flora functional pathway nodes, the edges are association lines connecting metabolites and flora functional pathways, and the association weight of the association lines is determined; screening key interaction nodes from the interaction network atlas; and quantitatively evaluating the functional state of the host intestinal flora based on the key interaction nodes.The application constructs a dynamic interaction network atlas by integrating metagenomic sequencing data and metabolome detection data, for the first time realizes the spatiotemporal association modeling of flora functional pathways and metabolites, overcomes the technical defects of the limited analysis perspective of single-omics, and significantly improves the systematicness and biological interpretability of the functional state evaluation.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data analysis and microbiology, more particularly, it relates to a method and system for analyzing the functional state of intestinal flora based on multi-omics joint analysis. BACKGROUND

[0002] As an important "invisible organ" of the human body, the functional state of intestinal flora is closely related to host metabolism, immunity and disease development. Precise evaluation of the functional activity of intestinal flora can not only reveal the mechanism of flora-host interaction, but also provide key biomarkers for personalized nutritional intervention, early warning of diseases and targeted treatment. Especially in the fields of metabolic diseases, autoimmune diseases and tumor microenvironment regulation, establishing a multi-dimensional and dynamic evaluation system for the function of flora has become a core requirement of current precision medicine research.

[0003] Currently, the functional analysis of intestinal flora is mostly limited to static analysis of single-omics data. For example, some existing technologies calculate the abundance of functional pathways based on metagenomic sequencing data, but do not integrate metabolomic information, resulting in the inability to capture the dynamic regulatory effects of microbial metabolites. Some existing technologies use metabolomic correlation analysis to screen for differential metabolites, but lack causal correlation modeling of microbial functional pathways, making it difficult to distinguish between symbiotic metabolism and host metabolism contribution. In addition, traditional methods mostly construct flora-metabolite networks through statistical correlation, such as Spearman correlation coefficient. This co-occurrence-based correlation analysis is susceptible to confounding factors and cannot identify driving nodes with biological significance.

[0004] Therefore, how to research and design a method and system for analyzing the functional state of intestinal flora based on multi-omics joint analysis that can overcome the above-mentioned defects is a problem we need to solve at present. SUMMARY

[0005] To solve the problems in the prior art, the purpose of the present application is to provide a method and system for analyzing the functional state of intestinal flora based on multi-omics joint analysis, which integrates metagenomic sequencing data and metabolomic detection data to construct a dynamic interaction network map, and for the first time realizes the spatiotemporal correlation modeling of microbial functional pathways and metabolites, overcoming the technical defects of single-omics analysis perspective limitations, and significantly improving the systematicness and biological interpretability of functional state evaluation.

[0006] The above technical purposes of the present application are achieved by the following technical solutions:

[0007] In a first aspect, a method for analyzing the functional state of intestinal flora based on multi-omics joint analysis is provided, comprising the following steps:

[0008] Obtaining metagenomic sequencing data and metabolomic detection data of the host intestinal flora;

[0009] constructing an interaction network graph comprising nodes and edges based on the metagenomic sequencing data and the metabolomic detection data, the nodes comprising metabolite nodes and gut microbiota functional pathway nodes, the edges being association lines connecting metabolites and gut microbiota functional pathways, and determining association weights of the association lines;

[0010] screening key interaction nodes from the interaction network graph;

[0011] quantitatively evaluating a functional state of the host gut microbiota based on the key interaction nodes, the functional state being a metabolic activity index, an immune regulation ability index, and / or a host interaction risk level.

[0012] Further, the determining of the association weights of the association lines comprises:

[0013] calculating mutual information between the metabolites and the gut microbiota functional pathways using a kernel density estimation method;

[0014] inputting metabolite concentrations and clinical phenotypes into a logistic regression model to calculate AUC values;

[0015] determining the association weights of the corresponding association lines in combination with the mutual information and the AUC values.

