Primer and method for detecting pathogens related to bacterial vaginitis and application

Abstract

The invention discloses a primer and a method for detecting bacterial vaginitis related pathogens and application of the primer and the method. The nucleic acid sequence of the primer group comprises the sequences as shown in SEQ ID NO.1 to SEQ ID NO.12. The invention further discloses a kit for detecting the content of the nucleic acid sequence. The invention provides a primer group and a method for evaluating bacterial vaginitis related pathogens by designing specific primers to carry out super-multiplex PCR amplification and enrichment, obtaining the detection data volume of each pathogen based on sequence reading of a gene sequencing platform and assisting by a machine learning model, and further provides a kit for evaluating bacterial vaginitis related pathogens and a kit for evaluating bacterial vaginitis related pathogens. The problems that an existing Amsel and Nutant scoring method is greatly influenced by subjective factors, the requirement for the sample size of multiple qPCR joint detection is high, and the 16s rDNA detection cost is high can be solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-throughput sequencing and gene detection, and relates to a primer, method and application for detecting bacterial vaginitis-related pathogens. Background Art

[0002] The female vagina is a complex micro-ecological system dominated by lactobacilli. If the normal flora in the vagina is out of balance, the lactobacilli decrease or disappear, and the facultative anaerobes and anaerobic bacteria increase, vaginal infection and inflammation will occur, which is clinically known as bacterial vaginosis (BV). This disease is common in women of childbearing age and is one of the common gynecological diseases.

[0003] At present, the diagnosis of BV is mainly based on the Amsel clinical diagnostic criteria and the Gram staining Nugent scoring diagnostic criteria. Among them, the Amsel diagnostic criteria are that at least 3 of the following 4 clinical indicators are positive to diagnose BV: (1) the vaginal discharge is thin, homogeneous, grayish white or yellow; (2) it releases a fishy odor after adding 10% potassium hydroxide solution; (3) the number of clue cells under microscopic examination is greater than 20% of the total number of vaginal epithelial cells; (4) the pH of vaginal discharge is greater than 4.5. Among them, clue cell positivity must meet the requirements. The Nugent scoring is a Gram staining microscopic examination of vaginal sample smears. The inspector identifies and statistically scores the bacterial morphology after staining. The results can be divided into normal (0-3 points), BV intermediate (4-6 points) and BV (7-10 points). Both diagnostic methods require certain skills and experienced inspectors. Although the Amsel diagnostic criteria are relatively simple, they are not sensitive enough to detect certain bacteria and are greatly affected by the subjective judgment of inspectors. Although the Nugent scoring method is more standardized, it requires professional technicians to perform microscopic examination and counting. The operation is relatively cumbersome, and since only some bacteria can be counted, there are intermediate judgment results.

[0004] With the rapid development of molecular biology in recent years, molecular diagnosis can accurately identify species (genus) and evaluate community state type (CST) by calculating the relative abundance of microorganisms. This method can provide more comprehensive information on vaginal flora, including those bacteria that are difficult to culture or detect by conventional methods, which helps to understand the microbiological basis of BV. The disadvantage is that 16S rDNA is expensive and requires professional bioinformatics analysis, so it is difficult to popularize in clinical diagnosis. Most of the previous studies used single-plex PCR, such as vaginal Gardnerella or vaginal Atopobium, but there is often a problem of low specificity. Even multiplex PCR is limited by the number of fluorescent channels of the instrument, and the same sample needs to be tested multiple times, which not only requires a large amount of nucleic acid, but also consumes a high cost.

[0005] CN102770559B discloses a method for diagnosing bacterial vaginitis, analyzing the presence of one or more lactobacilli and two or more pathogenic organisms in a patient test sample. PCR analysis of nucleic acid fragments corresponding to each target organism can be used to detect the presence of one or more lactobacilli and two or more pathogenic organisms. Then, the amount of the target organism can be used to determine a score indicating a diagnosis of bacterial vaginitis.

