Respiratory tract colonization and infectious pneumonia diagnosis biomarker group and application thereof

Abstract

The invention relates to an opportunistic pathogenic bacterium infection diagnosis biomarker group and application thereof, a diagnosis product and a diagnosis method, and relates to the fields of opportunistic pathogenic bacterium infection diagnosis technology and precision medical treatment. The biomarker group comprises an mRNA (messenger ribonucleic acid) of an ANKRD52 protein, an mRNA of a ZC3HAV1L protein, an mRNA of an SERPINE3 protein, an mRNA of a CDPF1 protein, an mRNA of a ZNF720 protein, an mRNA of a TAGLN3 protein and an mRNA of an LRRC15 protein. The diagnostic product comprises a reagent for detecting the mRNA level of each marker in the biomarker group. The diagnosis method comprises the following steps: acquiring the RPM value of microorganisms in a sample to be detected and the mRNA expression level (TPM value) of the seven proteins as input characteristics, and judging the diagnosis result of the sample to be detected according to the output result of the opportunistic pathogenic bacterium infection random forest algorithm diagnosis model.

Description

Technical Field

[0001] The present invention relates to the field of diagnostic technology, and in particular to a biomarker panel for diagnosing respiratory colonization and infectious pneumonia, its application, diagnostic products, and diagnostic methods. Background Art

[0002] Colonizing microorganisms, or colonizers, are colonies that reside in the host's body for a long time. They reside on the surfaces of various body parts, relying on nutrients provided by the host for growth and reproduction without triggering an inflammatory response in the host. Common colonization areas include the intestines, skin, and lungs. Infectious pathogens invade the host's body and trigger an inflammatory response in that area. Some microorganisms are opportunistic pathogens, meaning they are colonizers most of the time. However, when a patient becomes immunocompromised, the colonizers begin to multiply uncontrollably and invade the host's body, causing infection and disease.

[0003] While traditional microbial detection methods (such as culture and PCR) and existing metagenomic sequencing (mNGS) technologies can detect the presence of pathogens, they struggle to distinguish between colonization and infection, making it difficult to accurately determine the need for antibiotic treatment in clinical diagnosis. This limitation leads to the overuse of antibiotics and misdiagnosis of disease conditions in clinical practice.

[0004] The patent "A method, device and storage medium for distinguishing between colonization and infection of opportunistic pathogens" (application number: 202211503296.X) quickly distinguishes the colonization status and infection status of pathogens by analyzing the microbial composition in respiratory samples. This patent uses mNGS technology to perform high-throughput sequencing of microbial nucleic acids in respiratory samples to analyze the types of microorganisms in the samples and their relative abundance. After removing the host nucleic acid sequence, the remaining sequence is compared with the pathogen metagenomic database to obtain the relative abundance information of the microorganism. By calculating the number of reads per million (RPM) of the microorganism and comparing it with the preset RPM threshold, it can be determined whether the microorganism is in a colonization or infection state.

[0005] Although the method in the above patent provides a method for quickly distinguishing colonization from infection through metagenomic sequencing and RPM threshold analysis, this method mainly relies on metagenomic sequencing and RPM threshold analysis, and only relies on a single indicator of microbial abundance, lacks a comprehensive analysis of the host immune response, and the technical means are relatively simple, and the ability to analyze complex samples is limited, which can easily lead to insufficient diagnostic accuracy.

[0006] In summary, the following technical problems currently exist in this field: (1) Traditional methods and existing patents rely solely on microbial abundance or a single indicator, which cannot fully reflect the interaction between pathogens and hosts, resulting in insufficient diagnostic accuracy; (2) Existing technologies mainly focus on the abundance of pathogens and lack in-depth analysis of host immune responses, and cannot fully reflect the interaction mechanism between pathogens and hosts; (3) Existing technologies cannot accurately distinguish between colonization and infection, leading to increased abuse of antibiotics and drug resistance.

[0007] Therefore, in response to the above problems, there is an urgent need to develop effective diagnostic methods and / or models in this field to solve the problem that the existing technology cannot accurately distinguish between respiratory pathogen colonization and infection.

[0008] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute prior art known to ordinary technicians in this field. Summary of the Invention

[0009] The goal of this invention is to develop a highly accurate diagnostic model that can quickly and accurately distinguish between respiratory colonization and infectious pneumonia by comprehensively analyzing the microbial composition and host immune response in respiratory samples.

[0010] In the first aspect of the present invention, a biomarker panel for diagnosing opportunistic pathogen infection is provided, which includes mRNA of ANKRD52 protein, mRNA of ZC3HAV1L protein, mRNA of SERPINE3 protein, mRNA of CDPF1 protein, mRNA of ZNF720 protein, mRNA of TAGLN3 protein and mRNA of LRRC15 protein.

[0011] In a second aspect of the present invention, there is provided use of the biomarker panel described in the first aspect of the present invention in diagnosing opportunistic pathogen infection.

[0012] In a third aspect of the present invention, a product for diagnosing opportunistic pathogen infection is provided, comprising a detection reagent for detecting the level of mRNA of each protein in the biomarker panel described in the first aspect of the present invention.

[0013] In another preferred embodiment, the product is a kit, a detection reagent or a biochip.

[0014] In another preferred embodiment, the kit, detection reagent or biochip further comprises a label or instructions, wherein the label or instructions indicate that the kit, detection reagent or biochip is used for diagnosing opportunistic pathogen infection.

