Deep analysis method of microbial ribosome proteomics data set and application

Abstract

The invention discloses a deep analysis method and application of a microbial ribosome proteomics data set, and belongs to the technical field of laboratory medicine. The deep analysis method of the microbe ribosome proteomics data set is developed by taking the carbapenem-resistant klebsiella pneumoniae producing KPC enzyme as a specific detection object and taking the toxicity of the carbapenem-resistant klebsiella pneumoniae as a specific analysis phenotype. The method comprises the following steps: dividing carbapenem-resistant klebsiella pneumoniae capable of producing KPC into a non-toxicity group, a middle-toxicity group and a high-toxicity group according to the toxicity level, carrying out mass control analysis and data processing on a protein data set detected by a microbial mass spectrum, and further carrying out matrix analysis and clustering analysis; and continuing to analyze and screen out candidate molecular markers, evaluating the candidate markers of different combinations by combining clinical diagnosis efficacy evaluation indexes, and finally finding that three linked marker mass spectrum peaks can be used for distinguishing bacteria with different virulence levels, thereby forming a ribosomal proteome data analysis method which is large in sample size, high in reproducibility, complete and reliable.

Description

Technical Field

[0001] The present invention relates to a deep analysis method and application of a microbial ribosome proteomics data set, belonging to the technical field of laboratory medicine. Background Art

[0002] Matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) is a new biomass spectrometry technology developed in recent years for microbial identification and typing. It uses the principle that the mass-to-charge ratio (m / z) of ions is proportional to the flight time of ions to analyze ions, determine the molecular weight of samples, and analyze mixed biomacromolecules. It collects microbial protein molecular fingerprints in the range of 2000 to 20000 m / z and compares them with protein spectra in the database to determine the species of bacteria. Compared with immune and nucleic acid-based detection, MALDI-TOF-MS has the characteristics of high sensitivity, high accuracy, and high resolution. In addition, the reagent cost used is relatively low and does not require specially trained personnel to operate. It is a high-throughput screening and identification method that can identify a large number of strains in one run and draw conclusions within minutes. The application of MALDI-TOF-MS in microbial detection and identification is becoming more and more in-depth. Technical guidelines and expert consensus related to mass spectrometry detection have also been published. Mass spectrometry detection technology is becoming more and more mature. In the process of microbial strain identification, a huge amount of ribosomal proteome data has been obtained. However, how to effectively mine the massive data obtained is a technical challenge.

[0003] In clinical anti-infection treatment, bacterial resistance is a global problem. How to detect whether bacteria are resistant through the identification of bacterial species at the same time? In the Chinese invention patent with the publication date of February 3, 2023 and the publication number of CN115678958A, 4 mass spectrometry peaks are used to distinguish CSKP (carbapenem-sensitive Klebsiella pneumoniae) and CRKP (carbapenem-resistant Klebsiella pneumoniae); there are also studies that solve this problem by constructing a KPC-CRKP self-built mass spectrometry library through 40 characteristic peaks. Since 2020, there have been continuous reports in the literature that bacteria such as Klebsiella pneumoniae have a fusion of resistance phenotypes and virulence phenotypes. So whether it is possible to use MALDI-TOF-MS for bacterial species identification and drug resistance detection, while effectively mining ribosomal proteome data to achieve detection including bacterial virulence status, has gradually become a problem that needs to be solved urgently. Summary of the invention

[0004] The first objective of the present invention is to provide a method for in-depth analysis of microbial ribosomal proteomics data sets to solve the problem that the prior art cannot effectively mine microbial ribosomal proteomics data obtained by MALDI-TOF-MS mass spectrometry.

[0005] The second object of the present invention is to provide an application of a deep analysis method of a microbial ribosomal proteomics data set in identifying the virulence level of carbapenem-resistant Klebsiella pneumoniae, providing a fast and simple method for identifying the virulence level of carbapenem-resistant Klebsiella pneumoniae in the prior art.

[0006] In order to achieve the above-mentioned purpose, the technical solution adopted by the in-depth analysis method of a microbial ribosomal proteomics dataset in the present invention is:

[0007] A method for deep analysis of a microbial ribosomal proteomics dataset comprises the following steps:

[0008] S1. Sample pretreatment: The microbial samples to be analyzed are cultured and grouped according to preset phenotypes, and then subjected to matrix-assisted laser desorption ionization time-of-flight mass spectrometry to obtain a ribosomal protein spectrum data set;

[0009] S2. Preprocessing of mass spectrometry data set: preprocessing the ribosomal protein spectrum data set obtained in S1 to obtain a common mass spectrometry data set of microbial samples;

[0010] S3. Basic analysis of mass spectrometry data sets: quality control analysis, population representative mass spectrum peak generation and principal component analysis were performed on the mass spectrometry spectra obtained in S1 and S2 to obtain the population ribosomal protein expression profile of the microbial samples;

[0011] S4. Mass spectrometry dataset difference analysis: Matrix analysis and data processing are performed on the common mass spectrometry dataset of the microbial samples described in S2 to screen out differential mass spectrometry peaks, and hierarchical clustering analysis is performed on the population ribosomal protein expression profiles of the microbial samples described in S3 to output a visualized heat map for further cluster analysis;

[0012] S5. Molecular marker screening of mass spectrometry data set: The differential mass spectrometry peaks obtained in S4 were screened and analyzed to obtain candidate molecular markers. Combination analysis and clinical diagnostic efficacy evaluation of different candidate molecular markers were performed to determine the linked molecular markers for the preset phenotypic grouping of microorganisms.