[0006] Therefore, there is an urgent need to provide a method for detecting bacterial vaginosis that is highly accurate, specific, simple, efficient and low-cost. Summary of the invention

[0007] In view of the deficiencies of the prior art and actual needs, the present invention provides a primer, method and application for detecting pathogens related to bacterial vaginitis, which can specifically detect Atopobium vaginalis, Gardnerella vaginalis, Megasphaera type 1, bacterial vaginitis-related bacteria BVAB-2, Lactobacillus crispatus and Lactobacillus iners.

[0008] In order to achieve the purpose of the invention, the present invention adopts the following technical solutions:

[0009] In a first aspect, the present invention provides a primer set for detecting bacterial vaginitis-related pathogens, wherein the nucleic acid sequence of the primer set includes sequences shown as SEQ ID NO.1-SEQ ID NO.12.

[0010] In the present invention, specific primers are designed for super-multiplex PCR amplification enrichment, the detection data volume of each pathogen is obtained based on the sequence reading of the gene sequencing platform, and with the assistance of a machine learning model, a primer set and method for evaluating bacterial vaginitis-related pathogens are provided, which can solve the problems of large subjective factors in the existing Amsel and Nugent scoring methods, high sample size requirements for multiplex qPCR combined detection, and high cost of 16s rDNA detection.

[0011] SEQ ID NO. 1: GGTTAATACCGGATACTCCATATT.

[0012] SEQ ID NO. 2: CCGTGTTCCAGTCCCAATCT.

[0013] SEQ ID NO. 3: GTGACATGGTGCTAATCCCT.

[0014] SEQ ID NO. 4: GCTGCCCACTTTCATGACTT.

[0015] SEQ ID NO. 5: GATGCCAACAGTATCCGTCCG.

[0016] SEQ ID NO. 6: CCTCTCCGACACTCAAGTTCGA.

[0017] SEQ ID NO.7: CAAAGGAATTGACGGGGACC.

[0018] SEQ ID NO. 8: AATTCAGTCTCCTGAATCGTCAGA.

[0019] SEQ ID NO. 9: AACTAACAGATTTACTTCGGTAATGA.

[0020] SEQ ID NO. 10: AGCTGATCATGCGATCTGC.

[0021] SEQ ID NO. 11: AGTCTGCCTTGAAGATCGG.

[0022] SEQ ID NO. 12: CTTTTAAACAGTTGATAGGCATCATC.

[0023] The internal references in the present invention include: the conserved gene ACTB of the human genome, which is used to monitor whether the sampling and detection process is qualified.

[0024] The specific primer sequences for amplifying the human internal reference gene ACTB in the present invention include the nucleic acid sequences described in SEQ ID NO.13 and SEQ ID NO.14.

[0025] SEQ ID NO. 13: GCCACCAGAAGAGGTAG.

[0026] SEQ ID NO. 14: AATCTGGCACCACACCTTC.

[0027] Preferably, the bacterial vaginosis-related pathogens include any one or a combination of at least two of Atopobium vaginalis, Gardnerella vaginalis, Megasphaera type 1, bacterial vaginosis-related bacteria BVAB-2, Lactobacillus crispatus or Lactobacillus iners.

[0028] In a second aspect, the present invention provides use of the primer set for detecting bacterial vaginitis-related pathogens described in the first aspect in preparing a product for detecting bacterial vaginitis-related pathogens.

[0029] In a third aspect, the present invention provides a kit for detecting bacterial vaginitis-related pathogens, the kit comprising the primer set for detecting bacterial vaginitis-related pathogens described in the first aspect.

[0030] In a fourth aspect, the present invention provides use of the primer set for detecting bacterial vaginosis-related pathogens described in the first aspect in detecting bacterial vaginosis pathogens.

