Gastric cancer intestinal microbial marker, application thereof and method for constructing gastric cancer detection model

Abstract

The invention provides a gastric cancer intestinal microbial marker, application thereof and a method for constructing a gastric cancer detection model, and belongs to the technical field of biomedicine. The method comprises the following steps: extracting biomarkers related to gastric cancer pathogenesis and progress from metagenome data, constructing a random forest model by using a random forest algorithm, and training and testing the constructed random forest model by using a training set and a test set to obtain a gastric cancer detection model. According to the gastric cancer detection model constructed based on the gastric cancer intestinal microbial marker, the specificity is 97.2964%, the sensitivity is 94%, the accuracy is 93.40649%, the model efficiency and conformity are high, and the condition of gastric cancer can be accurately predicted and evaluated. Based on the gastric cancer intestinal microbiological marker and the gastric cancer detection model, the disease condition of a gastric cancer patient can be accurately predicted and evaluated, a doctor can make a more effective treatment scheme, the survival rate and life quality of the patient are improved, and the treatment effect is improved.

Description

Technical Field

[0001] The present invention relates to the field of biomedical technology, and in particular to a gastric cancer intestinal microbial marker and its application and a method for constructing a gastric cancer detection model. Background Art

[0002] Gastric cancer originates from gastric epithelial cells and is the most common malignant tumor of the digestive tract. It is also one of the most common malignant tumors worldwide, with high morbidity and mortality. According to data from the World Health Organization (WHO), there are about 1 million new cases of gastric cancer worldwide each year. In addition, early gastric cancer may have no obvious symptoms, or only non-specific symptoms such as upper abdominal discomfort and belching may occur, which are not easy to attract patients' attention, delaying the treatment of gastric cancer and causing the condition of gastric cancer patients to worsen.

[0003] At present, the main treatments for gastric cancer include surgery, chemotherapy, radiotherapy, targeted therapy, and traditional Chinese medicine. Diagnosing patients with gastric cancer in the early stages is an important means to improve the survival rate of gastric cancer patients. In recent years, studies have found that the gastric microecology is closely related to the occurrence and development of gastric cancer. Helicobacter pylori, other gastric flora, actinomycetes, fungi, viruses, etc. can cause the occurrence or further development of gastric cancer through various mechanisms. After gastric cancer patients undergo gastrectomy, their intestinal microbial flora will change significantly. This change may be related to changes in the patient's gastrointestinal environment after surgery and may affect the patient's prognosis. Therefore, the intestinal microbial flora is significantly related to the pathogenesis of gastric cancer, but the role of the intestinal microbial flora in the occurrence of early gastric cancer has not yet been clarified.

[0004] Currently, there is no marker with very high specificity and sensitivity that can be used for early screening of gastric cancer. Although some tumor markers such as CEA and CA19-9 may be elevated in gastric cancer patients, they are not specific and may also be elevated in other diseases. Therefore, to improve the early diagnosis rate and treatment effect of gastric cancer, further research and development of more sensitive and specific biomarkers are needed. Summary of the invention

[0005] The purpose of the present invention is to provide a gastric cancer intestinal microbial marker and its application and a method for constructing a gastric cancer detection model. By analyzing the composition and functional changes of the patient's intestinal microorganisms, signs of gastric cancer can be discovered earlier and the treatment effect can be monitored.

[0006] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0007] The present invention provides a gastric cancer intestinal microbial marker, wherein the intestinal microbial marker comprises the following species: Fusicatenibacter_saccharivorans, Streptococcus_anginosus_group, Bifidobacterium_dentium, Eubacterium_sp_CAG_274, Lactobacillus_salivarius, Lachnospira_pectinoschiza, Streptococcus_salivarius, Veillonella_parvula, Eubacterium_hallii, Anae rostipes_hadrus), Escherichia_coli, Streptococcus_oralis, Roseburia_faecis, Parabacteroides_merdae, Actinomyces_sp_ICM47, Streptococcus_gordonii, Blautia_wexlerae, Streptococcus_vestibularis, Streptococcus_parasanguinis, Roseburia_inulinivorans, Lactobacillus_mucosae.

[0008] The present invention also provides an application of a gastric cancer intestinal microbial marker, which includes a product for preparing a diagnosis of whether a patient has gastric cancer or the risk of gastric cancer, a target for gastric cancer condition assessment and treatment, and a gastric cancer detection model; the product includes any one of a detection reagent, a detection kit, and a detection chip.