[0013] The beneficial effect of the above scheme is that the in-depth analysis method of the microbial ribosomal proteomics data set of the present invention is a pioneering invention. The present invention takes carbapenem-resistant Klebsiella pneumoniae as the specific detection object and its virulence as the specific analysis phenotype, and develops a method for in-depth analysis of the microbial ribosomal proteomics data set. The present invention divides carbapenem-resistant Klebsiella pneumoniae into avirulence group, an intermediate virulence group and a high virulence group according to the virulence level, obtains a total of 389 common mass spectrometry peaks after quality control analysis and data processing of the protein data set in the microbial mass spectrometry detection, and further finds 130 differential protein peaks through matrix analysis and cluster analysis, and continues to perform data analysis to screen out 3 joint marker mass spectrometry peaks for distinguishing bacteria with different virulence levels, forming a large sample size, high reproducibility, complete and reliable ribosomal proteomics data analysis method.

[0014] As a further improvement, the preset phenotype in S1 includes the virulence or drug resistance of the microorganism.

[0015] The beneficial effect of the above scheme is that the virulence and drug resistance of pathogenic microorganisms are the key areas of focus for pathogenic microorganisms in clinical medical research and pathogenic microorganism research, and the research is relatively in-depth, with a clear and clear division of the relevant virulence and drug resistance.

[0016] Preferably, when the preset phenotype is microbial virulence or drug resistance, grouping can be performed after experimental testing according to the corresponding virulence or drug resistance detection methods disclosed in the prior art, such as: fluorescent quantitative PCR, drug sensitivity identification.

[0017] Specifically, when performing actual analysis, the preset phenotype may also be ST typing, surface antigen typing (serological typing) and the like of bacteria.

[0018] As a further improvement, the subset of the dataset described in S1 required an identification percentage ≥ 75% and a number of mass spectral peaks of 100-200.

[0019] As a further improvement, the common mass spectrometry dataset of the microbial samples described in S2 comes from different biological replicates and technical replicates.

[0020] As a further improvement, the quality control analysis in S3 includes an original mass spectrum data set and a common mass spectrum data set.

[0021] As a further improvement, the screening of the differential mass spectrum peaks described in S4 is as follows: ① excluding mass spectrum peaks with an appearance frequency of 100% in all strains; ② excluding mass spectrum peaks with an appearance frequency of <20% in all strains.

[0022] As a further improvement, the cluster analysis described in S4 is a two-way comprehensive analysis based on preset phenotypic groupings and mass spectrometry peak distributions.

[0023] As a further improvement, the screening of candidate molecular markers described in S5 is as follows: ① The mass spectrum peak to be analyzed has a POP < 20% in the target group of the preset phenotypic grouping; ② The mass spectrum peak to be analyzed is relatively highly expressed in other groups of the preset phenotypic grouping.

[0024] As a further improvement, the combination analysis described in S5 uses the method of permutation and combination, and the formula is as follows:

[0025]

[0026] In the formula, m represents the number of candidate protein joint marker combinations, and n represents the number of candidate molecular markers available for screening.

[0027] As a further improvement, the combined analysis described in S5 is followed by evaluation using clinical diagnostic efficacy evaluation indicators of molecular markers.

[0028] Specifically, it includes the assessment of sensitivity, specificity, accuracy, positive predictive value, and negative predictive value.

[0029] As a further improvement, the microbial sample to be analyzed in S1 is cultured to at least the F2 generation.

[0030] The beneficial effect of the above scheme is that after the microbial sample to be analyzed is cultured to the F2 generation, different offspring of the same sample (such as different colonies of the same strain) are taken for group testing to obtain mass spectrometry analysis data, which can reduce errors in the experimental analysis process.

[0031] In order to achieve the above-mentioned object, the technical scheme adopted by the application of a deep analysis method of a microbial ribosomal proteomics dataset in the identification of the virulence level of carbapenem-resistant Klebsiella pneumoniae in the present invention is:

[0032] Application of a deep analysis approach for microbial ribosomal proteomics datasets to characterize virulence levels in carbapenem-resistant Klebsiella pneumoniae.

[0033] The beneficial effect of the above scheme is that the application of the deep analysis method of the microbial ribosomal proteomics data set of the present invention in identifying the virulence level of carbapenem-resistant Klebsiella pneumoniae is a pioneering invention. By using the in-depth analysis method of the present invention, the combined labeling mass spectrum peaks of the highly toxic strains of carbapenem-resistant Klebsiella pneumoniae, the combined labeling mass spectrum peaks of the intermediate toxicity strains and the combined labeling mass spectrum peaks of the non-toxic strains were screened out. After analysis and verification, the protein molecular markers composed of the above three combined labeling mass spectrum peaks can accurately and quickly identify the virulence level of the carbapenem-resistant Klebsiella pneumoniae to be tested. In this way, a large amount of ribosomal proteomics data obtained in strain identification can be fully utilized, and the results of the analysis can be directly used for the extended analysis of clinical trial detection data, which will greatly improve the precision treatment of clinical anti-infection, and advance the time window of precision treatment and reduce the treatment cost of patients, and even save patients' lives. Moreover, compared with the detection of proteins and nucleic acids, the identification of the virulence of carbapenem-resistant Klebsiella pneumoniae using a protein molecule marker composed of three jointly labeled mass spectrometry peaks does not require additional experimental operations and testing costs, and there is no delay in detection time, which is suitable for large-scale, high-throughput detection and identification.

[0034] As a further improvement, the ribosomal proteomics dataset of the tested carbapenem-resistant Klebsiella pneumoniae was deeply analyzed, and the virulence level of the tested carbapenem-resistant Klebsiella pneumoniae was determined based on the combined labeled mass spectrometry peaks of different virulence.

[0035] As a further improvement, the combined labeled mass spectral peaks (m / z) of highly toxic carbapenem-resistant Klebsiella pneumoniae are 2368.64-2525.54-3130.11-3611.80, the combined labeled mass spectral peaks (m / z) of intermediate toxicity carbapenem-resistant Klebsiella pneumoniae are 2761.26-11178.83-11212.18-11873.99, and the combined labeled mass spectral peaks (m / z) of avirulent carbapenem-resistant Klebsiella pneumoniae are 2502.75-2580.52-4401.36-11178.83.

