A method for rapid identification of Streptococcus agalactiae and its highly virulent strains

Through the combination of deep learning and Raman spectroscopy, a classification model of GBS and III/ST17 high-virulence strains was established, which solved the problem of delay and time-consuming of traditional bacterial identification methods, and realized the high-virulence strains of GBS and III/ST17 high-virulence strains. Quick identification provides efficient and convenient diagnostic tools.

CN118603963BActive Publication Date: 2024-11-19ZHEJIANG UNIV
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
CN202411045173.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-01
Publication Date
2024-11-19
Estimated Expiration
2044-08-01

AI Technical Summary

Technical Problem

Traditional bacterial identification methods have delayed, expensive and time-consuming problems in diagnosing Streptococcus adenoid (GBS) and its high virulence strain III/ST17, especially in neonatal infections, which are severe for individuals. as a result of.

Method used

Combining deep learning and Raman spectroscopy technology, by establishing a classification model of GBS and III/ST17 high-virulence strains, data is collected using laser confocal Raman spectrometer, and classification is performed through convolutional neural networks to achieve rapid identification.

Benefits of technology

It realizes rapid and sensitive identification of GBS and III/ST17 high-virulence strains, provides efficient and convenient diagnostic tools, and overcomes the complexity and time-consuming problems of traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118603963B_ABST
    Figure CN118603963B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for rapid identification of Streptococcus agalactiae and its highly virulent strains. A laser confocal Raman spectrometer collects single-cell Raman spectra of clinical strains to establish a database. The data of highly virulent strains of III / ST17 are grouped into one group; the data of GBS non-III / ST17 strains are grouped into a second group; and the data of Escherichia coli are grouped into a third group. The three groups of data are randomly divided into a training set, a validation set, and a test set at a ratio of 7:1:2. Based on the ResNet18 residual neural network architecture, the 2D convolution in the network is changed to a 1D convolution suitable for 1-dimensional Raman spectroscopy, the optimal parameters are determined through simulation experiments, the weights of the model are trained in the training set, the optimal hyperparameters are selected through the results of the model on the validation set, and the trained model is determined. Compared with the prior art, the present invention saves time, reagents, and consumables.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the fields of bacterial identification methods, deep learning, and Raman spectroscopy technology, and specifically relates to a method for rapidly identifying Streptococcus agalactiae and its highly toxic strains. Background Art

[0002] Bacterial meningitis is an acute central nervous system infection caused by bacteria with high morbidity and mortality, affecting all populations, especially neonates. The neonatal period is the period of life with the highest risk of serious infection and the most vulnerable period for survival. Streptococcus agalactiae is a Gram-positive bacterium, also known as group B streptococcus (GBS), and is the main pathogen causing bacterial meningitis in neonates. Among the isolated clinical strains, serotypes III, Ia, V, Ib and II were found in most cases by serotyping of the capsular polysaccharide, with serotype III being the most prevalent. GBS can also be multilocus sequence typing (MLST), a single capsular serotype III GBS clone, belonging to the hypervirulent clone complex 17, is almost exclusively associated with cases of meningitis and late-onset neonatal GBS infection, and is often referred to as the hypervirulent clone, also known as type III / ST17 GBS hypervirulent strain.

[0003] Meningitis caused by highly virulent GBS strains III / ST17 is a major cause of neurological disability worldwide, with considerable emotional and economic impact on individuals, families, and society. Delayed diagnosis and treatment have serious, long-term consequences for individuals, and timely diagnosis and treatment are particularly important as neonatal GBS disease overlaps with critical periods of neurodevelopment.

[0004] Traditional methods for pathogen identification rely on phenotypic identification, mainly using staining, culture, and simple biochemical tests. Metagenomic next-generation sequencing technology can detect highly virulent strains of GBS type III / ST17. Compared with traditional PCR-based methods, it has the key advantage of being completely unbiased. This method is not limited by the prior selection of suspected pathogens and can identify potential infectious agents in a single test, which is particularly important for limited samples obtained from newborns, but it is expensive and time-consuming.