[0031] In a fifth aspect, the present invention provides a device for detecting bacterial vaginitis-related pathogens, the device comprising: a super-multiplex PCR amplification module, a high-throughput sequencing and quality control module, and a model building module;

[0032] The super-multiplex PCR amplification module is used to perform the following steps: extracting DNA from the sample to be tested as a template, and performing super-multiplex PCR amplification using the primer set for detecting bacterial vaginitis-related pathogens described in the first aspect to obtain a super-multiplex PCR library;

[0033] The high-throughput sequencing and quality control module is used to perform the following steps: performing high-throughput sequencing and quality control on the super-multiplex PCR library, normalizing the number of reads of the samples to be tested that have passed the quality control, and obtaining a pathogen normalized reads number matrix;

[0034] The model building module is used to perform the following steps: predicting the results of bacterial vaginosis detection through a machine learning model.

[0035] Preferably, the sample to be tested includes vaginal secretions.

[0036] Preferably, the quality control qualified standard is that the total number of sequencing reads of each sample to be tested is ≥20k, the Q30 proportion is ≥75%, and the number of internal reference normalized reads is ≥50; the Q30 proportion means that among all the bases generated by sequencing, at least 75% of the bases meet the Q30 quality standard, and the number of internal reference normalized reads = the number of reads aligned to the internal reference / the total number of reads of sample sequencing data × 100000.

[0037] Preferably, the machine learning model includes any one of a support vector machine model, a random forest model or a decision tree model.

[0038] Preferably, the input of the machine learning model is a normalized reads number matrix of bacterial vaginitis-related pathogens generated by the sample to be tested, and the output is a negative or positive prediction result of bacterial vaginitis for each sample to be tested, wherein the bacterial vaginitis-related pathogens include any one or a combination of at least two of vaginal Atopobium, vaginal Gardnerella, Megasphaera type 1, bacterial vaginitis-related bacteria BVAB-2, Lactobacillus crispatus or Lactobacillus iners.

[0039] In a sixth aspect, the present invention provides a method for constructing a diagnostic model for bacterial vaginitis amplified by super multiplex PCR, the construction method comprising the following steps:

[0040] (1) Designing specific primers for bacterial vaginitis-related pathogens, wherein the bacterial vaginitis-related pathogens include any one or a combination of at least two of Atopobium vaginalis, Gardnerella vaginalis, Megasphaera type 1, bacterial vaginitis-related bacteria BVAB-2, Lactobacillus crispatus or Lactobacillus iners, and the specific primers include sequences as shown in SEQ ID NO.1 to SEQ ID NO.12;

[0041] (2) a common sequence is set for the 5' end of the specific primer sequence, the common sequence is complementary to the 3' end of the sequencing adapter of the Illumina platform, and a sequence complementary to the common sequence is set for the 3' end of each sequencing adapter primer;

[0042] (3) preparing specific primers and amplification enzymes to form a reagent system 1, and performing super-multiplex PCR amplification of the target region on the extracted DNA of the sample to be tested;

[0043] (4) After the target region fragment is purified by magnetic beads, it is amplified with a sequencing adapter primer. The sequencing adapter primer and the amplification enzyme are prepared to form a library amplification reagent system 2, and the library is amplified to finally obtain a library structure of "Illumina sequencing adapter-public sequence-target region-public sequence-Illumina sequencing adapter";

[0044] (5) purifying each super-multiplex PCR library with magnetic beads, and after purification, mixing the libraries to obtain a bacterial vaginosis detection library;

[0045] (6) performing high-throughput sequencing on the bacterial vaginosis detection library obtained in step (5), using a bioinformatics analysis method to align the sequenced reads to a self-built genome database, obtaining the number of reads aligned to the internal reference and each pathogen, and performing quality control. The quality control pass standard is that the total number of sequencing reads for each sample to be tested is ≥20k, the Q30 ratio is ≥75%, and the internal reference normalized read number is ≥50; the Q30 ratio means that among all the bases generated by sequencing, at least 75% of the bases meet the Q30 quality standard, where Q30 is a sequencing error rate indicator, and the internal reference normalized read number = the number of reads aligned to the internal reference / the total number of reads of the sample sequencing data × 100000;