[0015] In another preferred embodiment, the detection reagent is a reagent used to detect the level of the biomarker panel described in the first aspect of the present invention using high-throughput sequencing technology.

[0016] In a fourth aspect of the present invention, a method for diagnosing opportunistic pathogen infection is provided, comprising the following steps:

[0017] S1. Obtaining a sample to be tested, wherein the sample to be tested is taken from a patient suspected of having opportunistic pathogens;

[0018] S2. Using the opportunistic pathogen infection diagnostic product described in the third aspect of the present invention, detecting the expression levels of ANKRD52 protein mRNA, ZC3HAV1L protein mRNA, SERPINE3 protein mRNA, CDPF1 protein mRNA, ZNF720 protein mRNA, TAGLN3 protein mRNA, and LRRC15 protein mRNA in the test sample to obtain the mRNA expression result of the test sample; the mRNA expression result is the TPM value;

[0019] S3, obtaining the microbial RPM value in the microbial metagenomic sequencing results of the sample to be tested;

[0020] S4. Input the TPM value of the mRNA expression result of the test sample obtained in step S2 and the RPM value of the microorganism in the microbial metagenomic sequencing result of the test sample obtained in step S3 into the random forest algorithm diagnosis model for opportunistic pathogen infection to obtain the output result of the random forest algorithm diagnosis model for opportunistic pathogen infection; based on the above random forest algorithm diagnosis model that is particularly suitable for solving binary classification problems, perform a two-class prediction and judgment of "infection" and "colonization" on the test sample;

[0021] The order of the above steps S2 and S3 can be interchanged.

[0022] In another preferred embodiment, in step S1, the sample to be tested is selected from the following group: sputum sample, bronchoalveolar lavage fluid (BALF) sample.

[0023] In another preferred embodiment, the construction process of the random forest algorithm diagnostic model for opportunistic pathogen infection in step S3 is:

[0024] (i) Obtaining samples from subjects infected with opportunistic pathogens, and dividing the subjects into two experimental groups, the experimental groups comprising: an opportunistic pathogen infection group (IP group) and an opportunistic pathogen colonization group (CZ group);

[0025] (ii) obtaining differential microbial RPM values from the microbial metagenomic sequencing results of each sample in each experimental group to form dataset 1;

[0026] (iii) Obtain the TPM values of differentially expressed host gene DEGs in each sample in each experimental group to form dataset 2;

[0027] (iv) Combining Dataset 1 in (ii) above and Dataset 2 in (iii) above, a random forest algorithm diagnostic model for opportunistic pathogen infection was constructed using a machine learning algorithm; model performance was evaluated through 10-fold cross-validation, and model parameters were optimized;

[0028] The order of the above steps (ii) and (iii) can be interchanged.

[0029] In another preferred embodiment, the process of constructing the random forest algorithm diagnostic model for opportunistic pathogen infection in step S3 further includes the following steps:

[0030] (vi) The diagnostic model was validated using independent sample sets (including colonization and infection groups) to evaluate the accuracy and robustness of the model.

[0031] In another preferred embodiment, the data in the independent sample set includes:

[0032] (1) 10%-30% of the data in Dataset 1 consisting of the differential microbial RPM values in the microbial metagenomic sequencing results of each sample in each experimental group, and,

[0033] (2) The TPM values of the differentially expressed host gene DEGs in each sample in each experimental group constitute 10%-30% of the data in dataset 2.

[0034] In a fifth aspect of the present invention, a method for constructing a random forest algorithm diagnostic model for opportunistic pathogen infection is provided, comprising the following steps:

[0035] (i) Obtaining samples from subjects infected with opportunistic pathogens, and dividing the subjects into two experimental groups, the experimental groups comprising: an opportunistic pathogen infection group (IP group) and an opportunistic pathogen colonization group (CZ group);

[0036] (ii) obtaining differential microbial RPM values from the microbial metagenomic sequencing results of each sample in each experimental group to form dataset 1;

[0037] (iii) Obtain the TPM values of differentially expressed host gene DEGs in each sample in each experimental group to form dataset 2;

[0038] (iv) Combining Dataset 1 in (ii) above and Dataset 2 in (iii) above, a random forest algorithm diagnostic model for opportunistic pathogen infection was constructed using a machine learning algorithm; model performance was evaluated through 10-fold cross-validation, and model parameters were optimized;

[0039] The order of the above steps (ii) and (iii) can be interchanged.

[0040] In another preferred embodiment, the method for constructing the random forest algorithm diagnostic model for opportunistic pathogen infection further comprises the steps of:

[0041] (vi) The diagnostic model was validated using independent sample sets (including colonization and infection groups) to evaluate the accuracy and robustness of the model.

[0042] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings are incorporated into and constitute a part of this specification, illustrate embodiments consistent with the present invention, and together with the description, serve to explain the principles of the present invention. Obviously, the drawings described below are only some embodiments of the present invention, and it is clear that those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0044] Figure 1 This figure shows the ranking of gene importance in the random forest model, with importance values on the horizontal axis and gene names on the vertical axis. ANKRD52 has the highest importance, followed by ZC3HAV1L and SERPINE3, and LRRC15 has the lowest importance. Importance values range from 0 to 3, reflecting the relative importance of each gene in the model.