[0005] Raman spectroscopy is a powerful and reliable technology for bacterial detection and characterization. Raman spectra originating from the vibration of molecular groups are regarded as the "fingerprint spectrum" of the structure and composition of biological molecules, and have been widely used in many fields such as biomedicine, chemical testing, and mineral analysis. Deep learning and machine learning methods can mine the implicit knowledge in data in a data-driven way. Due to their powerful characterization capabilities, they are widely used in many fields such as image processing, speech recognition, and classification and identification of different substances. The present invention proposes to combine Raman spectroscopy technology with deep learning, collect Raman spectral data of clinical strains, and quickly detect GBS and highly toxic strains of III / ST17. Summary of the invention

[0006] In order to overcome the deficiencies in the prior art, the present invention provides a detection method for Streptococcus agalactiae and highly virulent strains based on deep learning combined with Raman spectroscopy.

[0007] Raman spectroscopy can reflect the energy of chemical bonds and judge the structure of compounds. The single-cell Raman spectrum of bacteria can reflect the composition and concentration of biomacromolecules and metabolites in bacteria, cell membranes, and cell walls. The cell wall structures of high-virulence and low-virulence Streptococcus agalactiae are significantly different, so their Raman spectra also have corresponding characteristics. However, the difference in Raman spectra is difficult to observe with the naked eye. Therefore, the present invention uses a deep learning method to train the Raman spectra of bacteria, thereby establishing an algorithm model that can correctly identify and classify. Bacteria are divided into three categories: GBS-HIGH (high-virulence strain of Streptococcus agalactiae), GBS-OTHERS (non-high-virulence strain of Streptococcus agalactiae) and ECO (Escherichia coli). The training group data is subjected to deep learning analysis, and then a deep learning analysis model is established based on the analysis results. Finally, the validation group data is used for verification, the model parameters are adjusted, and finally the Raman spectrum of the bacteria to be tested is used for testing.

[0008] In order to achieve the above-mentioned invention object, the technical solution adopted to solve the technical problem is as follows:

[0009] A method for detecting Streptococcus agalactiae and highly virulent strains based on deep learning combined with Raman spectroscopy, comprising the following steps:

[0010] Step 1: Establish a Raman spectrum database;

[0011] Step 2: Establish a typing model for GBS and III / ST17 highly virulent strains;

[0012] Step 3: Evaluate the performance of the typing model for GBS and III / ST17 hypervirulent strains.

[0013] Step 1 includes the following:

[0014] Step 1.1 Strain identification: Resuscitate the clinical isolates to Columbia blood agar medium, culture overnight, pick an appropriate amount of single colonies, and evenly smear them on the target point of the MALDI target plate to form a thin layer. Cover the sample point with 1 μl HCCA standard solvent and dry it naturally at room temperature. Place the target plate into matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF-MS) and perform bacterial identification according to the automatic identification process. When the identification result score is ≥2.0 and the consistency classification is A, the strain is confirmed to be GBS;

[0015] Step 1.2 Determination of GBS III / ST17 highly virulent strains: Extract DNA from the strains verified in step 1.1, and use a BioDrop UV spectrophotometer to detect the purity and concentration of the extracted DNA. The absorbance A260 / A280 is 1.8~2.0, and the DNA concentration is >50 ng / μl. Use a 150 bp double-end sequencing strategy for whole genome sequencing, and use Unicycler to splice the original short-read sequence data for sequencing, compare and search with the corresponding database, and divide GBS into two groups: III / ST17 highly virulent strains and non-III / ST17 strains according to the typing information;

[0016] Step 1.3 Strain processing flow: Treat GBS III / ST17 highly virulent strain group, GBS non-III / ST17 strain and control group strains, pick an appropriate amount of colonies and wash them, then suspend them in 500 μl sterile deionized water in a 1.5 ml centrifuge tube, centrifuge at 12000 rpm / min for 3 min, repeat the washing 3 times until the turbidity reaches 0.5 McFarland, and finally keep 200 μl for use.

[0017] Step 1.4 Establishment of Raman spectroscopy database: Use laser confocal Raman spectrometer to detect GBS III / ST17 high-virulence strain group, GBS non-III / ST17 strain group and Escherichia coli group to stimulate the formation of different Raman spectra, collect and save them.

[0018] Sample pretreatment before collecting Raman spectra: 2 μl of sample solution was taken and dropped on an aluminum sheet to form a droplet with a diameter of about 2 mm, and then dried naturally.

[0019] The parameters for Raman spectrum collection are: laser wavelength: 532 nm; microscope objective magnification: 100x (numerical aperture NA = 0.9); laser power: 1%; grating line density: 1200 gr / mm; signal collection time: 50 seconds, single time; pinhole aperture size: 300 μm; slit width: 200 μm; signal collection frequency band range: 500cm -1-2000cm -1.