[0046] (7) For samples that pass the quality control, firstly, the number of reads aligned to pathogenic microorganisms is obtained through bioinformatics statistical analysis, and then the maximum and minimum normalization formula is used to normalize each pathogen of all samples to be tested, and a pathogen normalized read number matrix corresponding to the samples to be tested is generated, wherein the maximum and minimum normalization is to convert the number of reads of all pathogens into a value between 0 and 1, which is regarded as the relative abundance between pathogens. The normalization formula is as follows: Assuming that the number of reads of the pathogen is a list x_list, and now a pathogen read number x is given, then the normalization x'=(x-min(x_list)) / (max(x_list)-min(x_list)), where min(x_list) means obtaining the minimum value in the list x_list, and max(x_list) means obtaining the maximum value in the list x_list;

[0047] (8) A batch of negative and positive results of bacterial vaginosis corrected by Nugent and Amsel are used as the gold standard, and the operation is carried out according to steps (1) to (7), and the normalized reads number matrix of the detected pathogen is used as the eigenvalue, and is input into the support vector machine, random forest and decision tree machine learning model training respectively. According to the test set sample results, the model with the best prediction performance is selected as the final model for the prediction analysis of bacterial vaginitis, and the test set samples with known negative and positive bacterial vaginitis after correction are selected as references, and predictions are made through the trained machine learning model, and the sensitivity and specificity values ​​of different models are calculated respectively. Finally, according to the Youden index, the prediction model with the best analysis performance is selected, wherein the calculation formula of the Youden index is: Youden index = sensitivity + specificity - 1.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] (1) The present invention improves the detection sensitivity by designing specific primers for the target and performing super-multiplex PCR amplification enrichment. The detection data volume of each pathogen is obtained based on the sequence reading of the gene sequencing platform (Illumina), and with the assistance of a machine learning (ML) model, a primer set and method for evaluating bacterial vaginitis-related pathogens are provided, which can solve the problems of the existing Amsel and Nugent scoring methods, such as the large subjective factors, the high sample volume requirements of multiplex qPCR joint detection, and the high cost of 16s rDNA detection;

[0050] (2) The detection samples in the present invention can directly use clinical routine vaginal microecological detection samples without additional sampling. The constructed detection system combined with the prediction model has many advantages for predicting BV, such as non-invasiveness, high sensitivity, accuracy and objectivity. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is the peak diagram of the library of the present invention;

[0052] Figure 2 A schematic diagram of the optimal process of the molecular prediction method for bacterial vaginosis based on machine learning models;

[0053] Figure 3 Predict performance graphs for support vector machine machine learning models;

[0054] Figure 4 Predict performance graphs for random forest machine learning models;

[0055] Figure 5 Predictive performance graph for a decision tree machine learning model. DETAILED DESCRIPTION

[0056] To further illustrate the technical means and effects of the present invention, the present invention is further described below in conjunction with the embodiments and drawings. It should be understood that the specific implementation methods described herein are only used to explain the present invention, rather than to limit the present invention.

[0057] If no specific techniques or conditions are specified in the examples, the techniques or conditions described in the literature in the field or the product instructions are used. If no manufacturer is specified for the reagents or instruments used, they are all conventional products that can be purchased through regular channels.

[0058] Example 1

[0059] Clinical sample testing.

[0060] The primer set was used to detect the collected clinical samples, and the method was as follows:

[0061] 1. Sample sampling: In the gynecology outpatient department of the hospital, 211 female vaginal secretion samples were collected and Amsel diagnosis was performed at the same time. The collected samples were sent to the laboratory for Gram staining Nugent scoring diagnosis. For intermediate samples with scores of 4-6, corrections were made according to the Amsel diagnosis results, so that all samples had BV positive and negative diagnosis results. The sampling method is as follows: Use the distributed sampling sterile swab to rotate and stay for 20 seconds from the 1 / 3 side wall mucosa of the female vagina to obtain secretions, ensure that the sample volume is sufficient, place the swab after collecting the sample into the sampling tube, and rotate and roll on the inner wall of the sampling tube to fully mix the sample with the preservation solution, break the upper part of the swab at the mark, and tighten the lid to prevent spillage, place the preservation tube in the refrigerator at 2℃-8℃, and complete the test within 2 days.