[0045] Figure 2 The ROC curve and its corresponding AUC value for the classification model are shown. The horizontal axis represents 1-specificity (false positive rate), and the vertical axis represents sensitivity (true positive rate). The red curve represents the ROC curve for the training set, with an AUC value of 0.951 (95% confidence interval 0.934–0.968). The blue curve represents the ROC curve for the validation set, with an AUC value of 0.875.

[0046] Figure 3 contains a volcano plot (Volcano Plot) ( Figure 3A ) and two pathway enrichment analysis diagrams ( Figure 3B and Figure 3C). The volcano plot shows the significant changes in gene expression, among which the significantly up-regulated genes (red, 894 genes) and significantly down-regulated genes (blue, 980 genes) meet the conditions of p < 0.05 and log2FC > 1 or < -1, respectively. Non-significant genes are represented in gray. Some genes (such as CXCL8, IL1B, etc.) are annotated in the figure. The pathway enrichment analysis diagram shows the enrichment of the MAPK signaling pathway and the Wnt signaling pathway, including enrichment curves, NES values (MAPK is 1.389, Wnt is 1.545), p-value (both 0.002) and p.adjust (both 0.0182). DETAILED DESCRIPTION

[0047] Through extensive and in-depth research, the inventors have integrated three key technologies: metagenomic sequencing (mNGS), host transcriptomic analysis, and diagnostic model construction to construct a diagnostic model based on metagenomics (mNGS) and host transcriptomic analysis to distinguish between respiratory colonization and infectious pneumonia. The core of this invention is to develop a highly accurate diagnostic tool by analyzing the microbial composition in respiratory samples and the gene expression characteristics of the host immune response to quickly and accurately determine whether the patient is in a pathogen colonization state or an infection state. . The present invention was completed on this basis.

[0048] Specifically, the present invention combines metagenomic sequencing (mNGS) and host transcriptomic analysis, while considering the gene expression characteristics of microbial composition and host immune response, to improve diagnostic accuracy through multi-dimensional data analysis. For example, the diagnostic model of the present invention integrates microbial and host gene expression characteristics through a random forest algorithm, with an AUC of 0.951 for the training set and 0.875 for the validation set, which is significantly higher than the accuracy of single-metric analysis.

[0049] The present invention uses host transcriptomics analysis to detect gene expression pathways related to tissue repair and immune regulation (such as the Wnt signaling pathway and the MAPK signaling pathway), and combines this with microbial composition data to comprehensively reflect the interaction between pathogens and hosts. For example, the present invention found that the Wnt signaling pathway is significantly upregulated in the colonized state, while the transcription levels of microbial ribosomal subunits are significantly increased in the infected state. This difference provides a theoretical basis for accurate diagnosis.

[0050] The present invention uses a highly accurate diagnostic model to precisely distinguish between colonization and infection, avoiding unnecessary antibiotic treatment for colonized bacteria and reducing antibiotic abuse. For example, by analyzing microbial diversity and host gene expression characteristics, the present invention found that microbial diversity was reduced in the infected group, while the Wnt signaling pathway was significantly upregulated in the colonized group. These characteristics can be used for precise diagnosis.

[0051] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be embodied in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete and will fully convey the concept of the example embodiments to those skilled in the art.

[0052] Unless otherwise defined, all terms and phrases used herein include the meanings that the terms and phrases have in the art, unless explicitly stated to the contrary or clearly apparent from the context in which the terms and phrases are used. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, specific methods and materials are now described.

[0053] the term

[0054] Random forest algorithm diagnosis model for opportunistic pathogen infection

[0055] The present invention combines metagenomic sequencing (mNGS) and host transcriptomic analysis to simultaneously consider gene expression signatures of microbial composition and host immune response.

[0056] In a specific embodiment of the present invention, the construction process of the random forest algorithm diagnostic model for opportunistic pathogen infection is as follows:

[0057] (i) Obtaining samples from subjects infected with opportunistic pathogens, and dividing the subjects into two experimental groups, the experimental groups comprising: an opportunistic pathogen infection group (IP group) and an opportunistic pathogen colonization group (CZ group);

[0058] (ii) obtaining differential microbial RPM values from the microbial metagenomic sequencing results of each sample in each experimental group to form dataset 1;

[0059] (iii) Obtain the TPM values of differentially expressed host gene DEGs in each sample in each experimental group to form dataset 2;

[0060] (iv) Combining Dataset 1 in (ii) above and Dataset 2 in (iii) above, a random forest algorithm diagnostic model for opportunistic pathogen infection was constructed using a machine learning algorithm; model performance was evaluated through 10-fold cross-validation, and model parameters were optimized;

[0061] The order of the above steps (ii) and (iii) can be interchanged.

[0062] Microbial RPM value

[0063] Microbial RPM-normalized reads (READS) refers to the number of reads per million (RPM) of total sequencing reads in high-throughput sequencing data analysis. RPM normalization is a commonly used normalization method used to eliminate differences in sequencing depth (i.e., total amount of sequencing) between samples, thereby making microbial abundance or gene expression levels comparable across samples.

[0064] Specifically, the RPM normalized reading value is calculated using the following formula:

[0065] RPM = number of reads of target gene or microorganism / total sequencing reads (the sum of reads of all genes or microorganisms) × 10 6

[0066] where the denominator is the sum of all sequencing reads in the sample, multiplied by 10 6 This is to adjust the results to the level of reads per million to facilitate comparison across samples.

