Method for granger causality analysis between multiple flora based on pathogenic microorganism metagenome

Abstract

The application discloses a method for Granger causality analysis among multiple bacterial flora based on pathogenic microorganism metagenomics, which has the principle that: the obtained pathogenic microorganism metagenomics data is subjected to cluster analysis and each bacterial species is divided, the species abundance of each bacterial species at a corresponding time node of disease development is calculated, then the time series of each group of variables at each time node is established and modified, and finally the modified data is subjected to Granger causality analysis to find the main pathogenic bacteria. The high-throughput sequencing platform is used to identify multiple bacteria, and then the Granger causality analysis method is combined to find the causality relationship among the bacteria, so as to provide a fast identification method for finding one or several pathogenic bacteria from pathogenic microorganism samples, and to help clinicians analyze and judge more quickly to realize clinical diagnosis.

Description

Technical Field

[0001] This invention relates to the fields of high-throughput sequencing and analysis technology, specifically to a Granger causal analysis method for multiple bacterial communities based on pathogenic microbial metagenomics. Background Technology

[0002] Diagnosing infections of unknown origin, complex and critical conditions, and immunodeficiency has always been a challenging problem for clinicians. The causes commonly fall into three categories: infection, rheumatic and immune diseases, and malignant tumors. These patients often present difficulties in diagnosing the pathogen, which is numerous and complex. Because the pathogen is often unclear, treatment is mostly limited to empirical anti-infective therapy, lacking precision. Therefore, accurate and early identification of the pathogen and targeted anti-infective treatment are crucial for the prognosis of these patients.

[0003] Metagenomics next-generation sequencing (mNGS) based on high-throughput sequencing can detect DNA or RNA from all species in a sample, enabling rapid analysis of the overall microbiome in patient samples as well as the human host genome and transcriptome. Therefore, mNGS has significant advantages in discovering novel pathogens and detecting humans in healthy and diseased states. The literature “Boulange CL, Neves AL, Chilloux J et al. Impact of the gut microbiota on inflammation, obesity, and metabolic disease[J]. Genome Med, 2016, 8(1): 42.” uses high-throughput sequencing to analyze the close relationship between gut microbiota and diseases such as obesity, diabetes, and inflammatory bowel disease. The literature “Allegretti M, Fabi A, Bugliani S, Martayan A et al. Tearing down the walls: FDA approves next generation sequencing (NGS) assays for actionable cancer genomic aberrations [J]. ExpClin Cancer Res, 2018, 37(1): 47.” found that mNGS can identify cancer-related viruses and be used to guide drug use in advanced tumors. Therefore, targeting microorganisms and using mNGS to intervene in the microbiome will inevitably become a new direction for clinical treatment.

[0004] The continuous improvement of mNGS technology is driving the development of precision medicine. Since metagenomic sequencing of pathogenic microorganisms can detect up to 6549 bacteria with known genome sequences, typically consisting mainly of host-derived sequences and a very small portion of pathogen sequences, it presents a "needle in a haystack" problem for clinical diagnosis. Therefore, establishing a background microbial database (including normal microorganisms and contaminant microorganisms detected during mNGS sequencing) is crucial. We first perform cluster analysis on the acquired pathogenic microbial metagenomic data to classify each species, then calculate the species abundance of each species at corresponding time points in disease development. We then establish time series for each variable at each time point and make corrections. Finally, we perform Granger causality analysis on the corrected data to identify the main pathogenic species, thus assisting clinicians in making faster analyses and judgments for clinical diagnosis. Summary of the Invention

[0005] This invention utilizes a high-throughput sequencing platform to identify a variety of bacteria, and then combines Granger causal analysis to find the causal relationships between these bacteria. This provides a rapid identification method for finding one or more pathogenic bacteria from pathogenic microbial samples, and helps clinicians to analyze and judge more quickly to achieve clinical diagnosis.

[0006] To achieve the above-mentioned technical effects, the present invention is implemented through the following technical solution: The Granger causal analysis method based on pathogenic microbial metagenomics is based on the following principle: cluster analysis is performed on the acquired pathogenic microbial metagenomic data to classify each species, the species abundance of each species at the corresponding time point of disease development is calculated, time series of each variable is established for each time point and corrected, and finally Granger causal analysis is performed on the corrected data to find the main pathogens. The specific steps are as follows: S1: Nucleic acid extraction from pathogenic microorganism samples; S2: Next-generation high-throughput sequencing of nucleic acids of pathogenic microorganisms; S3: Assembly and comparison of raw sequencing data; S4: The data for each bacterial species is divided according to the principle of OTU clustering analysis. OTU clustering is obtained by comparing data from the pathogenic microorganism gene database. S5: Statistically analyze the time points and corresponding species abundance of each data set during the disease development process, and establish time series data for each species abundance; S6: Use Matlab to test the stationarity of the time series of each data set and select several data sets with relatively stable time series. S7: Use a second-order autoregressive model to perform Granger causality tests on the selected data to obtain the corresponding δ value for each group of data; S8: Predict the main pathogens based on the comparison of δ values.