[0036] As a further improvement, the tolerance of the detection data of each mass spectrum peak in the combined marker mass spectrum peaks of the highly, intermediately and avirulent carbapenem-resistant Klebsiella pneumoniae is 800 ppm.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. The combined mass spectrometry peak protein molecular markers screened in the deep analysis method of the microbial ribosomal proteomics data set of the present invention determine the virulence status of the strain by analyzing the mass spectrometry peak data of the strain identification based on the microbial mass spectrometry detection, which is an extended analysis of clinical experimental detection data.

[0039] 2. The three groups of combined mass spectrometry peak protein markers in the present invention can be directly applied to clinical diagnosis and treatment, without the need for additional experimental operations and testing costs, and without delays in detection time. The mass spectrometry peak data detected can be directly compared with the clinical routine microbial mass spectrometry strain identification: 2368.64-2525.54-3130.11-3611.80, 2761.26-11178.83-11212.18-11873.99, 2502.75-2580.52-4401.36-11178.8, and the virulence level of carbapenem-resistant Klebsiella pneumoniae can be determined.

[0040] 3. The purpose of the present invention is to provide a method for analyzing a microbial ribosomal proteome dataset with a large sample size, high reproducibility, complete reliability and easy operation. Compared with existing transcriptomics, metabolomics and proteomics technologies, each strain used for research in the present invention has multiple biological experiments, and the representative mass spectra generated by merging multiple mass spectra are used as the mass spectrometry dataset of the strain, thereby ensuring the high reproducibility and high stability of subsequent data analysis results; the present invention can perform data analysis and processing of large sample sizes, and directly screen out molecular markers that distinguish related phenotypes based on the discovery of differentially expressed molecules. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 The figure is a schematic diagram of the specific process of the deep analysis method of the microbial ribosomal proteomics data set of the present invention;

[0042] Figure 2 This is a quality control analysis diagram of the original mass spectrometry spectra of the ribosomal proteomics dataset in Example 1 of the present invention (wherein a is the pop distribution of the three datasets of CRKP-KPC dataset, b is the mass spectrometry peak quantity distribution of the three datasets of CRKP-KPC dataset, c is the minimum / maximum mass spectrometry peak distribution of the three datasets of CRKP-KPC dataset, and d is the pop distribution of all subsets of CRKP-KPC dataset);

[0043] Figure 3 This is a quality control analysis diagram of the common mass spectrum after the spectra of the ribosomal proteomics data set are merged in Example 1 of the present invention (wherein a is the pop distribution of the labeled matched peak list set data set, b is the mass spectrum peak number distribution of the labeled matched peak list set data set, c is the minimum / maximum mass spectrum peak distribution of the labeled matched peak list set data set, and d is the pop distribution of all subsets of the labeled matched peak list set);

[0044] Figure 4 A matrix Venn diagram of the network relationship of the ribosomal proteomics dataset subsets in Example 1 of the present invention;

[0045] Figure 5 This is a heat map analysis of differential property spectrum peaks of the ribosomal proteomics dataset in Example 1 of the present invention. DETAILED DESCRIPTION

[0046] In the prior art, when using matrix-assisted laser desorption ionization time-of-flight mass spectrometry to identify microbial species, a large amount of ribosomal proteomic data is obtained, but there is currently no analytical method to effectively mine it, resulting in a large amount of waste of available biological information. Based on this, the present invention takes carbapenem-resistant Klebsiella pneumoniae as the specific detection object and its virulence as the specific analytical phenotype, and develops a deep analysis method for microbial ribosomal proteomic data sets. The specific process diagram is as follows Figure 1 This method has the characteristics of large sample size, high reproducibility, rapidity, high efficiency and low cost, which is of great significance for the further analysis of microbial ribosomal proteomics data and can be applied in microbial ribosomal proteomics analysis and related molecular marker screening.

[0047] At the same time, the present invention also provides a specific application of a deep analysis method of a microbial ribosomal proteomics data set in identifying the virulence of carbapenem-resistant Klebsiella pneumoniae, through which the combined labeling mass spectrum peaks of highly toxic strains of carbapenem-resistant Klebsiella pneumoniae, the combined labeling mass spectrum peaks of intermediate toxic strains, and the combined labeling mass spectrum peaks of non-toxic strains are screened. After analysis and verification, the protein molecule markers composed of the above three combined labeling mass spectrum peaks can accurately and quickly identify the virulence level of the carbapenem-resistant Klebsiella pneumoniae to be tested. In this way, a large amount of ribosomal proteomics data obtained in strain identification can be fully utilized, and the results of the analysis can be directly used for the extended analysis of clinical trial detection data, which will greatly improve the precision treatment of clinical anti-infection, and advance the time window of precision treatment and reduce the treatment cost of patients, and even save patients' lives. Moreover, compared with the detection of proteins and nucleic acids, the identification of the virulence of carbapenem-resistant Klebsiella pneumoniae using a protein molecule marker composed of three jointly labeled mass spectrometry peaks does not require additional experimental operations and testing costs, and there is no delay in detection time, which is suitable for large-scale, high-throughput detection and identification.

[0048] The present invention is further described in detail below in conjunction with specific examples. Unless otherwise specified, the equipment and reagents used in each embodiment, experimental example and comparative example can be obtained from commercial sources.