[0067] Host gene expression level TPM

[0068] TPM (Transcripts Per Million) is a commonly used method for normalizing gene expression, particularly in RNA sequencing (RNA-seq) data analysis. It provides a fairer way to compare gene expression by simultaneously accounting for the effects of sequencing depth and gene length.

[0069] Definition and calculation of TPM

[0070] TPM represents the ratio of each gene's expression level to the total expression level of all genes in the sample and is normalized in units of transcripts per million. The calculation formula for TPM is as follows:

[0071]

[0072] READS i is the number of sequencing reads (READS) of gene i.

[0073] Gene Length i is the length of gene i (usually in base pairs)

[0074] · All genes sum.

[0075] 10 6 The results were normalized to the ratio per million transcripts.

[0076] Different microorganisms

[0077] Differential microorganisms refer to microorganisms whose composition, structure or function of microbial communities are significantly different under different conditions.

[0078] In the present invention, the composition, structure, or function of the microbial communities in the opportunistic pathogen infection group (IP group) and the opportunistic pathogen colonization group (CZ group) are significantly different, and the microbial metagenomic sequencing results can fully reflect these differences. The differential microbial RPM values from the microbial metagenomic sequencing results for each sample in each experimental group (IP group and CZ group) are obtained to form Dataset 1.

[0079] Differential host gene DEGs

[0080] Differential host genes refer to genes whose expression levels are significantly different under different conditions.

[0081] In the present invention, there are significant differences in the expression levels of host genes in the opportunistic pathogen infection group (IP group) and the opportunistic pathogen colonization group (CZ group). The host transcriptomic analysis results can fully reflect the differences between the two groups. The TPM values of the differentially expressed host gene DEGs in each sample in each experimental group (IP group and CZ group) are obtained to form Dataset 2.

[0082] method

[0083] This invention integrates three key technologies: metagenomic sequencing (mNGS), host transcriptomics analysis, and diagnostic model construction. Specifically:

[0084] (1) Metagenomic sequencing (mNGS): Through high-throughput sequencing of microbial nucleic acids in respiratory samples, a comprehensive analysis of the microbial species and their relative abundance in the samples is conducted. mNGS can detect pathogens that are difficult to detect with traditional methods and can provide information on microbial activity, such as the transcription level of ribosomal subunits.

[0085] (2) Host transcriptomic analysis: Analyze the host's immune response to pathogens by detecting changes in host cell gene expression. This study specifically focuses on gene expression pathways related to tissue repair and immune regulation, such as the Wnt signaling pathway and the MAPK signaling pathway. Upregulation of these pathways is closely related to pathogen colonization, while infection results in distinct gene expression patterns.

[0086] (3) Diagnostic model construction: Combining microbial composition (such as differential microbial RPM values) and host gene expression characteristics (such as differential host gene DEG TPM values), a diagnostic model is constructed using a machine learning algorithm (such as random forest). This model can quickly distinguish between colonization and infection states based on the comprehensive characteristics of microbial and host gene expression.

[0087] Typically, the present invention mainly achieves the purpose of distinguishing respiratory colonization from infection through three major technical means: multi-dimensional data analysis, high-throughput sequencing technology, and machine learning algorithms.

[0088] (1) Multidimensional data analysis: Traditional methods and existing technologies mainly rely on single indicators (such as microbial abundance or gene expression levels), while the present invention combines multidimensional data of microbial composition and host immune response to comprehensively reflect the interaction between pathogens and hosts;

[0089] (2) High-throughput sequencing technology: mNGS technology can detect pathogens that are difficult to detect with traditional methods and provide information on the activity of microorganisms. At the same time, host transcriptomic analysis can reveal the host's immune response to pathogens, providing richer information for diagnosis;

[0090] (3) Machine learning algorithm: A diagnostic model is constructed using the random forest algorithm to screen key biomarkers and optimize model parameters. This data-driven model can automatically learn the relationship between sample characteristics and colonization / infection status, improving the accuracy and robustness of diagnosis.

[0091] In one embodiment of the present invention, a method for diagnosing pathogenic bacteria infection is provided, comprising the following steps:

[0092] S1. Obtaining a sample to be tested, wherein the sample to be tested is taken from a patient suspected of having opportunistic pathogens;

[0093] S2. Using the opportunistic pathogen infection diagnostic product of the present invention, detecting the expression levels of ANKRD52 protein mRNA, ZC3HAV1L protein mRNA, SERPINE3 protein mRNA, CDPF1 protein mRNA, ZNF720 protein mRNA, TAGLN3 protein mRNA, and LRRC15 protein mRNA in the test sample to obtain the mRNA expression result of the test sample; the mRNA expression result is the TPM value;

[0094] S3, obtaining the RPM value of the microbial metagenomic sequencing results in the sample to be tested;

[0095] S4. Input the mRNA expression result TPM value of the test sample obtained in step S2 and the RPM value of the microbial metagenomic sequencing result of the test sample obtained in step S3 into the random forest algorithm diagnosis model for opportunistic pathogen infection to obtain the output result of the random forest algorithm diagnosis model for opportunistic pathogen infection; based on the above random forest algorithm diagnosis model that is particularly suitable for solving binary classification problems, perform a two-class prediction and judgment of "infection" and "colonization" on the test sample;

[0096] The order of the above steps S2 and S3 can be interchanged.