[0009] The present invention also provides a method for constructing a gastric cancer detection model, comprising the following steps:

[0010] (1) Total DNA was extracted from the feces of gastric cancer patients and healthy people, and high-throughput sequencing was performed after pretreatment. The sequencing results were annotated with species and analyzed for composition, gene abundance, function annotation, and difference analysis;

[0011] (2) Using random forest to screen gastric cancer intestinal microbial markers, constructing a random forest model, optimizing parameters, and training and testing the constructed random forest model with a training set and a test set to obtain a gastric cancer detection model;

[0012] (3) The metagenome and abundance information of the patient to be tested are input into the gastric cancer detection model for identification.

[0013] Furthermore, in step (1), the species annotation and composition spectrum analysis includes: alpha diversity analysis, beta diversity analysis; flora sample composition, sample heatmap, and species interaction network analysis.

[0014] Furthermore, in step (1), the gene abundance analysis includes gene Core-Pan curve, gene dilution curve, and correlation analysis between samples of gene numbers.

[0015] Furthermore, in step (1), the functional annotation includes: KEGG data rate functional annotation analysis, HUMAnN3: quantitative pathway abundance, KO functional abundance and analysis, Module functional abundance and analysis, Pathway functional abundance and analysis.

[0016] Furthermore, in step (1), the difference analysis includes: Lefse analysis, dominant species difference, multivariate variance adonis, MaAsLin, and rank sum test difference.

[0017] Furthermore, in step (2), the parameters for optimizing the parameters are determined according to GridSearchCV.

[0018] The beneficial effects of the present invention compared with the prior art are:

[0019] (1) The present invention extracts biomarkers related to the pathogenesis and progression of gastric cancer from metagenomic data, uses a random forest algorithm to construct a prediction model, uses a random forest algorithm to construct a random forest model, and trains and tests the constructed random forest model with a training set and a test set to obtain a gastric cancer detection model. Random forest is a powerful machine learning algorithm that can process high-dimensional data and performs well in processing nonlinear relationships and complex interactions. It can divide data by randomly selecting features, which can increase the generalization ability and resistance to overfitting of the model.

[0020] (2) The gastric cancer detection model of the present invention has a specificity of 97.2964%, a sensitivity of 94%, and an accuracy of 93.40649%. It has high model efficiency and conformity, and can accurately predict and evaluate the condition of gastric cancer. In addition, the ROC curve of the gastric cancer detection model is close to the upper left corner, and has a high true positive rate at each threshold. The AUC value reaches 0.8963, indicating that the classifier performs well in distinguishing positive and negative samples.

[0021] (3) The gastric cancer detection model of the present invention can accurately predict and evaluate the condition of gastric cancer patients, help doctors formulate more effective treatment plans, improve the survival rate and quality of life of patients, and provide personalized treatment for gastric cancer patients to improve the treatment effect. It can also detect early gastric cancer and find signs of gastric cancer earlier, achieving the effect of early diagnosis and early treatment, and can also monitor the monitoring and treatment effect of gastric cancer. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0023] Figure 1 A flow chart for constructing a gastric cancer detection model in Example 1;

[0024] Figure 2 This is a graph showing the results of ten cross-validation runs in Example 1;

[0025] Figure 3 It is the "Mean DecreaseAccuracy" indicator in the random forest model of Example 1;

[0026] Figure 4 This is the ROC curve in Example 1. Figure 5 This is the judgment standard of the gastric cancer detection model in Example 1. DETAILED DESCRIPTION

[0027] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as limiting the present invention, but should be understood as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0028] It should be understood that the terms described in the present invention are only for describing special embodiments and are not intended to limit the present invention. In addition, for the numerical range in the present invention, it should be understood that each intermediate value between the upper and lower limits of the scope is also specifically disclosed. Each smaller range between the intermediate value in any stated value or stated range and any other stated value or intermediate value in the described range is also included in the present invention. The upper and lower limits of these smaller ranges can be independently included or excluded in the scope.

[0029] Unless otherwise indicated, all technical and scientific terms used herein have the same meanings as those generally understood by those skilled in the art. Although the present invention describes only preferred methods and materials, any methods and materials similar or equivalent to those described herein may also be used in the implementation or testing of the present invention. All documents mentioned in this specification are incorporated by reference to disclose and describe the methods and / or materials associated with the documents. In the event of a conflict with any incorporated document, the content of this specification shall prevail.

[0030] It will be apparent to those skilled in the art that various modifications and variations may be made to the specific embodiments of the present invention description without departing from the scope or spirit of the present invention. Other embodiments derived from the present invention description will be apparent to the skilled artisan. The present invention description and examples are exemplary only.

[0031] The words “include,” “including,” “have,” “contain,” etc. used in this document are open-ended terms, meaning including but not limited to.