[0049] A specific embodiment of a deep analysis method of a microbial ribosomal proteomics dataset and its application in the present invention:

[0050] Example 1

[0051] In this example, carbapenem-resistant Klebsiella pneumoniae producing KPC enzymes is used as the specific detection object, and its virulence is used as the specific analysis phenotype, and the obtained ribosomal proteomics data is deeply analyzed. The specific implementation operations are as follows:

[0052] 1. Sample pretreatment

[0053] The carbapenem-resistant Klebsiella pneumoniae strains producing KPC enzymes were preserved, revived and subcultured, and the expression of their resistance genes and virulence genes was detected. In epidemiological studies in my country, more than 90% of the high drug resistance of the bacteria was caused by the production of KPC enzymes. The main high virulence marker genes of Klebsiella pneumoniae are iroN, rmpA, rmpA2 and iutA, and the strains were grouped according to the different expression levels of virulence genes; at the same time, each strain was tested and verified in the RUO mode of the VITEK MS mass spectrometer and the data information of the mass spectrum was collected. The specific method is as follows:

[0054] (1) Source and preservation of strains: The test strains were derived from sputum, blood, urine, cerebrospinal fluid, and wound secretion specimens submitted by inpatients of the First Affiliated Hospital of Henan University of Science and Technology from January to December 2018. The clinical pathogenic strains were isolated after routine microbial inoculation and culture, and were identified as Klebsiella pneumoniae by strain detection using a VITEK MS mass spectrometer in IVD mode. These Klebsiella pneumoniae were 2Compact drug sensitivity meter was used for drug sensitivity test, and the results were interpreted with reference to the MIC drug sensitivity breakpoints of Enterobacteriaceae in CLSIM100 (32nd edition). When the drug sensitivity breakpoint of imipenem was ≥4μg / mL, the strain was judged to be resistant to carbapenem antibiotics. Strains resistant to meropenem, imipenem, ertapenem, aztreonam, cefepime, cefotaxime, piperacillin / tazobactam, amikacin, and ciprofloxacin were preserved, and the collected strains were placed in a -80℃ refrigerator for long-term storage using the carrier preservation method. The quality control strain for mass spectrometer identification was Escherichia coli ATCC8739, and the quality control strains for drug sensitivity test were Escherichia coli ATCC25922 and Pseudomonas aeruginosa ATCC27853.

[0055] (2) Strain recovery and subculture: The carbapenem-resistant Klebsiella pneumoniae strain stored in a -80°C ultra-low temperature freezer was revived and inoculated on blood agar medium (Antu Biotechnology, China), and cultured in a 5% CO2 incubator at 35°C for 16-20 hours. Typical grayish white, moist, and smooth monoclonal colonies grew, which were the F0 generation. The culture was continued to the F2 generation.

[0056] (3) Molecular biological detection and grouping: 1.0 mL of sterile ultrapure water was added to a 1.5 mL sterile EP tube, and then a small amount of colonies on the F2 generation culture medium was placed in the EP tube and mixed. The genome of CRKP was extracted by boiling method, and the KPC gene detection method of ESNAYIGIT et al. (Novel Carbapenem-Hydrolyzing b-Lactamase, KPC-1, from a Carbapenem-Resistant Strain of Klebsiella pneumoniae, Antimicrobial Agents and Chemotherapy, 2001) was used, with upstream primers: TGTCACTGTATCGCCGTC (as shown in SEQ ID NO.1), and downstream primers: CTCAGTGCTCTACAGAAAACC (as shown in SEQ ID NO.2). Ordinary PCR technology was used for amplification experiments, and the virulence gene of carbapenem Klebsiella pneumoniae producing KPC enzyme was detected.

[0057] With reference to Fangyou Yu et al. (Multiplex PCR Analysis for Rapid Detection of Klebsiella pneumoniae Carbapenem-Resistant (Sequence Type 258 [ST258] and ST11) and Hypervirulent (ST23, ST65, ST86, and ST375) Strains, J Clin Microbiol, 2018) for the detection method of the high virulence gene of Klebsiella pneumoniae, the upstream primer of the rmpA gene is: ATGTGGCTTGACGTTTCGGGGG (as shown in SEQ ID NO.3), the downstream primer is: GCCGTGGATAATGGTTTACAATTCGGC (as shown in SEQ ID NO.4), the upstream primer of the rmpA2 gene is: GGATGTGGCTTGACATTTCGGGGG (as shown in SEQ ID NO.5), the downstream primer is: TTCATGGATGCCCTCCCTCCTG (as shown in SEQ ID NO.6), the upstream primer of the iroN gene is: CCGCAAAGAGACGAACCGCCTT (as shown in SEQ ID NO.7). ID NO.7), downstream primer: CGGGCAATCCCCGCTTTGACTT (as shown in SEQ ID NO.8), iutA gene upstream primer: AATCACCTGGGGGCTGGATGCT (as shown in SEQ ID NO.9), downstream primer: CCGCACCTTCCACGCCGTAAAT (as shown in SEQ ID NO.10). According to the expression of high virulence genes in 59 KPC-CRKP strains, they can be divided into 5 types, among which group A (iutA-rmpA-iroN-rmpA2-), group B (iutA+rmpA-iroN-rmpA2+) and group C (iutA+rmpA+iroN+rmpA2+) have higher detection rates (23.7%, 50.9% and 22.0%, see Table 1). According to the different expression conditions of the four high virulence genes, the strains in group A, group B and group C were named as avirulent subtype, intermediate virulence subtype and high virulence subtype, respectively.