[0097] Typically, the mRNA expression result is a sequencing result (eg, a TPM value).

[0098] use

[0099] The application of the present invention is mainly reflected in the following aspects:

[0100] Clinical diagnosis: Quickly and accurately determine whether the patient is infected with pathogens, avoid unnecessary antibiotic treatment for colonized bacteria, and reduce antibiotic abuse and the development of drug resistance.

[0101] Disease monitoring: Through dynamic monitoring of respiratory microbial communities, potential infection risks can be discovered in a timely manner, providing a basis for early intervention.

[0102] Research tools: Provide a technical platform for studying the interaction between respiratory microbial communities and host immune responses, which will help to gain a deeper understanding of the pathogenesis of respiratory diseases such as pneumonia.

[0103] The random forest algorithm diagnostic model for pathogenic bacterial infection constructed by the present invention only requires the RPM of the microbial metagenomic sequencing results of deep sputum or alveolar lavage fluid samples and the TPM of seven biomarkers (ANKRD52, ZC3HAV1L, SERPINE3, CDPF1, ZNF720, TAGLN3 and LRRC15) in the human transcriptome sequencing results. It can be used to predict whether the respiratory microorganisms in outpatients and inpatients (including ICU patients) are in a colonization or infection state, providing guidance for doctors for subsequent treatment and rational use of drugs.

[0104] The beneficial effects of the present invention mainly include:

[0105] 1. Significantly improve diagnostic accuracy

[0106] Limitations of existing technologies: Traditional diagnostic methods (such as culture and PCR) and existing patents (such as BGI's RPM threshold analysis) rely solely on a single indicator of microbial abundance, which makes it difficult to fully reflect the interaction between pathogens and hosts, resulting in insufficient diagnostic accuracy.

[0107] Effect of the present invention: The present invention significantly improves the diagnostic accuracy by combining multi-dimensional data analysis of microbial composition and host immune response. Experimental results ( Figure 2 ) showed that the AUC value of the diagnostic model of the present invention on the training set reached 0.951 (95% confidence interval: 0.934-0.968), and the AUC value on the independent validation set was 0.875.

[0108] In comparison, the AUC values of existing technologies (such as RPM threshold analysis) are generally between 0.7 and 0.8, and the diagnostic accuracy of the present invention is significantly higher.

[0109] 2. Comprehensively reflect the interaction between pathogens and hosts

[0110] Limitations of existing technologies: Existing technologies mainly focus on the abundance of pathogens, lack in-depth analysis of host immune responses, and cannot fully reflect the interaction between pathogens and hosts.

[0111] Effects of the Invention: The present invention uses host transcriptomic analysis to detect gene expression signatures related to tissue repair (such as the Wnt signaling pathway) and immune regulation (such as the MAPK signaling pathway). Experimental data (Figure 3) show that the Wnt signaling pathway is significantly upregulated in the colonized state and downregulated in the infected state.

[0112] This multidimensional analysis method can not only distinguish between colonization and infection states, but also provide a new perspective for studying the interaction mechanism between pathogens and hosts.

[0113] 3. Discovery and application of key biomarkers

[0114] Limitations of existing technologies: Existing technologies lack accurate biomarkers and are unable to provide reliable diagnostic basis.

[0115] Effects of the Invention: Using machine learning algorithms, the present invention screened a series of key biomarkers, including ANKRD52, ZC3HAV1L, SERPINE3, CDPF1, ZNF720, TAGLN3, and LRRC15. These markers exhibited significant differences in expression between colonization and infection, providing new targets for clinical diagnosis.

[0116] The data showed that the expression levels of these biomarkers were highly correlated with the pathogen status, significantly improving the accuracy and reliability of diagnosis.

[0117] Compared with BGI's patent "A method, device and storage medium for distinguishing colonization and infection of opportunistic pathogens" (application number: 202211503296.X), which mainly relies on metagenomic sequencing and RPM threshold analysis, the present invention combines metagenomic sequencing and host transcriptomics analysis. By comprehensively considering the microbial composition and host immune response, it can more comprehensively reflect the pathogen status and host response, and has higher diagnostic accuracy.

[0118] This method not only enables rapid differentiation between colonization and infection, but also provides deeper technical support for studying host-pathogen interactions, contributing to the understanding of disease mechanisms. Furthermore, by combining host transcriptomics analysis, a comprehensive diagnostic model has been developed that can more accurately determine pathogen status and provide new insights into disease mechanisms.

[0119] The biomarker panel for distinguishing colonization and infection of opportunistic pathogens provided by the present invention is further described below with reference to specific examples.

[0120] It should be understood that these examples are only used to illustrate the present invention and are not intended to limit the scope of the present invention. The experimental methods for which detailed conditions are not specified in the following examples are generally based on conventional conditions such as those described in "Molecular Cloning Laboratory Manual" (translated by Huang Peitang et al., Beijing: Science Press, 2002) by Sambrook, J. et al., USA, or according to the conditions recommended by the manufacturer (e.g., product instructions). Unless otherwise stated, percentages and parts are calculated by weight. The experimental materials and reagents used in the following examples are all available from commercial sources or can be prepared according to literature methods unless otherwise specified.

