Label-free SERS (Surface Enhanced Raman Scattering) detection method for rapidly identifying enterohemorrhagic Escherichia coli O157 strain

Abstract

The invention belongs to the field of microbiological detection, and particularly relates to an SERS (Surface Enhanced Raman Scattering) detection method for rapidly identifying enterohemorrhagic Escherichia coli O157: H7 strains, which mainly comprises the following steps: (1) collecting SERS spectrums of different strains and different batches of Escherichia coli; (2) using SERS spectrums of different strains as a training set and a test set, optimizing and investigating prediction models learned by different types of machines, and selecting an optimal model; (3) separating and enriching a to-be-detected sample, and collecting an SERS spectrum of the to-be-detected sample; and (4) inputting the SERS spectrum of a to-be-detected sample into the optimal machine learning model to realize rapid identification of the enterohemorrhagic Escherichia coli O157: H7. According to the method, the enterohemorrhagic specific strain of the escherichia coli can be specifically recognized, no recognition unit needs to be marked, the method has the advantages of being easy and convenient to operate and low in cost, and the detection accuracy is 99% or above.

Description

Technical Field

[0001] The invention belongs to the field of medicine or food detection, and specifically relates to a SERS detection method for rapidly identifying enterohemorrhagic Escherichia coli O157 strain (Escherichia coli O157:H7 strain). Background Art

[0002] Enterohemorrhagic Escherichia coli O157 (EHEC) is an important foodborne pathogen that can be transmitted to humans and animals. It can cause infection at very low doses. Infection may lead to serious diseases such as hemorrhagic colitis, hemolytic anemia, and thrombocytopenic purpura, which poses a greater threat to the elderly, children, and those with immunodeficiency. Therefore, timely detection is a key means to prevent and control EHEC infection.

[0003] At present, conventional EHEC detection methods mainly include plate culture method, immunological method and molecular biology method. As the gold standard, the plate culture method is highly reliable, but it takes a long time and is difficult to meet clinical and rapid testing needs. Immunological methods have attracted widespread attention due to their convenience, rapidity, strong specificity and low cost, but they require the use of recognition units such as labeled antibodies, which has high detection costs and complex operations. Molecular biological methods have high sensitivity, but the complex biological matrix components will affect the purity of DNA extraction. At the same time, their expensive equipment requirements and complex primer design limit the popularization and application of this technology. Therefore, how to develop a fast, accurate and cost-effective EHEC detection method has become a challenge that needs to be solved urgently.

[0004] Surface-enhanced Raman spectroscopy (SERS) has shown unique advantages in pathogen detection due to its ultra-high sensitivity, portability, rapid detection capability and relatively low cost. Researchers have developed SERS-based EHEC detection methods, especially antibody / aptamer-based labeled SERS methods, which have high sensitivity and specificity, but are complex to operate and have high detection costs (Analytica Chimica Acta, 2019, 1081: 146-156; Frontiers in Microbiology, 2020, 11: 596005; Food Chemistry, 2023, 424: 136433.). In contrast, the label-free SERS detection method is simpler and can be directly detected through the SERS fingerprint of the bacteria to be tested. There are many reports on label-free SERS detection methods for E. coli (Talanta, 2016, 146: 457-463; Frontiers in Microbiology, 2023, 14: 1101357; Analyst, 2016, 141 (4): 1356-1362; CN202010480668; CN202010020498). However, due to the genetic differences between different strains of E. coli, their SERS spectra may also show significant differences. This means that although the SERS spectral data of E. coli have been collected, these data are difficult to directly apply to the specific detection of EHEC. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the prior art and provide a new method for analyzing enterohemorrhagic Escherichia coli O157:H7 which is fast, simple, stable and reliable.

[0006] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is as follows:

[0007] (1) Collecting SERS spectra of different strain types and batches of Escherichia coli;

[0008] (2) performing spectral preprocessing on the collected bacterial SERS spectra;

[0009] (3) Divide the SERS spectra of E. coli of different strain types into training sets and test sets;

[0010] (4) Use different machine learning algorithms to build a binary classification model, and optimize the best classification algorithm through classification accuracy evaluation;

[0011] (5) After the sample pretreatment step, collecting the SERS spectrum of the sample to be tested;

[0012] (6) The SERS spectrum of the sample to be tested obtained in step (5) is input into the binary classification model established in step (4) after spectral preprocessing to obtain the identification result.