[0058] Table 1 Expression levels and grouping of high virulence genes in carbapenem-resistant Klebsiella pneumoniae producing KPC enzyme

[0059]

[0060] (4) Mass spectrometry detection: CRKP-KPC strains in groups A, B, and C were detected in the RUO mode of microbial mass spectrometry. The bacterial components from 1 to 2 monoclonal colonies on the F2 generation culture medium were evenly spread on the target plate wells of the VITEK MS mass spectrometer, and 1.0 μL of CHCA matrix solution was immediately added. The operation was repeated in the order of sample numbers. After all samples were dried, the sample information was entered in the SARAMIS Target Manager software interface and transferred to the launchpadsoftware connected to the mass spectrometer. The VITEK MS microbial mass spectrometer RUO mode was used to detect bacterial components in the mass range of 2000 to 20000 m / z in the positive ion linear operation mode. The launchpad software collected mass spectra and transmitted them to the SARAMIS Premium database for strain verification and spectrum data storage. The quality control strain for mass spectrometry detection is Escherichia coli ATCC8739. Each strain is dipped into different colonies on the F2 generation culture medium for 5-8 biological replicates and technical replicates to ensure the repeatability and accuracy of the mass spectrometry spectrum.

[0061] (5) Mass spectrometry dataset: SARAMIS Premium database evaluates the quality of mass spectrometry spectra of detected strains based on identification of% and datacount, and exports the ribosomal protein spectra of bacteria with identification percentage ≥75% and mass spectrometry peak number between 100-200 in text format. Five mass spectrometry spectra that meet the requirements are saved for each strain based on the mass spectrometry detection quality, thereby obtaining the mass spectrometry detection dataset of each strain sample, named CRKP-KPC dataset. The CRKP-KPC dataset file contains sample data of group A, group B, and group C, named folderA, folderB, and folderC respectively. File folderA contains all sample data of group A, and the subfolders are named folderA1, folderA2, folderA3, etc. according to the sample number. The mass spectrometry data sets of each strain detection are numbered in sequence. For example, the mass spectrometry data text files of the first strain sample detection in group A are named textA1.1, textA1.2, textA1.3, textA1.4, and textA1.5, and saved in the folderA1 subfolder, and so on. The CRKP-KPC dataset contains mass spectrometry data of 14 strains in group A, 28 strains in group B, and 12 strains in group C.

[0062] 2. Mass spectrometry dataset preprocessing

[0063] The visualization analysis of the ribosomal protein dataset CRKP-KPC dataset detected by MALDI-TOF MS mass spectrometry was performed using Mass UP software (http: / / www.sing-group.org / mass-up / ) for data preprocessing, including dataset format conversion, import and peak matching, and finally generating a common mass spectrum for each sample.

[0064] (1) Format conversion: The text format documents in the CRKP-KPC dataset are converted into CSV format documents, and unnecessary data information is filtered out. Only the two columns of mass spectrometry data, Mass and Intensity, are retained. For example, the documents in folderA1 should be CSV A1.1, CSV A1.2, CSV A1.3, CSV A1.4, CSV A1.5, and so on.

[0065] (2) Data import: Select the grouped experiment type "LABELED" in Mass UP software, import and upload the CSV files in folderA, folderB and folderC into the software in sequence according to the experimental grouping and sample number, and save the muc format file and the original data set.

[0066] (3) Data preprocessing: The original data sets folder A, folder B, and folder C were imported into the preprocessing program. This data set has been subjected to spectral processing such as noise removal and smoothing in MALDI-TOF MS detection. Here, the parameters are set to the default values, and peakintensity is set to None.

[0067] (4) Peak matching: The three data sets of folder A, folder B and folder C use the MALDIquant matching method (same as VITEK MS mass spectrometry analysis method) for both intra-sample and inter-sample matching, with a tolerance of 0.0002 and a percentage of presence (POP) of 50 by default. First, for intra-sample matching, each sample has 5 replicates for peak matching to find common peaks; for inter-sample matching, the spectra of all samples are matched separately, and finally a common mass spectrum of each sample is created, namely the labeled matched peak list set. The common mass spectrum data is stored according to the sample information of strains in groups A, B and C, and the folders are named AINTRA-MATCHED INTER-MATCHED (matched A), B INTRA-MATCHED INTER-MATCHED (matched B) and CINTRA-MATCHED INTER-MATCHED (matched C).

[0068] 3. Analysis of ribosomal protein expression profiles of CRKP-KPC populations (basic analysis of mass spectrometry datasets)

[0069] The preprocessed MALDI-TOF MS ribosomal protein dataset CRKP-KPC dataset was visualized and analyzed using Mass UPsoftware (http: / / www.sing-group.org / mass-up / ), including quality control analysis, population representative mass spectrum peak generation and principal component analysis, to obtain the ribosomal protein expression profile of the CRKP-KPC population.

[0070] (1) Original spectrum quality control analysis: The quality analysis of the data sets among the three groups of samples in folder A, folder B and folder C was performed using the MALDIquant comparison method (same as the VITEK MS mass spectrometry analysis method), with a tolerance of the default value of 0.0002. The frequency of occurrence of each mass spectrometry peak in all samples was statistically analyzed, and the POP values ​​were analyzed in three intervals of 60%, 80% and 100%. Among the 270 replicates of 54 strains, the average number of mass spectrometry peaks with POP ≥ 60% in each of the three groups of samples was approximately 120, the average number of mass spectrometry peaks with POP ≥ 80% was approximately 90, and the average number of mass spectrometry peaks with POP ≥ 100% was approximately 60 ( Figure 2 The number of mass spectrometry peaks of the three groups of samples is about 150 ( Figure 2 However, the maximum mass peak value of the three groups of samples is about 10000m / z for group B samples, and close to 15000m / z for groups A and C ( Figure 2The percentage of each peak in each mass spectrometry spectrum of the 270 replicates in all sample mass spectrometry data sets was ≥40% ( Figure 2 Quality control analysis showed that the mass spectrometry peaks used for data set analysis had a high percentage of occurrence, and the frequency of mass spectrometry peaks and the distribution of peak numbers between samples were highly consistent, thus ensuring the repeatability and stability of mass spectrometry analysis.