[0020] Step 2 includes the following:

[0021] Step 2.1 The Raman spectral data set collected in step 1 is grouped into the following groups: GBS III / ST17 highly virulent strain, GBS non-III / ST17 strain, and Escherichia coli group. Each group is randomly divided into a training set, a validation set, and a test set in a ratio of 7:1:2;

[0022] Step 2.2 Dataset processing: Self-normalization is used as preprocessing for the Raman spectra collected in step 1;

[0023] Step 2.3 Model structure design: Based on the ResNet18 residual neural network architecture, the 2D convolution in the network is changed to 1D convolution suitable for 1-dimensional Raman spectroscopy, the neural network structure model is improved, and the optimal parameters are determined through simulation experiments;

[0024] Step 2.4 Model training: Use cross entropy as the loss function and ADAM as the training optimizer to train the model weights in the training set, then select the optimal hyperparameters based on the results of the model on the validation set to determine the trained model;

[0025] Step 2.5 Model testing: For the trained convolutional neural network model, use the test set to test the classification effect of the model.

[0026] Step 3 includes the following:

[0027] The laboratory specificity and accuracy of the constructed typing model were evaluated using the GBS III / ST17 highly virulent strain, GBS non-III / ST17 strain, Escherichia coli and blank control group collected subsequently. The constructed typing model was evaluated for application using 8 clinical strains to evaluate the sensitivity, specificity and accuracy of the typing model. The evaluation was based on the cerebrospinal fluid culture results, and the divergent results were verified using the metagenomics.

[0028] Due to the adoption of the above technical solution, the present invention has the following advantages and positive effects compared with the prior art:

[0029] 1. The present invention overcomes the traditional method of pathogen phenotypic identification, innovatively uses laser confocal Raman spectrometer to construct a typing model for GBS and III / ST17 highly virulent strains, and realizes the rapid identification of GBS and III / ST17 highly virulent strains, which has the advantages of being simple, rapid and highly sensitive;

[0030] 2. The present invention overcomes the shortcomings of traditional analysis methods that require complex preprocessing of Raman spectra. It uses the deep learning method of convolutional neural networks to innovatively establish a Raman fingerprint database of GBS and highly virulent strains of III / ST17, providing an efficient and convenient rapid diagnostic tool for highly virulent strains of GBS of III / ST17. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the following briefly introduces the drawings required for describing the embodiment, wherein:

[0032] Figure 1 The Raman spectra of GBS III / ST17 highly virulent strain, non-III / ST17 strain and Escherichia coli;

[0033] Figure 2 This is a PCA analysis diagram of three types of data, namely, GBS III / ST17 highly virulent strain, non-III / ST17 strain and Escherichia coli in Example 1 of the present invention.

[0034] Figure 3 This is the identification result of the three types of data of GBS III / ST17 highly virulent strain, non-III / ST17 strain and Escherichia coli in Example 1 of the present invention. DETAILED DESCRIPTION

[0035] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with specific embodiments.

[0036] like Figure 1 As shown, this embodiment discloses a method for detecting Streptococcus agalactiae and highly virulent strains based on deep learning combined with Raman spectroscopy, comprising the following steps:

[0037] Step 1: Establish a Raman spectrum database;

[0038] Step 1 includes the following:

[0039] Step 1.1 Strain identification: Resuscitate the clinical isolates to Columbia blood agar medium, culture overnight, pick an appropriate amount of single colonies, and evenly smear them on the target point of the MALDI target plate to form a thin layer. Cover the sample point with 1 μl HCCA solution and dry it naturally at room temperature. Put the target plate into MALDI-TOF-MS and perform bacterial identification according to the automatic identification process. When the identification result score is ≥2.0 and the consistency classification is A, the strain is confirmed to be GBS;

[0040] Step 1.2 Determination of GBS III / ST17 highly virulent strains: Extract DNA from the strains compounded in step 1.1, and use a BioDrop UV spectrophotometer to detect the purity and concentration of the extracted DNA. The absorbance A260 / A280 is 1.8~2.0, and the DNA concentration is >50 ng / μl. Use a 150 bp double-end sequencing strategy for whole genome sequencing, and use Unicycler to splice and sequence the original short-read sequence data, compare and search with the corresponding database, and divide GBS into two groups: III / ST17 highly virulent strains and non-III / ST17 strains according to the typing information;

[0041] Step 1.3 Strain processing flow: Treat the GBS III / ST17 highly virulent strain group, GBS non-III / ST17 strain and Escherichia coli group, pick an appropriate amount of colonies and suspend them in 500 μl of sterile deionized water in a 1.5 ml centrifuge tube, centrifuge at 12000 rpm / min for 3 min, repeat washing 3 times to make the turbidity reach 0.5 McFarland, and finally keep 20 μl for use.