[0013] In step (1), there are at least 10 different strain types, and the collected SERS spectra need to contain 3 fingerprint spectrum patterns to examine the SERS spectra of all strain types of Escherichia coli.

[0014] The different batches in step (1) are at least 3 independent batches, and at least 30 bacterial SERS spectra are collected from each batch to fully investigate the variation characteristics of the SERS spectra;

[0015] The different strains in step (1) need to contain the enterohemorrhagic Escherichia coli O157:H7 strain. This step collects the SERS spectrum of the enterohemorrhagic Escherichia coli O157:H7 strain to improve the specificity of the method.

[0016] The spectral preprocessing in step (2) mainly includes spectral smoothing, baseline correction and normalization;

[0017] In step (3), the data is randomly divided into a training set and a test set, wherein the training set accounts for 70% to 80% of the total data set, and the number of the test set accounts for 20% to 30% of the total data set;

[0018] Before data division in step (3), the SERS spectra of enterohemorrhagic Escherichia coli and other Escherichia coli need to be labeled (Label);

[0019] The SERS spectra of pathogenic bacteria in steps (1) and (5) are obtained by using a silicon-based SERS substrate material and under irradiation with a 785 nm excitation light source;

[0020] The classification accuracy in step (4) is achieved by evaluating the classification efficiency (Efficient rate, EFR) of the test set.

[0021] The EFR calculation formula described in step (4) is as follows:

[0022]

[0023] In the above formula, TP and TN are the number of samples identified as true positive and true negative, and FP and FN are the number of samples identified as false positive and false negative.

[0024] The best machine learning classification model in step (4) is the one with the best EFR value;

[0025] The binary classification machine learning algorithms in step (4) are mainly extreme learning machine (ELM), decision tree (Decision Tree), random forest (RF), linear discriminant analysis (LDA), SoftMax classifier (SMC), stacked autoencoder (SAE), support vector machine (SVM), Naive Bayes model (NBM), K nearest neighbor (KNN), generalized regression neural network (GRNN) and at least one of them;

[0026] In step (5), the SERS spectrum of the sample to be tested is collected using at least three independent SERS substrate materials, and the total number of spectra is not less than 15;

[0027] In step (6), only when 90% of the SERS spectra of the samples to be tested are classified as enterohemorrhagic Escherichia coli O157:H7, the strain to be tested is determined to be enterohemorrhagic Escherichia coli O157:H7.

[0028] Compared with the prior art, the present invention has the following advantages:

[0029] (1) The present invention summarizes the differences in SERS spectra of different strains of E. coli and successfully distinguishes the SERS spectra of EHEC O157 from other E. coli, which can be used for rapid detection of specific strains;

[0030] (2) The method does not require any identification unit to be marked, has a low detection cost, and is relatively easy to operate;

[0031] (3) The identification accuracy of the method of the present invention is high, reaching more than 99%. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 The SERS spectra of E. coli in three different modes;

[0033] Figure 2 The PCA analysis results of SERS spectra of different E. coli species;

[0034] Figure 3 Flow chart of the SERS detection method of the present invention. DETAILED DESCRIPTION

[0035] The specific implementation modes of the present invention are further described in detail below in conjunction with the accompanying drawings.

[0036] Embodiment 1:

[0037] This example provides a SERS identification method for Escherichia coli, which is specifically carried out according to the following steps:

[0038] (1) Eleven different strains of Escherichia coli were taken, and six batches of SERS spectra of the different strains were collected using a silicon-based SERS substrate material, wherein at least 30 SERS spectra were collected for each batch;

[0039] (2) The collected E. coli SERS spectra were spectrally smoothed and baseline corrected to obtain a total E. coli SERS spectrum data set, which showed three different fingerprint patterns: Figure 1 );

[0040] (3) The SERS spectrum of enterohemorrhagic Escherichia coli O157:H7 is marked as "1", and the SERS spectra of other Escherichia coli are marked as "2";

[0041] (4) Randomly divide 70% of the total data set into a training set and 30% into a test set;