[0071] (2) Quality control analysis of the common spectra: The mass spectra of each strain after merging were further analyzed. The matched A, matched B, and matched C in the labeled matched peak list set in step 2. (4) were imported into the Mass UP software. The operation method of the three mass spectrometry data sets was the same as that of the original spectra. The POP distribution characteristics, mass peak numbers, and maximum values ​​of the three combined spectra were highly consistent with the original spectra ( Figure 3 The peak values ​​of each strain in the 54 combined spectra are mainly concentrated in the range of 20% to 80% ( Figure 3 (shown in d).

[0072] (3) Representative mass spectrometry peaks of the CRKP-KPC population: The protein spectrum data sets of groups A, B, and C were used for inter-class biomarker analysis. The matched A, matched B, and matched C data in the labeled matched peak list set in step 2. (4) were imported into the biomarker discovery (inter-class analysis) module for analysis. The P value in the data analysis was calculated using a randomized independence test (based on 10,000 repetitions), and the q value was calculated using the Benjamini Hochberg method to calculate the FDR. The representative mass spectrometry peak list of the samples of the CRKP-KPC population was obtained by comprehensive analysis of the three strains in the CRKP-KPC population and output in matrix form, named CRKP-KPC matrix. The columns in the table are samples, showing all samples in the data set, except for two columns for p-values ​​and q-values; the rows are peaks, indicating all mass spectrometry peaks that appear in all strains in this data set. The value of each cell represents the percentage of the peak in that row in the samples in that column. "0" defaults to non-existence, which means that the strain in this column does not express the ribosomal protein represented by the mass spectrum peak here; "100" defaults to existence, which means that the strain in this column expresses the ribosomal protein represented by the mass spectrum peak here. A total of 389 representative mass spectrum peaks are displayed in the matrix form of the ribosomal protein expression spectrum of the CRKP-KPC population.

[0073] (4) Principal component analysis: The matched A, matched B, and matched C data in the labeled matched peak list set in step 2. (4) peak matching were imported into the Principal Component Analysis module, with Max.Components set to the default value of -1 and Variance Covered set to 0.95. The data set was analyzed using the correlation coefficient matrix. The principal component eigenvalues ​​of PC0, PC1, and PC2 in the three-dimensional principal component analysis graph were 32.9354, 20.0658, and 17.4351, respectively, indicating that the ribosomal protein spectrum can distinguish the different virulence subtypes (genotypes) of strains in group A (iutA-rmpA-iroN-rmpA2-), group B (iutA+rmpA-iroN-rmpA2+), and group C (iutA+rmpA+iroN+rmpA2+).

[0074] 4. Difference analysis of the three virulence subtypes of CRKP-KPC (difference analysis of mass spectrometry data sets)

[0075] The CRKP-KPC ribosomal protein spectrum dataset was analyzed using the online software Multiple List Comparator (https: / / molbiotools.com / listcompare.php), and data processing was used to screen out differential property peaks. Mass UP software was used to perform hierarchical clustering analysis on these differential proteins and output them in the form of a visual heat map, and further cluster analysis was performed.

[0076] (1) Matrix analysis

[0077] 1) Data processing: Step 2. (4) The mass spectrometry data of 54 strains in matched A, matched B and matched C in the labeled matched peak list set in peak matching were processed, and only the mass spectrometry peak digital information in the "m / z" column in the original file was retained. The mass spectrometry peak intensity data in the "Intensity" column was filtered out, and the files were named in CSV format according to the sample group and sample number, such as A1, A2, ..., B1, B2, ..., C1, C2, ..., and the converted data set was named matrix dataset.

[0078] 2) Data import: In the Compare Lists interface of the online software Multiple List Comparator, first set the "Number of lists" and enter the number of analysis sample lists (n, 54); then upload the CSV format document in the "list input" item and enter the file name. After the operation is completed, click "compare".

[0079] 3) Data analysis: The item list of 54 data subsets in the "Venn Diagram" entry has 49 intersections. The parameters of the "Pairwise Intersections" entry are set to Display as: Full symmetrical matrix, Color by: Jaccard index. After executing the operation, a matrix analysis form is generated. For each cell in the matrix form, the first row of data is the number of mass spectral peaks used for subset analysis between the two samples; the second row of data is the Jaccard coefficient of the two sample subset analysis, which indicates the degree of data overlap between the two samples; the matrix form data is stored as a matrix-based diagram ( Figure 4 The "Subset Item Listing" entry displays the analysis information of the dataset subset items, taking the analysis of group A and group B samples as an example: "AB", "A∩B", "A∪B", and displays the corresponding mass spectrum peak data information. The "SharedItems" entry displays the mass spectrum peak data information used for data analysis in all samples, and is output in txt format and named list.comparison.results.

[0080] 4) Screening of differential mass spectrum peaks: Convert list.comparison.results data from txt format to execl format, showing that there are 310 mass spectrum peaks in the CRKP-KPC comparison analysis list output by matrix analysis. Use execl to process data: i. Exclude mass spectrum peaks that exist in all strains analyzed by the matrix, that is, the frequency of occurrence = 100% (Occurences, 54); ii. Exclude mass spectrum peaks with low frequencies in all strains in the matrix analysis, and at the same time maximize the presence of the screened mass spectrum peaks in the data sets of the three components, that is, the frequency of occurrence is <20% (Occurences, <10.8). After screening, a total of 130 differential mass spectrum peaks were found. The matrix form is stored in execl format and named list.discrimination.results. The specific mass spectrum data information is shown in Table 2.