[0016] Further, the determining of the association weights of the association lines comprises:

[0017] determining time-series abundances of each of the gut microbiota functional pathways in the metagenomic sequencing data, and taking a ratio of the time-series abundance to a maximum historical abundance of the corresponding gut microbiota functional pathway as a first parameter;

[0018] determining time-series concentrations of each of the metabolites in the metabolomic detection data, and taking a ratio of the time-series concentration to a maximum historical concentration of the corresponding metabolite as a second parameter;

[0019] inputting the first parameter and the second parameter into a long short-term memory network to determine the association weights of the corresponding association lines.

[0020] Further, the screening of the key interaction nodes from the interaction network graph comprises:

[0021] analyzing causal directions of the gut microbiota functional pathways and the metabolome by a LiNGAM model;

[0022] calculating causal effect values of the edges according to the causal directions;

[0023] screening strong causal edges with causal effect values greater than or equal to a first threshold value, and taking nodes connected to the strong causal edges as the key interaction nodes.

[0024] Further, the screening of the key interaction nodes from the interaction network atlas comprises:

[0025] dividing functional subnets according to the metagenomic sequencing data;

[0026] calculating the functional synergy gain of the nodes in the functional subnets;

[0027] screening the nodes with the functional synergy gain greater than or equal to a second threshold value as the key interaction nodes.

[0028] Further, the screening of the key interaction nodes from the interaction network atlas comprises:

[0029] evaluating the time sequence resilience index of each node in the dynamic network according to the time sequence data in the metabolomic detection data;

[0030] screening the nodes with the time sequence resilience index greater than or equal to a third threshold value as the key interaction nodes.

[0031] Further, if the functional state is the metabolic activity index, the quantitative evaluation comprises:

[0032] pairing the metabolites and the functional pathways of the flora in all the key interaction nodes to obtain paired combinations;

[0033] calculating the product of the time sequence abundance and the time sequence concentration in each paired combination, multiplying the corresponding correlation weight, and obtaining metabolic contribution results;

[0034] accumulating the metabolic contribution results of all the paired combinations to obtain preliminary metabolic activity results;

[0035] normalizing the preliminary metabolic activity results by using a normalization coefficient to obtain the metabolic activity index.

[0036] Further, if the functional state is the immune regulation ability index, the quantitative evaluation comprises:

[0037] determining the probiotic gain parameters and the pathogenic bacteria inhibition parameters corresponding to all the key interaction nodes respectively;

[0038] determining the net regulation ability of the host intestinal flora to the immune system by the ratio of the probiotic gain parameters to the pathogenic bacteria inhibition parameters;

[0039] determining a balance correction coefficient according to the concentration ratio of anti-inflammatory factors to pro-inflammatory factors;

[0040] taking the product of the net regulation ability and the balance correction coefficient as the immune regulation ability index.

[0041] Further, if the functional state is a host interaction risk level, the quantitative evaluation includes:

[0042] determining a causal effect value of each microbiota node to a host phenotype node directly or indirectly regulated downstream;

[0043] determining a clinical severity of the host phenotype node;

[0044] multiplying the causal effect value and the clinical severity, and accumulating to determine the host interaction risk level;

[0045] wherein the microbiota node is a biological entity or a pathway with functional attributes in the host intestinal microbiota, and the host phenotype node is a clinical preparation or a pathological feature directly related to the host health status.

[0046] In a second aspect, a functional state analysis system of intestinal microbiota based on multi-omics joint analysis is provided, which is used to implement the functional state analysis method of intestinal microbiota based on multi-omics joint analysis as described in any one of the first aspect, and includes:

[0047] a data acquisition module configured to acquire metagenomic sequencing data and metabolome detection data of a host intestinal microbiota;

[0048] a graph construction module configured to construct an interaction network graph containing nodes and edges based on the metagenomic sequencing data and the metabolome detection data, wherein the nodes include metabolite nodes and microbiota functional pathway nodes, and the edges are association lines connecting metabolites and microbiota functional pathways, and an association weight of the association lines is determined;

[0049] a node screening module configured to screen key interaction nodes from the interaction network graph;

[0050] a state evaluation module configured to quantitatively evaluate a functional state of the host intestinal microbiota based on the key interaction nodes, wherein the functional state is a metabolic activity index, an immune regulation ability index and / or a host interaction risk level.