[0032] The present invention provides a method for constructing a gastric cancer detection model, which specifically comprises the following steps:

[0033] (1) Total DNA was extracted from the feces of gastric cancer patients and healthy people, and high-throughput sequencing was performed after pretreatment. The sequencing results were annotated with species and analyzed for composition, gene abundance, function annotation, and difference analysis;

[0034] The species annotation and composition spectrum analysis include:

[0035] Alpha diversity: Shannon, Simpson, Insimpson index analysis;

[0036] beta diversity: Bray-CurtisPCoa / PCA / NMDS / UPGMA analysis;

[0037] Bacteria sample composition, sample heatmap;

[0038] Species interaction network analysis.

[0039] The gene abundance analysis includes gene Core-Pan curve, gene dilution curve, and correlation analysis between samples of gene numbers.

[0040] The functional annotations include: KEGG data rate functional annotation analysis, HUMAnN3: quantitative pathway abundance, KO functional abundance and analysis, Module functional abundance and analysis, Pathway functional abundance and analysis.

[0041] The difference analysis includes: Lefse analysis, dominant species difference, multivariate variance adonis, MaAsLin, and rank sum test difference.

[0042] (2) Using random forest to screen gastric cancer intestinal microbial markers, constructing a random forest model, optimizing parameters, and training and testing the constructed random forest model with a training set and a test set to obtain a gastric cancer detection model;

[0043] Among them, the screened microbial markers include the following species: Fusicatenibacter_saccharivorans, Streptococcus_anginosus_group, Bifidobacterium_dentium, Eubacterium_sp_CAG_274, Lactobacillus_salivarius, Lachnospira_pectinoschiza, Streptococcus_salivarius, Veillonella_parvula, Eubacterium_hallii, Anaerostipes _hadrus), Escherichia coli, Streptococcus oralis, Roseburia faecis, Parabacteroides merdae, Actinomyces sp ICM47, Streptococcus gordonii, Blautia wexlerae, Streptococcus vestibularis, Streptococcus parasanguinis, Roseburia inulinivorans, Lactobacillus mucosae.

[0044] (3) The metagenome and abundance information of the patient to be tested are input into the gastric cancer detection model for identification.

[0045] Example 1

[0046] Embodiment 1 of the present invention provides a method for constructing a gastric cancer detection model, which specifically includes the following steps:

[0047] Fecal samples from 14 gastric cancer patients and 86 healthy controls were collected clinically to ensure that the patients had not received antibiotics or other related treatments that might affect the intestinal flora, and were divided into training and test sets. Figure 1 The process shown uses sterile containers to collect stool samples to avoid cross contamination, and the MagBeads Fecal Genomic Kit is used to extract fecal DNA. Take 300 mg of fecal sample and 1000 μL Pre-Wash Buffer, vortex until the sample is completely uniform; centrifuge the sample at 14000 rpm for 5 minutes, discard the supernatant, add 980 μL Lysis Buffer F1, 10 μL RNase A Solution and 120 μL Lysis Buffer F2 to the Lysing Matrix E tube, and vortex for 10 minutes; centrifuge at 14000 rpm / min for 5 minutes to separate particle fragments; transfer 900 μL supernatant to a 2.0 mL microcentrifuge tube, add 250 μL PPS, and mix by inverting 20 times; centrifuge at 14000 rpm for 5 minutes, transfer 1000 μL supernatant to a new 2.0 mL microcentrifuge tube, and then add 1000 μL Binding Buffer F and 10 μL Magnetic Beads and vortex to mix; place the test tube on the magnetic stand for 3 minutes to allow the magnetic beads to precipitate, and then discard the supernatant; add 1000 μL Wash to the test tube Buffer F1, shake on a shaker for 3 minutes, place the test tube on the magnetic rack for 1 minute to allow the magnetic beads to settle, and then discard the supernatant; add 1000μL Wash Buffer F2 to the test tube, shake on a shaker for 3 minutes, place the test tube on the magnetic rack for 1 minute, and then discard the supernatant; repeat the operation and wash again with Wash Buffer F2; then air-dry the magnetic beads at 55℃; add 100μL TE Buffer and incubate at 55℃ for 5min; place the test tube on the magnetic rack until the magnetic beads settle, transfer the supernatant (eluted DNA) to a clean 1.5mL microcentrifuge tube and store it at -20℃ for a long time.