[0121] Example 1 Screening of detection markers (groups), model construction and effect verification of the present invention

[0122] 1. Sample Collection and Processing

[0123] The subjects were classified into the infection group (IP group) and the colonization group (CZ group) according to the imaging and laboratory criteria for infectious lung lesions.

[0124] Inclusion criteria for the IP group included tracheotomy or endotracheal intubation due to treatment needs, sputum specimen collection and chest CT on the same day, and chest imaging showing patchy or patchy infiltration or interstitial changes in the lungs, possibly accompanied by other symptoms such as pleural effusion; exclusion criteria were the presence of potential or confirmed infection other than lung infection, difficulty in obtaining deep respiratory secretions, inconsistency between lung imaging manifestations and clinical symptoms and signs, and lack of sufficient evidence to support infection.

[0125] The inclusion criteria for the CZ group included tracheotomy or endotracheal intubation due to treatment needs, sputum specimen collection and chest CT performed on the same day, and no pneumonia shown in the CT report; the exclusion criteria were the same as those for the IP group.

[0126] From December 2020 to April 2022, the inventors enrolled 17 patients with colonization and 27 patients with infectious pneumonia according to the inclusion criteria. Deep sputum samples were collected from these patients for metagenomic transcriptome and metagenomic sequencing. These deep sputum samples can better reflect the microbial composition of the lower respiratory tract and the host immune response.

[0127] Sample processing: Samples need to follow the DNA and RNA extraction processes required by the sequencing platform (such as BGI or Ilumina) to prepare for metagenomic sequencing and transcriptomic analysis.

[0128] 2. Metagenomic sequencing (mNGS)

[0129] Technical principle: Through high-throughput sequencing technology, the whole genome of microbial nucleic acids in the sample is sequenced to obtain information on the type, abundance and activity of microorganisms.

[0130] In the metagenomic sequencing protocol, 1 ml of sputum sample was taken and treated with 1 unit of benzonase and 0.5% Tween 20 to remove host nucleic acids. Subsequently, the solution was placed in an incubator at 37 ° C and incubated for 5 minutes. After the incubation, 600 microliters of solution were taken and mechanically broken (bead beating) in a Minilys Personal TGrinder H24 homogenizer with ceramic beads. Then, nucleic acids were extracted from 400 microliters of the broken sample using the QIAamp UCP Pathogen Mini Kit (QIAamp UCP Pathogen Mini Kit, Qiagen, cat.50214), and their concentrations were determined by the Qubit dsDNA High Sensitivity Detection Kit (Qubit dsDNA HS Assay Kit, Thermo Fisher Scientific, cat.Q32854). Thirty microliters of extracted DNA was used to construct sequencing libraries using the Nextera DNA Flex Kit (Illumina, cat. 20018704). Library concentration was determined using the Qubit dsDNA High Sensitivity Detection Kit, and library integrity and quality were assessed using an Agilent 2100 Bioanalyzer and a High Sensitivity DNA kit (Thermo Fisher Scientific, cat. Q32854). Finally, the libraries were sequenced on the Illumina NextSeq 550 sequencing platform using a 75-cycle single-end sequencing strategy.

[0131] After sequencing, fastp software was used to check the quality of the sequencing results and remove low-quality alignments. Fastp software was used for data filtering in this analysis, with the following parameters: remove reads less than 15 bp in length using the Nextera DNAFlex kit (Illumina, cat. 20018704), remove reads with a low-quality base ratio greater than 40%, and remove reads with more than 5 N bases.

[0132] Removal of human RNA or cDNA: The obtained high-quality reads were aligned to the human genome (GRCh38) using HTSAT2 (version 2.2.1), and any reads that were not aligned were removed.

[0133] The remaining reads were aligned using Kraken to the Pathogens Metagenomics Database (PMDB), which consists of genomes of bacteria, fungi, viruses, and parasites downloaded from NCBI (ftp: / / ftp.ncbi.nlm.nih.gov / genomes / ). Read counts were obtained for all microorganisms during the alignment. Two differentially expressed microorganisms were identified using LefSe analysis.

[0134] The RPM values in the sequencing results are used to construct the following model.

[0135] 3. Host Transcriptomics Analysis

[0136] Technical principle: Through high-throughput sequencing technology, the gene expression profile of host cells is analyzed, focusing on gene expression pathways related to immune response and tissue repair.

[0137] The data after sequencing must be standardized.

[0138] In terms of macrotranscriptome sequencing, 1 ml of sputum sample was taken and cell material was collected by centrifugation. Subsequently, the precipitate was lysed, RNA was extracted, and RNA extraction was performed using the Direct-zol RNA Miniprep kit (ZYMO RESEARCH, R2052). 10 microliters were taken from the purified RNA and a sequencing library was constructed using the Ovation Trio RNA-Seq library preparation kit (NuGEN Technologies, 0357-96). The concentration of the library was determined using the Qubit dsDNA High Sensitivity Detection Kit (Qubit dsDNA HS Assay Kit, Thermo Fisher Scientific, cat. Q32854), and the integrity and quality of the library were assessed by an Agilent 2100 Bioanalyzer and a High Sensitivity DNA kit (High Sensitivity DNA kit, Thermo Fisher Scientific, cat. Q32854). Finally, the library was sequenced on the Illumina NextSeq 550 sequencing platform using a 75-cycle single-end sequencing strategy.