[0051] Compared with the prior art, the present application has the following beneficial effects:

[0052] 1. The functional state analysis method of intestinal microbiota based on multi-omics joint analysis provided by the present application integrates metagenomic sequencing data and metabolome detection data to construct a dynamic interaction network graph, which firstly realizes the spatiotemporal association modeling of microbiota functional pathways and metabolites, overcomes the technical defects of the perspective limitation of single-omics analysis, and significantly improves the systematicness and biological interpretability of the functional state evaluation.

[0053] 2、The application fuses the correlation weight by nuclear density estimation and logistic regression model, effectively distinguishes statistical correlation and clinical significant correlation, avoids the problem of insufficient capture of non-linear relationship of traditional mutual information method, and can improve the screening accuracy of key interaction nodes;

[0054] 3、The application quantifies the causal effect value by using the LiNGAM model, breaks through the pseudo-correlation interference of the traditional correlation network, accurately identifies the key functional module of the driving host phenotype, and provides a verifiable molecular target for targeted intervention;

[0055] 4、The application introduces the time series resilience index to evaluate the dynamic stability of the node, solves the problem of instantaneous fluctuation misjudgment caused by single time point analysis, and can improve the consistency of metabolic activity index prediction and clinical outcome. BRIEF DESCRIPTION OF DRAWINGS

[0056] The drawings described herein are used to provide further understanding of the embodiments of the application, constitute a part of the application, and do not constitute a limitation on the embodiments of the application. In the drawings:

[0057] Figure 1 is the flow chart in embodiment 1 of the application;

[0058] Figure 2 is the system block diagram in embodiment 2 of the application. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical scheme and advantages of the application more clear and obvious, the application is further described in detail below in combination with embodiments and drawings, and the illustrative embodiments of the application and the description thereof are only used to explain the application, and do not constitute a limitation on the application.

[0060] Embodiment 1: A method for analyzing the functional state of intestinal flora based on multi-omics joint analysis, as shown in Figure 1 The method comprises the following steps:

[0061] S1: Obtain the metagenomic sequencing data and the metabolome detection data of the host intestinal flora;

[0062] S2: Construct an interaction network atlas containing nodes and edges based on the metagenomic sequencing data and the metabolome detection data;

[0063] S3: Screen out key interaction nodes from the interaction network atlas;

[0064] S4: Quantitatively evaluate the functional state of the host intestinal flora based on the key interaction nodes, and the functional state is a metabolic activity index, an immune regulation ability index and / or a host interaction risk level.

[0065] In step S1, the host gut microbiota is generally a mammal, such as a human and a cynomolgus monkey. The present application is described in detail taking a cynomolgus monkey caused by a mutation in the hemoglobin beta (HBB) gene as an example.

[0066] The metagenomic sequencing data is mainly based on KEGG (bioinformatics database) functional annotation to calculate the abundance of the gene pathways of the microbiota. The metabolomic detection data is mainly to detect the differential metabolites through UPLC-MS (ultra-performance liquid chromatography-mass spectrometry).

[0067] For example, for the metabolomic detection data, the metabolites in the fecal sample are detected using the UPLC-MS technology, peak extraction and annotation (matching the HMDB and mzCloud databases) are performed through the Compound Discoverer software; the original peak area is then log2 transformed, batch effect corrected (such as the ComBat algorithm), and the differential metabolites are screened (Fold Change>1.2, p<0.05).