[0048] Use an ultrasonic disruptor to randomly break high molecular weight DNA into fragments of about 300 bp, repair the ends of the fragmented DNA to make them blunt ends; connect specific adapter sequences to both ends of the repaired DNA fragments, and remove excess adapters, primers and other impurities through magnetic beads or column purification; use specific primers to PCR amplify the DNA fragments with adapters to enrich a sufficient amount of library fragments; use fastp software to perform quality control on the sequencing data, including removing adapters and low-quality sequences; use Bowtie2 to filter out reads that may come from the host, and then use the Illumina platform for high-throughput sequencing. The sequencing results are sorted according to Figure 1 The species annotation and composition spectrum analysis, gene abundance analysis, functional annotation, and differential analysis processes were analyzed, and the results were input into the random forest model. The random forest was used to screen the intestinal microbial markers of gastric cancer, and the random forest model was constructed. The biomarkers of 21 bacterial genera that changed significantly between gastric cancer patients and healthy people were found: Fusicatenibacter_saccharivorans, Streptococcus_anginosus_group, Bifidobacterium_dentium, Eubacterium_sp_CAG_274, Lactobacillus_salivarius, Lachnospira_pectinoschiza, Streptococcus_salivarius, Veillonella_parvula, and Horace Eubacterium hallii, Anaerostipes hadrus, Escherichia coli, Streptococcus oralis, Roseburia faecis, Parabacteroides merdae, Actinomyces sp ICM47, Streptococcus gordonii, Blautia wexlerae, Streptococcus vestibularis, Streptococcus parasanguinis, Roseburia inulinivorans, Lactobacillus mucosae.

[0049] The training set of gastric cancer patients and healthy people was used to train the random forest model using the python software sklearn.Ensemble.RandomForestClassifier module to obtain the gastric cancer detection model; the model training parameters were set as follows: gradient n_estimators = [100,500,1000]; max_depth = [1,2,3,7,9]; max_features = ["log2","sqrt"].

[0050] The stability and reliability of the gastric cancer detection model were verified by ten-fold cross validation. The python software GridSearchCV module was used to search for the optimal parameters. The random forest model was constructed using the optimal parameters. The gastric cancer detection model was obtained by training and testing with the training set and test set. The results are shown in the figure. Figure 2 As shown in Table 1.

[0051] Depend on Figure 2 It can be seen that the vertical axis reflects the value of the cross-validation error, which ranges roughly from 0.10 to 0.30. There are several curves in the line chart, which represent different variable selection methods. The black curve clearly shows the trend of the cross-validation error as the number of variables increases. This curve initially shows a downward trend, indicating that the prediction error of the model is gradually decreasing in the process of adding variables, and both the model training set and the test set have a high accuracy rate, indicating that the model results are good.

[0052] Fusicatenibacter_saccharivorans, Streptococcus_anginosus_group, Bifidobacterium_dentium, Eubacterium_sp_CAG_274, Lactobacillus_salivarius, Lachnospira_pectinoschiza, Streptococcus_salivarius, Veillonella_parvula, Eubacterium_hallii, Anaerostipes_hadrus, Escherichia coli The retention performance of 21 bacterial genera was the best, including Streptococcus cherichia coli, Streptococcus oralis, Roseburia faecis, Parabacteroides merdae, Actinomyces sp ICM47, Streptococcus gordonii, Blautia wexlerae, Streptococcus vestibularis, Streptococcus parasanguinis, Roseburia inulinivorans, and Lactobacillus mucosae.

[0053] Table 1 Gastric cancer detection model indicators

[0054]

[0055]

[0056] In Table 1, "Mean DecreaseAccuracy" and "Mean Decrease Gini" are two important indicators in the random forest model. "Mean DecreaseAccuracy" indicates the degree of reduction in the prediction accuracy of the random forest. The larger the value, the greater the importance of the variable (see the results Figure 3 ). "Mean Decrease Gini" calculates the impact of each variable on the heterogeneity of observations at each node of the classification tree, thereby comparing the importance of the variables. The larger the value, the greater the importance of the variable.

[0057] ROC curve is used to comprehensively evaluate the performance of classification models. The horizontal axis (X-axis) of the ROC curve represents 1-Specificity, that is, the false positive rate, which ranges from 0 to 1 and reflects the proportion of negative examples that the model incorrectly predicts as positive examples. The vertical axis (Y-axis) represents Sensitivity, that is, the true positive rate, which measures the ability of the model to correctly identify positive examples. The results are shown in Figure 4 .