[0139] Data analysis: In the field of host gene expression analysis, high-quality datasets were aligned with the human genome reference sequence hg38 using the HISAT2 alignment tool. Subsequently, quantitative analysis was performed at the gene level using the FeatureCounts tool in the Subread software package, which was used to summarize gene counts. Differentially expressed genes (DEGs) were identified using the DESeq2 tool using established thresholds of false discovery rate (FDR) < 0.05 and fold change > 1.5. Gene set enrichment analysis was performed on these differentially expressed genes based on the KEGG database, and significant pathways were identified by Fisher's exact test, and the Benjamin and Hochberg method was used for correction. The sets of immune-related genes and interferon-stimulated genes were obtained from the ImmPort database, respectively, and the references were used. 1,2 The research results were verified.

[0140] Gene set enrichment analysis (GSEA) and weighted gene co-expression network analysis (WGCNA) were performed to identify differentially expressed genes (DEGs) associated with colonization and infection states. For example, the Wnt signaling pathway was significantly upregulated in the colonization state, while inflammation-related pathways (such as the MAPK signaling pathway) were significantly upregulated in the infection state. The results are shown in Figure 3.

[0141] Through analysis, seven key biomarkers were screened out: ANKRD52, ZC3HAV1L, SERPINE3, CDPF1, ZNF720, TAGLN3 and LRRC15. The expression levels of these genes were significantly different between colonization and infection states ( Figure 1 ).

[0142] The TPM values of the above 7 biomarkers in the sequencing results were used as input for the following model construction.

[0143] 4. Diagnostic Model Construction

[0144] Technical principle: Combining microbial composition and host gene expression characteristics, a diagnostic model is constructed using machine learning algorithms (such as random forest).

[0145] Model construction: The differential microbial RPM values and the differential host gene DEG TPM values were used as input features, and the model performance was evaluated through 10-fold cross validation to optimize the model parameters.

[0146] These features formed our initial pool of candidate features. We then employed a random forest algorithm (implemented using the R package "mlr3"), combined with a 10-fold cross-validation (CV) strategy, to evaluate model performance and estimate the 95% confidence interval (CI) for the area under the receiver operating characteristic curve (AUC). All analyses were performed in R version 4.1.0 using RStudio.

[0147] Key biomarkers: The model screened out key biomarkers, such as ANKRD52, ZC3HAV1L, SERPINE3, CDPF1, ZNF720, TAGLN3 and LRRC15, whose expression levels were significantly different between colonization and infection states.

[0148] 5. Diagnostic Model Validation

[0149] Verification method: The above diagnostic model was verified using independent sample sets (including colonization group and infection group) to evaluate the accuracy and robustness of the model.

[0150] The inventors validated the model using a validation set consisting of 2 colonization cases and 4 infection cases, using the random forest algorithm (via the "mlr3" package in R) and implementing a 10-fold cross-validation (CV) strategy to evaluate model performance and estimate the 95% confidence interval (CI) of the area under the receiver operating characteristic curve (AUC).

[0151] Verify the result( Figure 2 ) showed that the AUC value of the training set reached 0.951 (95% confidence interval: 0.934-0.968), and the AUC value of the validation set was 0.875, indicating that the model has high diagnostic accuracy.

[0152] discuss

[0153] 1. Multidimensional data analysis: combining microbial composition with host immune response

[0154] Technical features: The present invention analyzes the microbial composition in respiratory samples through metagenomic sequencing (mNGS), and combines it with host transcriptomics analysis to detect host gene expression characteristics, especially gene expression pathways related to immune response and tissue repair (such as Wnt signaling pathway, MAPK signaling pathway, etc.).

[0155] Analytical Reasoning: This multidimensional data analysis method can fully reflect the interaction between pathogens and hosts, rather than relying solely on microbial abundance or a single indicator. By comprehensively analyzing microbial composition and host responses, this invention can more accurately distinguish between colonization and infection states.

[0156] Corresponding technical issues: Solved the problem that existing technologies (such as BGI’s patent) rely solely on a single indicator of microbial abundance, resulting in insufficient diagnostic accuracy.

[0157] 2. Construction and validation of a high-accuracy diagnostic model

[0158] Technical features: The present invention uses the random forest algorithm to construct a diagnostic model, takes the differential microbial RPM values and the differential host gene DEG TPM values as input features, optimizes the model parameters through 10-fold cross-validation, and screens out key biomarkers (such as ANKRD52, ZC3HAV1L, SERPINE3, CDPF1, ZNF720, TAGLN3 and LRRC15, etc.).

[0159] Analysis and reasoning: The model achieved an AUC of 0.951 on the training set and 0.875 on the independent validation set, significantly exceeding the diagnostic accuracy of existing technologies. This highly accurate diagnostic model is one of the core innovations of this invention.

[0160] Corresponding technical problems: Solve the problem that existing technologies cannot provide highly accurate diagnostic tools, especially the problem of difficulty in distinguishing colonization from infection in complex samples.

[0161] 3. Screening and application of key biomarkers

[0162] Technical Features: This invention uses machine learning algorithms to screen a series of key biomarkers whose expression levels differ significantly between colonization and infection. For example, genes related to the Wnt signaling pathway (such as WNT8B and WNT7A) are significantly upregulated in colonization, while genes related to inflammation (such as CTNNB1 and PIK3R1) are significantly upregulated in infection.