[0068] For example, for the metagenomic sequencing data, the fecal sample is subjected to metagenomic sequencing (Illumina NovaSeq), contigs (continuous gene fragments) are assembled using MEGAHIT (a metagenomic assembly tool), and genes are predicted using MetaGeneMark (a prokaryotic gene prediction tool); the genes are then mapped to the KEGG pathways through HUMAnN3 (a metagenomic functional pathway analysis tool) to calculate the pathway abundance (RPKM standardization).

[0069] In step S2, the interaction network graph can be constructed using a graph convolution network, and the nodes in the interaction network graph include metabolite nodes and microbiota functional pathway nodes.

[0070] In some examples, the metabolite nodes in the interaction network graph represent the detected host or microbiota metabolites, such as HMDB0254633 (3-oxooctadecanoic acid) and short-chain fatty acids SCFA.

[0071] In some examples, the microbiota functional pathway nodes in the interaction network graph represent the KEGG functional pathways in which the microbiota is involved, such as alanine metabolism and linoleic acid metabolism.

[0072] In some examples, the edges in the interaction network graph are the association lines connecting the metabolites and the microbiota functional pathways, representing the interaction relationship therebetween, such as positive regulation and negative inhibition. The edges in the interaction network graph are configured with two attributes of association weight and causal effect value.

[0073] In some examples, the interaction network atlas can also be divided into local network modules according to functions or mechanisms, such as metabolic activity subnets and immune regulation subnets. Each subnet is configured with two attributes of function label and synergistic effect, such as the function label can be classified into first-level metabolism and second-level immunity based on KEGG hierarchy. In addition, the synergistic effect is the synergistic relationship between subnets captured by the attention mechanism, such as metabolic subnets driving immune regulation.

[0074] In some examples, considering that traditional mutual information (MI) calculation relies on discrete binning, which may lose continuous data information, and adopting kernel density estimation (KDE) to estimate continuous probability density through a smoothing kernel function (such as a Gaussian kernel), the non-linear correlation can be more accurately captured, such as the U-shaped relationship between metabolite concentration and pathway abundance. In addition, considering that mutual information (MI) relying only on statistical association may screen out metabolite-pathway pairs unrelated to the condition, the present application quantifies the discriminant ability of metabolites for lesion phenotypes by AUC value (obtained by ROC curve analysis), and ensures that the weight is biased towards the association with clear clinical significance.

[0075] To this end, the association weight of the association line is determined, including: calculating the mutual information between the metabolite and the functional pathway of the flora using the kernel density estimation method; inputting the metabolite concentration and the clinical phenotype into a logistic regression model to calculate the AUC value; and determining the association weight of the corresponding association line in combination with the mutual information and the AUC value.

[0076] Specifically, the calculation formula of mutual information is as follows:

[0077]

[0078] wherein, MI(m a , g b ) represents the mutual information between the a a,b th metabolite m a and the b b th functional pathway of the flora g a ; x represents the binning interval of the metabolite concentration, and each metabolite can correspond to multiple binning intervals; y represents the binning interval of the flora pathway abundance, and each flora functional pathway can correspond to multiple binning intervals; p(x, y) represents the joint distribution probability of the binning interval x and the binning interval y; p(x) represents the marginal distribution probability of the binning interval x; and p(y) represents the marginal distribution probability of the binning interval y.

[0079] Specifically, the calculation formula of the association weight is as follows:

[0080]

[0081] wherein, w a,b represents the association weight of the a a th metabolite m b and the b a th functional pathway of the flora g a .The association weight of the correlation lines between them; AUC(m a ) represents the a-th metabolite m a The ability to discriminate lesion phenotypes; MI(m k ,g b ) represents the k-th metabolite m k With the functional pathway of the bth microbial community g b Mutual information between them; AUC(m) k ) represents the k-th metabolite m k The ability to distinguish lesion phenotypes; N represents the total amount of metabolites.

[0082] In some examples, determining the association weights of the association lines includes: determining the temporal abundance of each microbial functional pathway in the metagenomic sequencing data, and using the ratio of the temporal abundance to the historical maximum abundance of the corresponding microbial functional pathway as a first parameter; determining the temporal concentration of each metabolite in the metabolomics detection data, and using the ratio of the temporal concentration to the historical maximum concentration of the corresponding metabolite as a second parameter; inputting the first and second parameters into a long short-term memory network to determine the association weights of the corresponding association lines.