[0042] Step 1.4 Raman spectroscopy database establishment: Use a laser confocal Raman spectrometer to detect the GBS III / ST17 high-virulence strain group, GBS non-III / ST17 strain group and Escherichia coli group, including the bacterial suspension in step 1.3, to stimulate the formation of different Raman spectra, which are collected and stored.

[0043] Sample pretreatment before collecting Raman spectra: 2 μl of sample solution was taken and dropped on an aluminum sheet to form a droplet with a diameter of about 2 mm, and then dried naturally.

[0044] The parameters for Raman spectrum collection are: laser wavelength: 532 nm; microscope objective magnification: 100x (numerical aperture NA = 0.9); laser power: 1%; grating line density: 1200 gr / mm; signal collection time: 30 seconds, single time; pinhole aperture size: 300 μm; slit width: 200 μm; signal collection frequency band range: 500 cm -1 -2000 cm -1 .

[0045] Step 2: Establish a typing model for GBS and III / ST17 highly virulent strains;

[0046] Step 2 includes the following:

[0047] Step 2.1 The Raman spectral data set collected in step 1 is grouped into the following groups: GBS III / ST17 highly virulent strain, GBS non-III / ST17 strain, and Escherichia coli. Each group is randomly divided into a training set, a validation set, and a test set in a ratio of 7:1:2;

[0048] Step 2.2 Dataset processing: Self-normalization is used as preprocessing for the Raman spectra collected in step 1;

[0049] Step 2.3 Model structure design: Based on the ResNet18 residual neural network architecture, the 2D convolution in the network is changed to 1D convolution suitable for 1-dimensional Raman spectroscopy, and the optimal parameters are determined through simulation experiments;

[0050] Step 2.4 Model training: Use cross entropy as the loss function and ADAM as the training optimizer to train the model weights in the training set, then select the optimal hyperparameters based on the results of the model on the validation set to determine the trained model;

[0051] Step 2.5 Model testing: For the trained convolutional neural network model, the classification effect of the model was tested using the test set. Specifically, the PCA function of the sklearn package in the pytyhon software was used to perform PCA analysis on the last layer of features extracted by the deep neural network, and deep learning analysis was performed on the test group data to identify highly virulent Streptococcus agalactiae type III / ST17.

[0052] Step 3: Evaluate the performance of the typing model for GBS and III / ST17 highly virulent strains;

[0053] Step 3 includes the following:

[0054] The laboratory specificity and accuracy of the constructed typing model were evaluated using the GBS III / ST17 highly virulent strain, GBS non-III / ST17 strain, and Escherichia coli group collected subsequently. The constructed typing model was evaluated for application using 8 clinical strains to assess the sensitivity, specificity, and accuracy of the typing model. The evaluation was based on the cerebrospinal fluid culture results, and the divergent results were verified by metagenomic sequencing.

[0055] Figure 1 The Raman spectra differences between GBS III / ST17 highly virulent strain, non-GBS III / ST17 and Escherichia coli are listed, where the horizontal axis represents Raman shift (Raman shift / cm -1 ), the vertical axis represents Raman intensity (Intensity), black represents Escherichia coli, red represents GBS III / ST17 highly virulent strain, and blue represents the Raman spectra of non-III / ST17 strains.

[0056] Figure 2 This is a PCA analysis diagram of three types of data, namely, GBS III / ST17 highly virulent strain, non-III / ST17 strain and Escherichia coli in Example 1 of the present invention.

[0057] The data is input into the established model, and deep learning analysis is performed on the test group data to identify whether it is a highly virulent strain of GBS type III / ST17. When the test data is tested, the model outputs the percentage of the test data identified as highly virulent strains. When a strain is identified as a highly virulent strain of GBS type III / ST17 by more than 80%, it means that the strain is a highly virulent strain of GBS type III / ST17. The identification accuracy rate is 93.18%. Figure 3 shown.