[0042] (5) Use KNN, LDA, NBM, DT, RF, ELM, SVM, GRNN, SAE and SMC algorithms to establish a binary classification model for E. coli, calculate the classification efficiency EFR of the test set, and select the best classification algorithm. Calculations show that the recognition rates of KNN, LDA, NBM, DT, RF, ELM, SVM, GRNN, SAE and SMC algorithms for EHEC are 97.29%, 60.09%, 55.11%, 75.47%, 93.76%, 71.00%, 97.09%, 97.33%, 98.76% and 99.71%, respectively. From the results, it can be seen that the SoftMax classifier has the best classification effect, so the SoftMax classifier is selected as the best classifier;

[0043] (6) taking three samples of the bacteria to be tested (one sample is an EHEC strain and the other two samples are other Escherichia coli strains), culturing and treating the bacteria, and using a silicon-based SERS substrate material to collect bacterial SERS spectra, wherein 15 spectra are collected for each strain;

[0044] (7) The SERS spectrum obtained in step (6) is input into the classification model established in step (5) after spectrum smoothing and baseline correction to obtain the identification result with an accuracy of 100%.

[0045] In summary, the present invention combines a label-free SERS detection method and a machine learning classification algorithm to propose a rapid identification method for enterohemorrhagic Escherichia coli, with an identification accuracy rate of more than 99%. The method of the present invention is easy to operate, stable and reliable, and low in cost. In particular, if combined with a portable Raman spectrometer, it can be used for on-site detection in the fields of medicine, food, etc. In addition, the above invention is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any relevant personnel familiar with the technical field, within the technical scope disclosed by the present invention, shall make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention, which shall be covered within the protection scope of the present invention.

Claims

1. A label-free SERS detection method for rapid identification of enterohemorrhagic Escherichia coli O157:H7 strains, characterized in that: The following steps are involved: (1) Collecting SERS spectra of Escherichia coli of different strain types and batches; (2) performing spectral preprocessing on the collected bacterial SERS spectra; (3) Divide the SERS spectra of E. coli of different strain types into training sets and test sets; (4) Use different machine learning algorithms to build binary classification models, evaluate the classification performance of different models through classification accuracy, and optimize the best classification algorithm; (5) After the sample pretreatment step, collecting the SERS spectrum of the sample to be tested; (6) The SERS spectrum of the sample to be tested obtained in step (5) is input into the binary classification model established in step (4) after spectral preprocessing to obtain the identification result.

2. The label-free SERS detection method according to claim 1, characterized in that: In step (1), the number of different strain types is at least 10, and the SERS spectrum collected needs to contain 3 fingerprint patterns; The different batches in step (1) are at least 3 independent batches, and at least 30 bacterial SERS spectra are collected from each batch; The different strains in step (1) need to contain the enterohemorrhagic Escherichia coli O157:H7 strain.

3. The label-free SERS detection method according to claim 1, characterized in that: The spectral preprocessing in step (2) mainly includes spectral smoothing, baseline correction and normalization.

4. The label-free SERS detection method according to claim 1, characterized in that: In step (3), the data are randomly divided into a training set and a test set, wherein the training set accounts for 70% to 80% of the total data set, and the number of test sets accounts for 20% to 30% of the total data set.

5. The label-free SERS detection method according to claim 1, characterized in that: Before data division in step (3), the SERS spectra of enterohemorrhagic Escherichia coli and other Escherichia coli need to be labeled.

6. The label-free SERS detection method according to claim 1, characterized in that: The SERS spectra of pathogenic bacteria in step (1) and step (5) are obtained by using a silicon-based SERS substrate material and under irradiation with a 785nm excitation light source.

7. The label-free SERS detection method according to claim 1, characterized in that: The classification accuracy in step (4) is achieved by evaluating the classification efficiency of the test set.

8. The label-free SERS detection method according to claim 1, characterized in that: The binary classification machine learning algorithm in step (4) is at least one of extreme learning machine, decision tree, random forest, linear discriminant analysis, SoftMax classifier and stacked autoencoder, support vector machine, naive Bayes, k-nearest neighbor algorithm, and generalized regression neural network.

9. The label-free SERS detection method according to claim 1, characterized in that: In step (5), the SERS spectrum of the sample to be tested is collected using at least three independent SERS substrate materials, and the total number of spectra is not less than 15.

10. The label-free SERS detection method according to claim 1, characterized in that: In step (6), when 90% of the SERS spectra of the samples to be tested are classified as enterohemorrhagic Escherichia coli O157:H7, the strain to be tested is determined to be enterohemorrhagic Escherichia coli O157:H7.

Images