[0081] Table 2 130 differential property spectrum peaks were screened out from the CRKP-KPC ribosomal protein spectrum dataset by matrix analysis

[0082]

[0083]

[0084] (2) Hierarchical cluster analysis

[0085] 1) Data preprocessing: The CRKP-KPC matrix form in the representative property spectrum peaks of the CRKP-KPC population in step 3.(2) above is converted from CSV format to execl format. Only the matrix form data of 130 differential property spectrum peaks (see Table 1) are retained for data processing using execl, and the file is named CRKP-KPC discrimination matrix. In the differential matrix form, the columns respectively show the expression of differential property spectrum peaks in the 54 strain samples. Step 2.(4) The 54 subsets of matched A, matched B and matched C samples in the labeled matched peak list set in the peak matching are compared with the CRKP-KPC discrimination matrix form data information to filter out the non-differential property spectrum peak data of each strain, and only retain the differential property spectrum peak data. The newly generated differential property spectrum peak data information is stored in a CSV format file, the names of the 54 subsets remain unchanged, the corresponding sub-data sets are named discriminant A, discriminant B and discriminant C, and the total data set is named discriminant peak list set.

[0086] 2) Data analysis: The three data sets of discriminant A, discriminant B and discriminant C in the discriminant peak list set were imported into the data set through the "MPL" method, and Clustering Analysis was performed. The parameters were set as Minimum variance: 0, Cluster Reference Value: average, Distance Function: hamming, Conversion Values: presence, and Intra-sample Minimum Presence: 0. Hierarchical clustering analysis was performed on the differential property peak data set and presented in the form of a heat map visualization ( Figure 5 Green indicates that the mass spectrum peak is missing, and red indicates that the mass spectrum peak exists. Cluster analysis was performed on 54 strain samples and 130 differentially expressed mass spectrum peaks.

[0087] 3) Heat map analysis:

[0088] a. Cluster analysis of strain samples: The strains of different components showed a certain degree of mixing in the cluster analysis. The 54 subset data can be divided into two major clusters, cluterα and cluterβ. cluterα is mainly composed of strains from group C and group A, while cluterβ is mainly composed of strains from group B.

[0089] b. Analysis of differential mass spectrometry peak clusters: The frequency distribution of mass spectrometry peaks in different strains showed two major clusters, cluterⅠ and cluterⅡ. The mass spectrometry peak of cluter1a(*) was almost completely present in all strains, followed by the rest, and the frequency of protein peaks in cluterⅡ was relatively low. There were more mass spectrometry peaks in cluterⅡ, which were divided into several different branches, indicating that the distribution of protein peaks in this part was diversified.

[0090] c. Overall analysis: In the cluterⅡ mass spectrometry peak cluster, the κ region in the cluterα1 strain cluster showed a concentrated lack of mass spectrometry peaks; similarly, in the cluterⅡ mass spectrometry peak cluster, the λ region in the cluterβ2 strain cluster also showed a lack of mass spectrometry peaks. Comprehensive analysis showed that the joint detection of multiple negative expression mass spectrometry peaks is expected to serve as a molecular marker to distinguish different components, that is, a molecular marker of different virulence levels.

[0091] 5. Combined protein markers of three virulence subtypes of CRKP-KPC (molecular marker screening of mass spectrometry dataset)

[0092] The differential property spectral peaks of 130 ribosomal proteins were screened and analyzed to obtain candidate molecular markers, and then different candidate molecular markers were combined for analysis. The sensitivity, specificity, accuracy, positive predictive value and negative predictive value of the clinical diagnostic efficacy evaluation indicators of molecular markers were combined to finally discover a combined molecular marker that distinguishes different virulence levels.

[0093] (1) Data preprocessing: In step 4. (1) Matrix analysis 4) list.discrimination.results form of differential mass spectrum peak screening, the number of strains detected in group A, group B and group C, as well as the percentage of occurrence in each group, were calculated for the 130 mass spectrum peaks.

[0094] (2) Candidate molecular markers: The candidate molecular markers were screened from the 130 differential mass spectrometry peaks according to the following criteria: i. The mass spectrometry peak had a very low positive rate in the group, POP < 20%; ii. The mass spectrometry peak was relatively highly expressed in the other two groups. After the data preprocessing in step (1), a total of 27 mass spectrometry peaks were found in the list.discrimination.results form as candidate molecular markers, as shown in Table 3 below.

[0095] Table 3 27 candidate molecular markers screened from the CRKP-KPC ribosomal protein profile dataset

[0096]

[0097] (3) Molecular marker combination: In order to improve the accuracy and specificity of molecular markers between groups, the method of combined molecular markers is used to distinguish different virulence levels. Through the method of permutation and combination, possible combined markers can be found. Two, three or more molecular markers can be combined respectively. The combination types have the following calculation formula:

[0098]

[0099] In the formula, m represents the number of candidate protein joint labeling combinations;

[0100] n represents the number of candidate molecular markers available for screening.

[0101] (4) Interpretation of joint markers: If the combined interlocking markers are all missing in the mass spectrometry of the same strain, the group of markers can be successfully included in the analysis; conversely, if the linked markers are detected with ≥1 mass peak in the mass spectrometry of the same strain, they cannot be included in the analysis. Since the screening of candidate molecular markers included in the analysis is based on negative expression to predict different virulence levels, the interlocking markers are represented as peak1-peak2…(-), and the number of strains that can be included in the analysis in groups A, B, and C for each interlocking marker is calculated.

[0102] (5) Evaluation of diagnostic efficacy of combined markers: Analysis of sensitivity, specificity, accuracy, positive predictive value, and negative predictive value of different combinations of linked markers. The specific calculation formula is as follows:

[0103] Sensitivity = a / a+c×100%;

[0104] Specificity = d / b + d × 100%;

[0105] Accuracy = a + d / N × 100%;

[0106] Positive predictive value = a / a+b×100%;

[0107] Negative predictive value = d / c + d × 100%;

[0108] In the formula, a represents the number of strains included in the analysis of ribosomal protein-linked markers in this group;

[0109] b represents the sum of the number of strains included in the analysis for ribosomal protein-linked markers in the other two groups;

[0110] c indicates the number of strains whose ribosomal protein-linked markers were not included in the analysis in this group;

[0111] d represents the sum of the number of strains whose ribosomal protein-linked markers were not included in the analysis in the other two groups.