[0083] Specifically, the expression for determining the association weight is:

[0084]

[0085] Among them, w a,b (t) represents the association weight of the correlation line between the a-th metabolite and the b-th microbial functional pathway at time t; M a (t) represents the time-series concentration of the a-th metabolite in the metabolomics data at time t; max(M a ) represents the historical maximum concentration of the a-th metabolite; G b (t) represents the temporal abundance of the b-th microbial community functional pathway at time t in metagenomic sequencing data; max(G b ) represents the historical abundance maximum of the b-th microbial community functional pathway.

[0086] This invention captures the dynamic synergistic or antagonistic effects of microbial communities and metabolites through long short-term memory networks, learns complex regulatory relationships through gating mechanisms, and finally eliminates data heterogeneity by normalizing based on historical maximum values. This can effectively identify the nonlinear regulation of pathway activity by metabolite concentration.

[0087] In step S3, since the interaction network of the host intestinal flora and metabolites usually contains thousands of nodes, direct full network analysis can cause computational complexity explosion; in addition, experimental errors (such as ion interference of mass spectrometry) and statistical false positives are common in multi-omics data. Therefore, the application filters out key interaction nodes of the core pathway and metabolites that are directly involved in the host pathology from the interaction network map to improve computational efficiency and ensure the accuracy of the evaluation results.

[0088] In some examples, filtering out the key interaction nodes from the interaction network map includes: analyzing the causal direction of the flora functional pathway and the metabolome by the LiNGAM model; calculating the causal effect value of the edge according to the causal direction; filtering out strong causal edges with a causal effect value greater than or equal to a first threshold, and the nodes connected with the strong causal edges as the key interaction nodes.

[0089] The LiNGAM (Linear Non-Gaussian Acyclic Model) model is a data-driven causal inference model that quantifies the causal effect between nodes by analyzing the linear relationship and non-Gaussian noise distribution between variables.

[0090] Specifically, first, the metabolite concentration matrix M (n x p) and the flora pathway abundance matrix G (n x q) are determined, and M and G are standardized by Z-score (mean 0, variance 1), where n is the sample number, p is the number of metabolite species, and q is the number of flora pathway species.

[0091] Next, the metabolites and flora pathways are combined into a mixed matrix X = [M, G], (n x (p+q)), for example, if 3 metabolites (HMDB0254633, SCFA, o-tyrosine) and 2 flora pathways (K11212, K07244) are detected, then X is an n x 5 matrix.

[0092] The matrix X is decomposed by independent component analysis (ICA) to find the causal direction, and the decomposition expression is: X = B ∈; where B is the causal effect coefficient matrix, with a dimension of (q+p) x (q+p), and the element b ij represents the causal effect value of variable X i (flora pathway abundance) on variable X j (metabolite concentration); ∈ is an independent non-Gaussian noise with a dimension of n x (p+q), representing unobserved random disturbances.

[0093] The matrix B is estimated by ICA, and the row vectors are permuted to convert B into a lower triangular matrix. For example, if the causal direction is X1→X3→X2, then the matrix B is:

[0094]

[0095] Based on the causal direction, estimate b by ordinary least squares (OLS) ij For each variable X j , only use the variables X i (i < j) before its causal direction to regress. For example, if the causal direction of X3 is after X1, b 31 represents the causal effect value of X1→X3.

[0096] In some examples, the key interaction nodes are screened from the interaction network map, including: dividing functional subnets according to metagenomic sequencing data; calculating the functional synergy gain of the nodes in the functional subnets; and screening the nodes with a functional synergy gain greater than or equal to a second threshold value as the key interaction nodes.