[0058] It should be 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 preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for rapidly identifying Streptococcus agalactiae and its highly virulent strains, characterized in that: (1) Sample pretreatment: The bacterial samples were divided into three groups: highly virulent strains of Streptococcus agalactiae, non-hypervirulent strains of Streptococcus agalactiae, and Escherichia coli. The bacterial samples were dropped on the surface of an aluminum sheet, dried naturally, and the Raman spectra were collected. (2) Establishment of Raman spectroscopy database: Use laser confocal Raman spectrometer to detect three groups of bacterial samples: highly virulent strains of Streptococcus agalactiae, non-hypervirulent strains of Streptococcus agalactiae, and Escherichia coli to stimulate the formation of different Raman spectra, which are collected and stored; (3) Establishing a deep learning analysis model: randomly divide three groups of bacterial samples, namely, the highly virulent strain of Streptococcus agalactiae, the non-hypervirulent strain of Streptococcus agalactiae and Escherichia coli, into a training set, a validation set and a test set in a ratio of 7:1:2; use self-normalization as a preprocessing for the Raman spectra collected in step (2); based on the ResNet18 residual neural network architecture, change the 2D convolution in the network to a 1D convolution suitable for 1-dimensional Raman spectra, and determine the optimal parameters through simulation experiments; train the weights of the model in the training set, and then select the optimal hyperparameters based on the results of the model on the validation set to determine the trained model; for the trained convolutional neural network model, use the test set to test the classification effect of the model, and evaluate the performance of the typing model for the highly virulent strain of Streptococcus agalactiae, the non-hypervirulent strain of Streptococcus agalactiae and Escherichia coli; (4) performing sample pretreatment on the sample to be tested according to step (1), collecting Raman spectra according to step (2), and identifying the sample to be tested using the deep learning analysis model of step (3); The highly virulent strain of Streptococcus agalactiae is a serotype III, sequence type 17 highly virulent Streptococcus agalactiae; In step (2), a laser confocal Raman spectrometer was selected, a 532 nm laser was used, the laser power was set to 1%, and the spectrometer grating was selected to be 1200 g / mm; in step (2), the spectral range was 500-2000 cm -1 .

2. The method for rapid identification of Streptococcus agalactiae and its highly toxic strains according to claim 1, characterized in that: In step (1), the bacteria sample is suspended in sterile water with a turbidity of 0.5 McFarland, and 2 μl is dropped on the surface of an aluminum sheet and allowed to dry naturally.

3. The method for rapid identification of Streptococcus agalactiae and its highly toxic strains according to claim 1 or 2, characterized in that: In step (2), a single cell of the bacterial sample to be tested was found under a 100x objective lens, the single spectrum acquisition time was set to 50 s, and 100 spectra were collected for each bacterial strain in the training group.

4. The method for rapid identification of Streptococcus agalactiae and its highly toxic strains according to claim 1 or 2, characterized in that: In step (2), the pinhole aperture size is 300 μm and the slit width is 200 μm.

5. The method for rapid identification of Streptococcus agalactiae and its highly toxic strains according to claim 1 or 2, characterized in that: In step (3), the PCA function of the sklearn package in the pytyhon software is used to perform PCA analysis on the last layer of features extracted by the deep neural network, and deep learning analysis is performed on the test group data to identify highly virulent Streptococcus agalactiae.

6. The method for rapid identification of Streptococcus agalactiae and its highly toxic strains according to claim 1 or 2, characterized in that: In step (3), the weight of the model is trained in the training set by using cross entropy as the loss function and ADAM as the training optimizer to train the weight of the model in the training set.

7. The method for rapid identification of Streptococcus agalactiae and its highly toxic strains according to claim 1 or 2, characterized in that: In step (4), the data of the sample to be tested is input into the trained convolutional neural network model, and the model will output the percentage of the sample to be tested that is identified as a highly virulent strain. When the percentage is greater than 80%, it is identified as a highly virulent Streptococcus agalactiae.

Citation Information

Patent Citations

  • Spectroscopic means and methods for identifying microorganisms in culture

    CN104136908A

  • Method for rapidly detecting eight bacillus marinus and application of method in fermented food

    CN115343272A

  • Method for rapidly detecting 10 yeasts and application of method in fermented food

    CN115343273A

  • Rice bacterial leaf blight and rice bacterial leaf streak pathogen detection method based on Raman spectrum

    CN117007575A

  • Raman spectrum bacteria classification method and system based on contrast learning

    CN117219176A