[0062] 2. Sample pretreatment: Place the sampling tube on a vortex mixer and vortex thoroughly for 30 seconds. Directly pipette 1.3 mL of sample into a 1.5 mL sterile centrifuge tube as the sample to be tested.

[0063] 3. Nucleic acid extraction: Use nucleic acid extraction or purification reagent (Guangzhou Jinqirui Biotechnology Co., Ltd., registration number Yuesui Xiebei 20220884) to extract according to the instructions.

[0064] 4. Preparation of primer mixture: Prepare all primers into primer mixture according to the ratio in Table 1 below. Prepare primer mixture according to the number of detection reactions, and each reaction requires 1.5 μL of primer mixture.

[0065] Table 1

[0066]

[0067]

[0068] The method for preparing a 100 μL primer mixture is as follows: after diluting the primers to 100 μM, use a pipette to respectively absorb 5 μL of SEQ ID NO.1 and SEQ ID NO.2, then use a pipette to respectively absorb 7.5 μL of SEQ ID NO.3, SEQ ID NO.4, SEQ ID NO.5, SEQ ID NO.6, SEQ ID NO.11 and SEQ ID NO.12, then use a pipette to respectively absorb 10 μL of SEQ ID NO.7, SEQ ID NO.8, SEQ ID NO.9 and SEQ ID NO.10, then use a pipette to respectively absorb 2.5 μL of SEQ ID NO.13 and SEQ ID NO.14, mix thoroughly, store at 4°C, use within one week, and if stored for a long time, place at -20°C.

[0069] 5. Library Preparation

[0070] (1) PCR1 reaction: Perform super multiplex PCR amplification on the nucleic acid template, prepare a library amplification reaction system with a volume of 25 μL according to 1.5 μL multiple primer mixture + 12.5 μL PCR reaction buffer + 11 μL nucleic acid, and obtain the target region fragment. The amplification program is as shown in Table 2:

[0071] Table 2

[0072]

[0073]

[0074] (2) After the above PCR1 reaction product is purified, a library amplification reaction system with a volume of 25 μL is prepared according to 1 μL adapter primer + 12.5 μL PCR reaction buffer + 11.5 μL purified enriched product, and library amplification is performed to finally obtain a library structure of "Illumina sequencing adapter-public sequence-target region-public sequence-Illumina sequencing adapter". The amplification program is as shown in Table 3:

[0075] Table 3

[0076]

[0077] (3) Purify each super-multiplex PCR library, and after purification, mix the sample libraries to obtain a library for BV-related pathogenic microorganism detection.

[0078] 6. Sequencing on the machine: The sequencer used for sequencing was KMMiniSeqDx-CN produced by Guangzhou Jinqirui Biotechnology Co., Ltd., adapted to the universal sequencing reaction kit (Guangzhou Jinqirui Biotechnology Co., Ltd., product number KS107-CXR, specification MR100), and the library was sequenced on the machine according to the instructions.

[0079] 7. Data analysis: For the fastq files obtained by sequencing, first use fastp software to remove the adapter sequence and filter the low-quality data in the original sequencing results. Then write a script to screen and retain the valid reads with primer sequences in the sequencing products. Then use bwa alignment software to align the valid reads product sequences to the self-built database and count the number of reads covered by each pathogen. The self-built database is the genome matched according to the product sequence of the primer, which comes from the Nucleotide database of NCBI. Each target contains at least one matching genome. For the same target, try to increase multiple genomes so that all possible product sequences of the primer can be perfectly aligned to the database.