[0112] By using molecular markers for diagnostic performance evaluation, it was found that the combined molecular markers 2761.26-11178.83-11212.18-11873.99(-) had extremely high sensitivity and specificity (96.43% and 92.31%) for the identification of strains with intermediate virulence levels, and at the same time had extremely high positive predictive value and negative predictive value (93.10% and 96.00%), with an accuracy of up to 94.44% (see Table 4); the combined markers 2502.75-2580.52-4 The sensitivity of 401.36-11178.83(-) in identifying avirulent strains was 78.57%, with high specificity and negative predictive value (80.00% and 88.89%) (see Table 4); the sensitivity of the combined marker 2368.64-2525.54-3130.11-3611.80(-) in identifying highly virulent strains was 75.00%, with an accuracy of 87.04%, with extremely high specificity and negative predictive value (90.48% and 92.68%) (see Table 4). Analysis of these data showed that 2761.26-11178.83-11212.18-11873.99(-), 2502.75-2580.52-4401.36-11178.83(-) and 2368.64-2525.54-3130.11-3611.80(-) could be used as molecular markers to distinguish different virulence levels of CRKP-KPC strains.

[0113] Table 4 Combined molecular markers for distinguishing different virulence levels screened from the CRKP-KPC ribosomal protein profile dataset

[0114]

[0115]

[0116] Finally, it is noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for deep analysis of microbial ribosomal proteomics datasets, characterized by: The following steps are involved: S1. Sample pretreatment: The microbial samples to be analyzed are cultured and grouped according to preset phenotypes, and then subjected to matrix-assisted laser desorption ionization time-of-flight mass spectrometry to obtain a ribosomal protein spectrum data set; S2. Preprocessing of mass spectrometry data set: preprocessing the ribosomal protein spectrum data set obtained in S1 to obtain a common mass spectrometry data set of microbial samples; S3. Basic analysis of mass spectrometry data sets: quality control analysis, population representative mass spectrum peak generation and principal component analysis were performed on the mass spectrometry spectra obtained in S1 and S2 to obtain the population ribosomal protein expression profile of the microbial samples; S4. Mass spectrometry dataset difference analysis: Matrix analysis and data processing are performed on the common mass spectrometry dataset of the microbial samples described in S2 to screen out differential mass spectrometry peaks, and hierarchical clustering analysis is performed on the population ribosomal protein expression profiles of the microbial samples described in S3 to output a visualized heat map for further cluster analysis; S5. Molecular marker screening of mass spectrometry data set: The differential mass spectrometry peaks obtained in S4 were screened and analyzed to obtain candidate molecular markers. Combination analysis and clinical diagnostic efficacy evaluation of different candidate molecular markers were performed to determine the linked molecular markers for the preset phenotypic grouping of microorganisms.

2. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The preset phenotypes described in S1 include virulence or drug resistance of the microorganism.

3. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The subset of the dataset described in S1 required an identification percentage ≥ 75% and a number of mass spectral peaks of 100-200.

4. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The common mass spectrometry datasets of the microbial samples described in S2 were obtained from different biological replicates and technical replicates.

5. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The quality control analysis described in S3 includes an original mass spectral spectrum dataset and a common mass spectral spectrum dataset.

6. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The screening of differential mass spectrometry peaks described in S4 is as follows: ① exclude mass spectrometry peaks with a frequency of 100% in all strains; ② exclude mass spectrometry peaks with a frequency of <20% in all strains.

7. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The cluster analysis described in S4 is a two-way comprehensive analysis based on preset phenotypic groupings and mass spectrometry peak distribution.

8. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The screening of candidate molecular markers described in S5 is as follows: ① the mass spectrometry peak to be analyzed has a POP < 20% in the target group of the preset phenotypic group; ② the mass spectrometry peak to be analyzed has a relatively high expression in other groups of the preset phenotypic group.

9. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The combination analysis described in S5 uses the permutation and combination method, and the formula is as follows: In the formula, m represents the number of candidate protein joint marker combinations, and n represents the number of candidate molecular markers available for screening.

10. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The clinical diagnostic efficacy evaluation index of molecular markers was used for evaluation after combined analysis as described in S5.

11. The method for deep analysis of microbial ribosomal proteomics dataset according to claim 1, characterized in that: The microbial sample to be analyzed in S1 is cultured to at least the F2 generation.

12. Use of the method according to any one of claims 1 to 11 in identifying the virulence level of carbapenem-resistant Klebsiella pneumoniae.

13. Use of the method according to claim 12 in identifying the virulence level of carbapenem-resistant Klebsiella pneumoniae, characterized in that: The ribosomal proteomics dataset of the tested carbapenem-resistant Klebsiella pneumoniae was deeply analyzed, and the virulence level of the tested carbapenem-resistant Klebsiella pneumoniae was determined based on the linked protein markers of different virulence levels.

14. Use of the method according to claim 12 or 13 in identifying the virulence level of carbapenem-resistant Klebsiella pneumoniae, characterized in that: The combined labeled mass spectral peaks (m / z) of highly virulence carbapenem-resistant Klebsiella pneumoniae were 2368.64-2525.54-3130.11-3611.80, the combined labeled mass spectral peaks (m / z) of intermediate virulence carbapenem-resistant Klebsiella pneumoniae were 2761.26-11178.83-11212.18-11873.99, and the combined labeled mass spectral peaks (m / z) of non-virulence carbapenem-resistant Klebsiella pneumoniae were 2502.75-2580.52-4401.36-11178.83.

Images