[0097] The functional synergy gain refers to the additional functional benefit generated by the node through cross-subnet regulation, that is, the enhancement effect of the overall network function (such as metabolic activity, immune regulation) when the node plays a pivotal role in multiple functional subnets. For example: a metabolite node (such as HMDB0254633) regulates alanine metabolism in the metabolic activity subnet, while inhibiting pro-inflammatory bacteria in the immune regulation subnet, and its cross-module synergy can improve the overall network stability.

[0098] If the node has high betweenness centrality in multiple subnets (such as metabolic and immune subnets), the value of the functional synergy gain is high, indicating that the synergy gain is strong.

[0099] The calculation process of the functional synergy gain is: determining the betweenness centrality of the node in the subnet, which is an important measure of the node's importance as a bridge within the subnet, such as the ability to control information flow; determining the association strength of the subnet with the clinical phenotype, such as the AUC value calculated by the regression model; calculating the ratio of the betweenness centrality to the total number of nodes in the subnet, and then combining the association strength for weighting, and finally accumulating the gains of different subnets to obtain the functional synergy gain.

[0100] In some examples, the key interaction nodes are screened from the interaction network map, including: evaluating the time sequence resilience index of each node in the dynamic network according to the time sequence data in the metabolome detection data; and screening the nodes with a time sequence resilience index greater than or equal to a third threshold value as the key interaction nodes.

[0101] It should be noted that the time sequence resilience index is a dynamic network analysis tool for quantifying the stability and regulation potential of the node in the time sequence data, so as to screen out the nodes that have a long-term key role in the network function. In the present application, the main purpose is to screen out biomarkers that continue to play a role in disease development or intervention treatment, such as the metabolite HMDB0254633 that regulates iron absorption for a long time, so as to avoid misjudgment caused by single-time-point analysis, such as temporary fluctuation nodes.

[0102] In some examples, the three key interaction node screening methods described above can be combined to screen key interaction nodes.

[0103] In step S4, the functional state evaluation extracts high-confidence functional features of the flora from the massive data, providing data-driven decision-making basis for disease mechanism analysis, precise diagnosis and treatment, and drug development. The metabolic activity index quantifies the activity of metabolic pathways in which the flora participates, the immune regulation ability index evaluates the regulation ability of the flora on the host immune system, and the host interaction risk level predicts the contribution of the flora function to the host pathology.

[0104] In some examples, if the functional state is the metabolic activity index, the quantitative evaluation includes: pairing the metabolites and flora functional pathways in all key interaction nodes to obtain paired combinations; calculating the product of the time series abundance and the time series concentration in each paired combination, and multiplying it by the corresponding correlation weight to obtain the metabolic contribution result; accumulating the metabolic contribution results of all paired combinations to obtain the preliminary metabolic activity result; and normalizing the preliminary metabolic activity result using a normalization coefficient to obtain the metabolic activity index.

[0105] In some examples, if the functional state is the immune regulation ability index, the quantitative evaluation includes: determining the probiotic gain parameters and pathogenic bacteria inhibition parameters corresponding to all key interaction nodes respectively; determining the net regulation ability of the host intestinal flora on the immune system by the ratio of the probiotic gain parameters to the pathogenic bacteria inhibition parameters; determining a balance correction coefficient according to the concentration ratio of anti-inflammatory factors to pro-inflammatory factors; and taking the product of the net regulation ability and the balance correction coefficient as the immune regulation ability index.

[0106] In some examples, if the functional state is the host interaction risk level, the quantitative evaluation includes: determining the causal effect value of each flora node to the host phenotype node directly or indirectly regulated in the downstream; determining the clinical severity of the host phenotype node; multiplying the causal effect value and the clinical severity, and accumulating to determine the host interaction risk level; wherein the flora node is a biological entity or pathway with functional attributes in the host intestinal flora, and the host phenotype node is a clinical preparation or pathological feature directly related to the health status of the host.