[0080] 8. Sample quality control: Quality control is performed on the data obtained by analysis. The minimum number of total sequencing reads for each sample is ≥20k, otherwise the original data is judged to be unqualified; the sequencing quality requires that the Q30 ratio is ≥75%, otherwise the sequencing quality is judged to be unqualified; the number of normalized reads for the internal reference is ≥50, otherwise it is judged to be unqualified for the internal reference. The Q30 ratio means that among all the bases generated by sequencing, at least 75% of the bases meet the Q30 quality standard, where Q30 is a sequencing error rate indicator. Specifically, Q30 means that the sequencing error rate does not exceed 1 / 1000, that is, the sequencing accuracy of each base is higher than 99.9%; the number of normalized reads for the internal reference = the number of reads aligned to the internal reference / the total number of reads of the sample sequencing data × 100000. In this embodiment, 211 samples passed the quality control according to the above detection process.

[0081] Example 2

[0082] Performance evaluation of 3 machine learning models.

[0083] In this embodiment, based on the first embodiment, the 211 samples that passed the quality control were divided into training set and test set samples according to a ratio of 7:3, resulting in 148 training set samples and 63 test set samples.

[0084] For the reads numbers of the six pathogens in each sample obtained in Example 1, each pathogen is normalized according to the maximum and minimum normalization formula to obtain its relative abundance, and a normalized reads number matrix of the six pathogens corresponding to the sample is generated; wherein, the normalization formula is as follows: Assuming that the number of reads of the six pathogens is a list x_list, and now a pathogen read number x is given, then the normalization x'=(x-min(x_list)) / (max(x_list)-min(x_list)), the data after normalization are all between 0 and 1, which can ensure that different data sets or features are numerically consistent, and in the formula, min(x_list) means obtaining the minimum value in the list x_list, and max(x_list) means obtaining the maximum value in the list x_list.

[0085] For the 148 samples in the training set, the BV positive and negative results after Nugent and Amsel diagnostic correction were used as the gold standard. The sklearn machine learning package of the Python programming language was used to write codes to input the normalized reads matrix into the three models of support vector machine (SVM), random forest (RF) and decision tree (DT) for training. All model parameters were default values. The constructed models were used to predict the 63 test set samples, and the BV positive and negative results after Nugent and Amsel diagnostic correction were used as references to analyze and evaluate the performance of each prediction model, calculate the sensitivity (Sensitivity) and specificity (Specificity) values, and finally select the prediction model with the best analysis performance based on the Youden's index. Among them, the calculation method of the Youden's index is: Youden's index = sensitivity + specificity - 1.

[0086] According to the above analysis method, the receiver operating characteristic (ROC) curve was drawn for the prediction performance results of the three machine learning models, as shown in Figure 2. Figure 3-Figure 5 As shown in the results, the Youden index of the support vector machine model is 0.95, which performs best among the three models. Therefore, the support vector machine model is selected as the optimal prediction model for the prediction analysis of bacterial vaginosis.

[0087] Example 3

[0088] The SVM model was used to validate the external validation set.

[0089] In the gynecological outpatient department of the hospital, according to the results of Nugent and Amsel diagnostic correction, 55 BV negative and positive vaginal secretion samples were retrospectively searched, and the primer set of the present invention was detected for the 110 samples with reference to Example 1, and bioinformatics analysis and statistics were performed to obtain the number of reads of 6 bacterial vaginosis-related pathogens for each sample. For samples that passed the quality control, each pathogen was normalized according to the maximum and minimum normalization formula to obtain its relative abundance and generate the pathogen normalized reads number matrix corresponding to the sample. Based on Example 2, the SVM model with the best performance was used as the bacterial vaginosis prediction model of the present invention.

[0090] According to the above analysis process, a total of 110 samples were tested. After removing 3 samples that failed quality control, 50 BV positive and negative samples were selected for model prediction. The corrected bacterial vaginosis results were used as a reference to calculate the prediction results of the SVM model as shown in Table 4:

[0091] Table 4

[0092]

[0093] The results showed that the quality control pass rate of vaginal secretions detected by the primer set of the present invention was 97.3%. 100 samples that passed the quality control were selected, and the accuracy of the external validation set using the SVM machine learning model was 99%, the sensitivity was 100%, and the specificity was 98%, indicating that the primer set of the present invention combined with the machine learning prediction model can accurately distinguish between negative and positive bacterial vaginosis.