[0007] The beneficial effects of this invention are: it utilizes a high-throughput sequencing platform to identify multiple bacteria, and then combines this with Granger causal analysis to find the causal relationships between these bacteria, providing a rapid identification method for finding one or more pathogenic bacteria from pathogenic microorganism samples, thus assisting clinicians in conducting analysis and judgment more quickly to achieve clinical diagnosis; the method provided by this invention reveals the interrelationships between pathogenic microorganism populations from a statistical perspective, and identifies major pathogenic microorganisms, and the test report generated by combining theoretical analysis with the final experimental results can have higher reliability. Attached Figure Description

[0008] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a flowchart of the Granger causal analysis method among multiple bacterial communities of the present invention. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0011] like Figure 1 As shown, this invention performs cluster analysis on the acquired pathogenic microbial metagenomic data and classifies each species, calculates the species abundance of each species at the corresponding time point in disease development, then establishes a time series of each group of variables for each time point and corrects it, and finally performs Granger causality analysis on the corrected data to find the main pathogens. The specific steps are as follows: Step 1: The obtained pathogenic microorganism sample was disrupted using a surfactant method; the disrupted sample was extracted using a centrifugal column extraction method; and the extracted nucleic acid was purified using a magnetic bead separation method to obtain high-purity pathogenic microorganism nucleic acid.

[0012] Step 2: Use a second-generation high-throughput sequencer to sequence the high-purity nucleic acid of pathogenic microorganisms; Step 3: Use FLASH to splice and compare the sequencing results to obtain the target gene sequence; Step 4: Use SeekDeep to perform OUT cluster analysis on the target gene sequence to obtain multiple bacterial species; Step 5: Analyze the time points and corresponding species abundance of each data set during the disease development process, establish time series data for each species abundance, and set time series parameters based on the characteristics of the microbial sample data. ,in This corresponds to the sampling time point of the nth sample. Using the time axis T as the x-axis, we solve for the independent variable X, attempting to find the relationship between the time axis T and the independent variable X. ; Step 6: Import the time series data of each species abundance into Matlab. Matlab will test the time series stationarity of each set of data and select several sets of data with relatively stable time series. Step 7: Perform a Granger causality test on the selected data using a second-order autoregressive model to obtain the corresponding δ value for each data group. The Granger causality test uses the past 'a' points of the time series to solve the regression equation for the current point, where 'a' is the order of the regression equation, i.e., lag. Next, the time range T:lag... Predict the point on [lag+1]. , Error And so on, over a time range T:(n-lag,…,n-1) Predict points on [n] , Error After obtaining a series of errors, the unbiased estimation method is used to solve for the unbiased estimation error δ generated by the joint regression. Step 8: Determine the causal variable X and the outcome variable Y by comparing the δ values ​​of the two time series, and predict the main pathogenic bacteria based on the comparison of the δ values.

[0013] In the sixth step above, given the uncertainty of the sampling time point of the gene data, it is necessary to test the time series stationarity of the sample and correct the data that does not meet the requirements of Granger causality analysis when necessary.

[0014] In step seven above, based on the characteristics of metagenomic data, the choice of the order of the regression equation will affect the accuracy of Granger causality analysis. Given a sufficiently large amount of data, a higher order is generally better; however, a higher order places higher demands on computational performance, thus requiring a reasonable selection.

[0015] Terminology Explanation

Claims

1. A Granger causal analysis method for multiple bacterial communities based on pathogenic microbial metagenomics, characterized in that, Specifically, the following steps are included: S1: Nucleic acid extraction from pathogenic microorganism samples; S2: Next-generation high-throughput sequencing of nucleic acids of pathogenic microorganisms; S3: Assembly and comparison of raw sequencing data; S4: The data for each bacterial species is divided according to the principle of OTU clustering analysis. OTU clustering is obtained by comparing data from the pathogenic microorganism gene database. S5: Develop time series data tailored to the characteristics of microbial sample data. ,in This corresponds to the sampling time point of the nth sample. Using the time axis T as the x-axis, we solve for the independent variable X, attempting to find the relationship between the time axis T and the independent variable X. ; Statistically analyze the time points and corresponding species abundance of each data set during the disease development process, and establish time series data for each species abundance; S6: Use Matlab to test the stationarity of the time series of each data set and select several data sets with relatively stable time series. S7: Perform a Granger causality test on the selected data using a second-order autoregressive model to obtain the corresponding δ value for each data group; the Granger causality test uses the past 'a' points of the time series to solve the regression equation for the current point, where 'a' is the order of the regression equation, i.e., lag. Next, the time range T:lag... Predict the point on [lag+1]. , Error And so on, over a time range T:(n-lag,…,n-1) Predict points on [n] , Error After obtaining a series of errors, the unbiased estimation method is used to solve for the unbiased estimation error δ generated by the joint regression. S8: Predict the main pathogens based on the comparison of δ values.

2. The Granger causality analysis method for multiple bacterial communities based on pathogenic microbial metagenomics according to claim 1, characterized in that, In step S1, the nucleic acid extraction of the pathogenic microorganism sample includes crushing, extraction, and purification.

3. The Granger causality analysis method for multiple bacterial communities based on pathogenic microbial metagenomics according to claim 1, characterized in that, In step S3, the splicing and comparison of the raw sequencing data are performed in FLASH software.

4. The Granger causality analysis method for multiple bacterial communities based on pathogenic microbial metagenomics according to claim 1, characterized in that, In step S4, the data for each bacterial species are subjected to OUT clustering analysis using SeekDeep software.

5. The Granger causality analysis method for multiple bacterial communities based on pathogenic microbial metagenomics according to claim 2, characterized in that, The pathogenic microorganism samples were broken up using a surfactant method.

6. The Granger causality analysis method for multiple bacterial communities based on pathogenic microbial metagenomics according to claim 2, characterized in that, The pathogenic microorganisms were extracted using a centrifugal column extraction method after the samples were broken up.

7. The Granger causality analysis method for multiple bacterial communities based on pathogenic microbial metagenomics according to claim 2, characterized in that, The pathogenic microorganism samples were purified using magnetic bead separation after extraction.

Images