[0107] Comparative test: To verify the sensitivity and specificity of the technical solutions described in the present application in evaluating the functional state of the intestinal flora, the metabolic activity index MAI, the immune regulation ability index IRI, and the host interaction risk level HRR of the gene-mutated cynomolgus monkey (HBB) and the wild-type cynomolgus monkey (WT) are compared to reveal the influence of gene mutation on flora-host interaction.

[0108] The metabolic activity index comparison results are shown in Table 1, and the results show that the amino acid metabolic pathway (phenylalanine metabolism) in the HBB group is significantly down-regulated, which is consistent with the decrease in MAI.

[0109] Table 1 Comparison of metabolic activity index

[0110] Group MAI mean ± SD Significance test (t-test) HBB mutation group 12.3±2.1 p＝0.002(↓35％) WT group 18.9±3.4

[0111] The comparison results of immune regulation ability index are shown in Table 2. The serum IL-6 / IL-10 concentration detected by ELISA is significantly negatively correlated with the IRI calculation result (r = -0.82, p < 0.01).

[0112] Table 2 Comparison of immune regulation ability index

[0113]

[0114] The comparison results of host interaction risk level are shown in Table 3. The serum ferritin level (≥ 1500 ng / mL) of the HBB group is significantly positively correlated with the HRR (r = 0.75, p = 0.003).

[0115] Table 3 Comparison of host interaction risk level

[0116]

[0117] The above experiments confirm that HBB mutation leads to impaired intestinal flora metabolic activity, immune regulation imbalance and increased host pathological risk in cynomolgus monkeys through multi-omics joint analysis.

[0118] Example 2: An intestinal flora functional state analysis system based on multi-omics joint analysis, which is used to implement the intestinal flora functional state analysis method based on multi-omics joint analysis as described in Example 1, as shown in Figure 2 The system includes a data acquisition module, a graph construction module, a node screening module and a state evaluation module.

[0119] The data acquisition module is configured to acquire metagenomic sequencing data and metabolomic detection data of the intestinal flora of the host. The graph construction module is configured to construct an interaction network graph containing nodes and edges based on the metagenomic sequencing data and the metabolomic detection data, wherein the nodes include metabolite nodes and flora functional pathway nodes, and the edges are association lines connecting the metabolites and the flora functional pathways, and the association weights of the association lines are determined. The node screening module is configured to screen out key interaction nodes from the interaction network graph. The state evaluation module is configured to quantitatively evaluate the functional state of the intestinal flora of the host based on the key interaction nodes, and the functional state is a metabolic activity index, an immune regulation ability index and / or a host interaction risk level.

[0120] Working principle: the application constructs a dynamic interaction network atlas by integrating metagenomic sequencing data and metabolome detection data, for the first time realizes the spatiotemporal correlation modeling of the functional pathway of the flora and the metabolic product, overcomes the technical defects of the single omics analysis perspective limitation, and significantly improves the systematicness and biological interpretability of the functional state evaluation.

[0121] Those skilled in the art will appreciate that embodiments of the application can be provided as methods, systems, or computer program products. Accordingly, the application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the application can be embodied in the form of a computer program product on one or more computer readable storage media (including, but not limited to, disk memory, CD-ROMs, optical storage media, etc.) having computer usable program code embodied therein.

[0122] The application is described with reference to the flowcharts and / or block diagrams according to the methods, devices (systems), and computer program products of the embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the flow Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks or flows.

[0123] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including instruction apparatus, which implements the flow Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks or flows.

[0124] These computer program instructions can also be loaded into a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to produce a computer implemented process, so that the instructions executed on the computer or other programmable device provide a process for implementing the flow Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more blocks or flows.