[0094] In summary, the present invention improves the detection sensitivity by designing specific primers for the target, performing super-multiplex PCR amplification enrichment, obtaining the detection data of each pathogen based on the sequence reading of the gene sequencing platform, and providing a primer set and method for evaluating bacterial vaginitis-related pathogens with the assistance of a machine learning model, which can solve the problems of large subjective factors in the existing Amsel and Nugent scoring methods, high sample size requirements in multiplex qPCR combined detection, and high cost of 16srDNA detection.

[0095] The applicant declares that the present invention illustrates the detailed method of the present invention through the above-mentioned embodiments, but the present invention is not limited to the above-mentioned detailed method, that is, it does not mean that the present invention must rely on the above-mentioned detailed method to be implemented. Those skilled in the art should understand that any improvement of the present invention, equivalent replacement of various raw materials of the product of the present invention, addition of auxiliary components, selection of specific methods, etc., all fall within the protection scope and disclosure scope of the present invention.

Claims

1. A primer set for detecting bacterial vaginitis-related pathogens, characterized in that: The nucleic acid sequences of the primer set include sequences shown as SEQ ID NO.1-SEQ ID NO.

12.

2. The primer set according to claim 1, characterized in that The bacterial vaginitis-related pathogens include any one or a combination of at least two of Atopobium vaginalis, Gardnerella vaginalis, Megasphaera type 1, bacterial vaginitis-related bacteria BVAB-2, Lactobacillus crispatus or Lactobacillus iners.

3. Use of the primer set for detecting bacterial vaginitis-related pathogens according to claim 1 or 2 in the preparation of a product for detecting bacterial vaginitis-related pathogens.

4. A kit for detecting bacterial vaginitis-related pathogens, characterized in that: The kit comprises the primer set for detecting bacterial vaginitis-related pathogens according to claim 1 or 2.

5. Use of the primer set for detecting bacterial vaginitis-related pathogens according to claim 1 or 2 in detecting bacterial vaginitis-related pathogens.

6. A device for detecting bacterial vaginosis-related pathogens, characterized in that: The device comprises: a super-multiplex PCR amplification module, a high-throughput sequencing and quality control module and a model building module; The super-multiplex PCR amplification module is used to perform the following steps: extracting DNA from the sample to be tested as a template, and performing super-multiplex PCR amplification using the primer set for detecting bacterial vaginitis-related pathogens according to claim 1 or 2 to obtain a super-multiplex PCR library; The high-throughput sequencing and quality control module is used to perform the following steps: performing high-throughput sequencing and quality control on the super-multiplex PCR library, normalizing the number of reads of the samples to be tested that have passed the quality control, and obtaining a pathogen normalized reads number matrix; The model building module is used to perform the following steps: predicting the results of bacterial vaginosis detection through a machine learning model.

7. The device according to claim 6, characterized in that The sample to be tested includes vaginal secretions; Preferably, the quality control qualified standard is that the total number of sequencing reads of each sample to be tested is ≥20k, the Q30 proportion is ≥75%, and the number of internal reference normalized reads is ≥50; the Q30 proportion means that among all the bases generated by sequencing, at least 75% of the bases meet the Q30 quality standard, and the number of internal reference normalized reads = the number of reads aligned to the internal reference / the total number of reads of sample sequencing data × 100000.

8. The device according to claim 6 or 7, characterized in that The machine learning model includes any one of a support vector machine model, a random forest model or a decision tree model.