[0058] exist Figure 4 In the figure, the solid red line clearly depicts the ROC curve of the model, while the shaded part represents the confidence interval (CI). The ROC curve of the present invention is close to the upper left corner, and has a higher true positive rate at each threshold. The area under the curve (AUC) is close to 1, indicating that the classification performance of the model is better. The AUC value of the present invention reaches 0.8963, indicating that the classifier performs well in distinguishing positive and negative samples. The CI ranges from 0.8473 to 0.9454, which provides a reliable evaluation of the AUC value, indicating the performance stability of the gastric cancer detection model.

[0059] The metagenome and abundance information of the patient to be tested are input into the gastric cancer detection model, and the patient is judged whether he or she is ill according to the prediction score. The judgment criteria are: when the risk value is 0-0.5, it is low risk; 0.5-0.7 is medium risk; 0.7-1 is high risk. The results are shown in Figure 5 .

[0060] Example 2

[0061] Example 2 of the present invention tested the specificity, sensitivity, and accuracy of the gastric cancer detection model of Example 1, and the specific method is as follows:

[0062] The results of the gastric cancer detection model were compared with the results of the macrogene detection using the formula to obtain the TPR (sensitivity), TNR (specificity), and ACC (accuracy) of the evaluation model.

[0063] TPR, also known as recall rate, measures the model's ability to correctly identify positive samples, TNR measures the model's ability to correctly identify negative samples, and ACC measures the proportion of the model's overall correct predictions.

[0064] The formula is as follows:

[0065] TPR = TP / (TP+FN);

[0066] TNR = TN / (TN + FP);

[0067] ACC=(TP+TN) / (TP+TN+FP+FN);

[0068] Among them, TP (True Positive), TN (True Negative), FP (False Positive) and FN (False Negative).

[0069] The calculation results show that the TPR is 97.2964% specificity, TNR is 94% sensitivity, and ACC is 93.40649% accuracy. The results show that the evaluation model has a high model efficiency and compliance.

[0070] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A gastric cancer intestinal microbial marker, characterized in that: The intestinal microbial markers include the following species: Fusicatenibacter_saccharivorans, Streptococcus_anginosus_group, Bifidobacterium_dentium, Eubacterium_sp_CAG_274, Lactobacillus_salivarius, Lachnospira_pectinoschiza, Streptococcus_salivarius, Veillonella_parvula, Eubacterium_hallii, Anaerostipes_hadrus, Escherichia_coli, Streptococcus oralis, Roseburia faecis, Parabacteroides merdae, Actinomyces sp ICM47, Streptococcus gordonii, Blautia wexlerae, Streptococcus vestibularis, Streptococcus parasanguinis, Roseburia inulinivorans, Lactobacillus mucosae.

2. A use of the gastric cancer intestinal microbial marker according to claim 1, characterized in that: The application includes the preparation of products for diagnosing whether a patient has gastric cancer or the risk of gastric cancer, using them as targets for gastric cancer disease assessment and treatment, and constructing gastric cancer detection models; the product includes any one of a detection reagent, a detection kit, and a detection chip.

3. A method for constructing a gastric cancer detection model using the gastric cancer intestinal microbial markers according to claim 1, characterized in that: The steps include: (1) Total DNA was extracted from the feces of gastric cancer patients and healthy people, and high-throughput sequencing was performed after pretreatment. The sequencing results were annotated with species and analyzed for composition, gene abundance, function annotation, and difference analysis; (2) Use random forest to screen intestinal microbial markers for gastric cancer, construct a random forest model, optimize parameters, and train and test the constructed random forest model using training sets and test sets to obtain a gastric cancer detection model; (3) The metagenome and abundance information of the patient to be tested are input into the gastric cancer detection model for identification.

4. The method for constructing a gastric cancer detection model according to claim 3, characterized in that: In step (1), the species annotation and composition spectrum analysis includes: alpha diversity analysis, beta diversity analysis; microbial sample composition, sample heatmap, and species interaction network analysis.

5. The method for constructing a gastric cancer detection model according to claim 3, characterized in that: In step (1), the gene abundance analysis includes gene Core-Pan curve, gene dilution curve, and correlation analysis between gene numbers in samples.

6. The method for constructing a gastric cancer detection model according to claim 3, characterized in that: In step (1), the functional annotation includes: KEGG data rate functional annotation analysis, HUMAnN3: quantitative pathway abundance, KO functional abundance and analysis, Module functional abundance and analysis, Pathway functional abundance and analysis.

7. The method for constructing a gastric cancer detection model according to claim 3, characterized in that: In step (1), the difference analysis includes: Lefse analysis, dominant species difference, multivariate variance adonis, MaAsLin, and rank sum test difference.

8. The method for constructing a gastric cancer detection model according to claim 3, characterized in that: In step (2), the parameters for optimizing the parameters are determined according to GridSearchCV.

Images