[0125] The above detailed description of the specific embodiments of the present application has been given to understand the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A system for analyzing the functional state of intestinal flora based on multi-omics joint analysis, characterized in that, The method comprises the following steps: a data acquisition module for acquiring host intestinal flora metagenomic sequencing data and metabolomics detection data; a graph construction module for constructing an interaction network graph containing nodes and edges based on the metagenomic sequencing data and the metabolomics detection data, the nodes including metabolite nodes and flora functional pathway nodes, and the edges being association lines connecting metabolites and flora functional pathways, and determining the association weight of the association lines; a node screening module for screening key interaction nodes from the interaction network graph; a state evaluation module for quantitatively evaluating the functional state of the host intestinal flora based on the key interaction nodes, the functional state being a metabolic activity index, an immune regulation ability index, and / or a host interaction risk level; The determination of the association weight of the association line comprises: using kernel density estimation method to calculate the mutual information between the metabolite and the flora functional pathway; inputting metabolite concentration and clinical phenotype into a logistic regression model to calculate AUC value; combining the mutual information and the AUC value to determine the association weight of the corresponding association line; The determination of the association weight of the association line comprises: determining the time-series abundance of each flora functional pathway in the metagenomic sequencing data, and taking the ratio of the time-series abundance to the maximum historical abundance of the corresponding flora functional pathway as a first parameter; determining the time-series concentration of each metabolite in the metabolomics detection data, and taking the ratio of the time-series concentration to the maximum historical concentration of the corresponding metabolite as a second parameter; inputting the first parameter and the second parameter into a long short-term memory network to determine the association weight of the corresponding association line; The screening of the key interaction nodes from the interaction network graph comprises: analyzing the causal direction of the flora functional pathway and the metabolome by LiNGAM model; calculating the causal effect value of the edge according to the causal direction; screening strong causal edges with a causal effect value greater than or equal to a first threshold value, and the nodes connected with the strong causal edges as the key interaction nodes; The screening of the key interaction nodes from the interaction network graph comprises: dividing functional subnets according to the metagenomic sequencing data; calculating the functional synergy gain of the nodes in the functional subnets; screening nodes with a functional synergy gain greater than or equal to a second threshold value as the key interaction nodes; Or, the screening of the key interaction nodes from the interaction network graph comprises: evaluating the time-series resilience index of each node in the dynamic network according to the time-series data in the metabolomics detection data; screening nodes with a time-series resilience index greater than or equal to a third threshold value as the key interaction nodes. 2.The system for analyzing the functional state of intestinal flora based on multi-omics joint analysis according to claim 1, wherein, If the functional state is the metabolic activity index, the quantitative evaluation comprises: pairing the metabolites and the flora functional pathways in all the key interaction nodes to obtain paired combinations; calculating the product of the time-series abundance and the time-series concentration in each paired combination, and multiplying it by the corresponding association weight to obtain metabolic contribution results; accumulating the metabolic contribution results of all the paired combinations to obtain preliminary metabolic activity results; The preliminary metabolic activity result is normalized by a normalization coefficient to obtain the metabolic activity index. 3.The system for analyzing the functional state of intestinal flora based on multi-omics joint analysis according to claim 1, wherein, If the functional status is the immune regulation ability index, the quantitative evaluation includes: determining the probiotic gain parameter and the pathogenic bacteria inhibition parameter corresponding to each of the key interaction nodes, respectively; determining the net regulation ability of the host intestinal flora to the immune system by the ratio of the probiotic gain parameter to the pathogenic bacteria inhibition parameter; determining a balance correction coefficient according to the concentration ratio of anti-inflammatory factors to pro-inflammatory factors; taking the product of the net regulation ability and the balance correction coefficient as the immune regulation ability index. 4.The system for analyzing the functional state of intestinal flora based on multi-omics joint analysis according to claim 1, wherein, If the functional status is the host interaction risk level, the quantitative evaluation includes: determining the causal effect value of each flora node to the host phenotype node directly or indirectly regulated downstream; determining the clinical severity of the host phenotype node; multiplying the causal effect value and the clinical severity, and accumulating to determine the host interaction risk level; wherein the flora node is a biological entity or a pathway with functional attributes in the host intestinal flora, and the host phenotype node is a clinical preparation or a pathological feature directly related to the host health status.

Images