9. The device according to any one of claims 6 to 8, characterized in that The input of the machine learning model is a normalized reads number matrix of bacterial vaginitis-related pathogens generated by the sample to be tested, and the output is a negative or positive prediction result of bacterial vaginitis for each sample to be tested. The bacterial vaginitis-related pathogens include any one or a combination of at least two of vaginal Atopobium, vaginal Gardnerella, Megasphaera type 1, bacterial vaginitis-related bacteria BVAB-2, Lactobacillus crispatus or Lactobacillus iners.

10. A method for constructing a diagnostic model for bacterial vaginosis based on super multiplex PCR amplification, characterized in that: The construction method comprises the following steps: (1) Designing specific primers for bacterial vaginitis-related pathogens, wherein the bacterial vaginitis-related pathogens include any one or a combination of at least two of Atopobium vaginalis, Gardnerella vaginalis, Megasphaera type 1, bacterial vaginitis-related bacteria BVAB-2, Lactobacillus crispatus or Lactobacillus iners, and the specific primers include sequences as shown in SEQ ID NO.1 to SEQ ID NO.12; (2) a common sequence is set for the 5' end of the specific primer sequence, the common sequence is complementary to the 3' end of the sequencing adapter of the Illumina platform, and a sequence complementary to the common sequence is set for the 3' end of each sequencing adapter primer; (3) preparing specific primers and amplification enzymes to form a reagent system 1, and performing super-multiplex PCR amplification of the target region on the extracted DNA of the sample to be tested; (4) After the target region fragment is purified by magnetic beads, it is amplified with a sequencing adapter primer. The sequencing adapter primer and the amplification enzyme are prepared to form a library amplification reagent system 2, and the library is amplified to finally obtain a library structure of "Illumina sequencing adapter-public sequence-target region-public sequence-Illumina sequencing adapter"; (5) purifying each super-multiplex PCR library with magnetic beads, and after purification, mixing the libraries to obtain a bacterial vaginosis detection library; (6) performing high-throughput sequencing on the bacterial vaginosis detection library obtained in step (5), using a bioinformatics analysis method to align the sequenced reads to a self-built genome database, obtaining the number of reads aligned to the internal reference and each pathogen, and performing quality control. The quality control pass standard is that the total number of sequencing reads for each sample to be tested is ≥20k, the Q30 ratio is ≥75%, and the internal reference normalized read number is ≥50; the Q30 ratio means that among all the bases generated by sequencing, at least 75% of the bases meet the Q30 quality standard, where Q30 is a sequencing error rate indicator, and the internal reference normalized read number = the number of reads aligned to the internal reference / the total number of reads of the sample sequencing data × 100000; (7) For samples that pass the quality control, firstly, the number of reads aligned to pathogenic microorganisms is obtained through bioinformatics statistical analysis, and then the maximum and minimum normalization formula is used to normalize each pathogen of all samples to be tested, and a pathogen normalized read number matrix corresponding to the samples to be tested is generated, wherein the maximum and minimum normalization is to convert the number of reads of all pathogens into a value between 0 and 1, which is regarded as the relative abundance between pathogens. The normalization formula is as follows: Assuming that the number of reads of the pathogen is a list x_list, and now a pathogen read number x is given, then the normalization x'=(x-min(x_list)) / (max(x_list)-min(x_list)), where min(x_list) means obtaining the minimum value in the list x_list, and max(x_list) means obtaining the maximum value in the list x_list; (8) A batch of negative and positive results of bacterial vaginosis corrected by Nugent and Amsel are used as the gold standard, and the operation is carried out according to steps (1) to (7), and the normalized reads number matrix of the detected pathogen is used as the eigenvalue, and is input into the support vector machine, random forest and decision tree machine learning model training respectively. According to the test set sample results, the model with the best prediction performance is selected as the final model for the prediction analysis of bacterial vaginitis, and the test set samples with known negative and positive bacterial vaginitis after correction are selected as references, and predictions are made through the trained machine learning model, and the sensitivity and specificity values ​​of different models are calculated respectively. Finally, according to the Youden index, the prediction model with the best analysis performance is selected, wherein the calculation formula of the Youden index is: Youden index = sensitivity + specificity - 1.

Images