[0163] Analytical reasoning: The discovery of these key biomarkers provides new targets for clinical diagnosis, which can more accurately reflect the pathogen status and host response. By detecting the expression levels of these markers, it is possible to quickly determine whether the patient is in a colonization or infection state.

[0164] Corresponding technical issues: It solves the problem of lack of accurate biomarkers in existing technologies and provides a more reliable basis for clinical diagnosis.

[0165] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the invention being indicated by the appended claims.

[0166] References

[0167] 1. ImmPort: disseminating data to the public for the future of immunology. Immunol Res. 2014 May; 58(2-3): 234-9.

[0168] 2. A diverse range of gene products are effectors of the type I interferon antiviral response. Nature. 2011 Apr 28; 472(7344): 481-5.

Claims

1. A biomarker panel for diagnosing opportunistic pathogen infection, characterized in that: The biomarker panel includes mRNA of ANKRD52 protein, mRNA of ZC3HAV1L protein, mRNA of SERPINE3 protein, mRNA of CDPF1 protein, mRNA of ZNF720 protein, mRNA of TAGLN3 protein and mRNA of LRRC15 protein.

2. Use of the biomarker panel as claimed in claim 1 in the diagnosis of opportunistic pathogen infection.

3. A product for diagnosing opportunistic pathogenic bacteria infection, characterized in that: The invention comprises a detection reagent for detecting the level of mRNA of each protein in the biomarker panel of claim 1.

4. The opportunistic pathogen infection diagnosis product according to claim 3, characterized in that: The product is a kit, a detection reagent or a biochip.

5. The kit, detection reagent or biochip according to claim 4, characterized in that: It also includes a label or instructions, which indicates that the kit, detection reagent or biochip is used for diagnosing opportunistic pathogen infection.

6. The opportunistic pathogen infection diagnostic product according to claim 3, characterized in that: The detection reagent is a reagent used to detect the level of the biomarker group according to claim 1 using high-throughput sequencing technology.

7. A method for diagnosing opportunistic pathogen infection, characterized in that: The following steps are involved: S1. Obtaining a sample to be tested, wherein the sample to be tested is taken from a patient suspected of having opportunistic pathogens; S2. Using the opportunistic pathogen infection diagnostic product according to any one of claims 3 to 6, detecting the expression levels of ANKRD52 protein mRNA, ZC3HAV1L protein mRNA, SERPINE3 protein mRNA, CDPF1 protein mRNA, ZNF720 protein mRNA, TAGLN3 protein mRNA, and LRRC15 protein mRNA in the test sample to obtain the mRNA expression results of the test sample; the mRNA expression results are TPM values; S3, obtaining the microbial RPM value in the microbial metagenomic sequencing results of the sample to be tested; S4. Input the TPM value of the mRNA expression result of the test sample obtained in step S2 and the RPM value of the microorganism in the microbial metagenomic sequencing result of the test sample obtained in step S3 into the random forest algorithm diagnosis model for opportunistic pathogen infection to obtain the output result of the random forest algorithm diagnosis model for opportunistic pathogen infection; based on the above random forest algorithm diagnosis model, which is particularly suitable for solving binary classification problems, perform a two-class prediction and judgment of "infection" and "colonization" on the test sample; The order of the above steps S2 and S3 can be interchanged.

8. The method for diagnosing opportunistic pathogen infection according to claim 7, characterized in that: In step S1, the sample to be tested is selected from the following group: sputum sample, alveolar lavage fluid sample.

9. The method for diagnosing opportunistic pathogen infection according to claim 7, characterized in that: The construction process of the random forest algorithm diagnosis model for opportunistic pathogen infection in step S3 is as follows: (i) Obtaining samples from subjects infected with opportunistic pathogens, and dividing the subjects into two experimental groups, the experimental groups comprising: an opportunistic pathogen infection group (IP group) and an opportunistic pathogen colonization group (CZ group); (ii) obtaining differential microbial RPM values from the microbial metagenomic sequencing results of each sample in each experimental group to form dataset 1; (iii) Obtain the TPM values of differentially expressed host gene DEGs in each sample in each experimental group to form dataset 2; (iv) Combining Dataset 1 in (ii) above and Dataset 2 in (iii) above, a random forest algorithm diagnostic model for opportunistic pathogen infection was constructed using a machine learning algorithm; model performance was evaluated through 10-fold cross-validation, and model parameters were optimized; The order of the above steps (ii) and (iii) can be interchanged.

10. A method for constructing a random forest algorithm diagnostic model for opportunistic pathogen infection, characterized in that: The following steps are involved: (i) obtaining samples from subjects infected with opportunistic pathogens, and dividing the subjects into two experimental groups, the experimental groups comprising: an opportunistic pathogen infection group and an opportunistic pathogen colonization group; (ii) obtaining differential microbial RPM values from the microbial metagenomic sequencing results of each sample in each experimental group to form dataset 1; (iii) Obtain the TPM values of differentially expressed host gene DEGs in each sample in each experimental group to form dataset 2; (iv) Combining Dataset 1 in (ii) above and Dataset 2 in (iii) above, a random forest algorithm diagnostic model for opportunistic pathogen infection was constructed using a machine learning algorithm; model performance was evaluated through 10-fold cross-validation, and model parameters were optimized; The order of the above steps (ii) and (iii) can be interchanged.

Images