A set of protein biomarkers for early screening of gastric cancer and a logistic regression prediction model constructed therefrom

Abstract

The present application relates to a protein biomarker-based and logistic regression prediction model for gastric cancer screening. The present application can obtain a probability prediction of gastric cancer with high specificity and sensitivity by monitoring whether a set of gastric cancer biomarkers is contained in a sample and using it for a logistic regression prediction model.

Description

Technical Field

[0001] This invention relates to protein biomarkers for gastric cancer screening, and a logistic regression prediction model using the aforementioned protein biomarkers. Background Technology

[0002] Gastric carcinoma (GC) is a malignant tumor originating from the gastric mucosal epithelium. In terms of incidence, gastric cancer is the fifth most common cancer worldwide and the third leading cause of cancer death, ranking first in China. Because early-stage gastric cancer often lacks specific gastrointestinal manifestations, most patients are diagnosed at a middle or late stage with poor prognosis and limited treatment options, resulting in a very low five-year survival rate. However, existing biomarkers for gastric cancer diagnosis and prognosis have low sensitivity and specificity. Therefore, current diagnosis of gastric cancer primarily relies on barium meal X-rays, gastroscopy, abdominal ultrasound, spiral CT, and positron emission tomography (PET). These methods all have their limitations; for example, imaging techniques struggle to detect small tumors, leading to a high rate of missed diagnoses in early screening. Invasive surgery is inconvenient, and the acceptance rate of gastroscopy and colonoscopy is low among the vast majority of people. This is one of the important reasons for the high mortality rate of gastrointestinal tumors in my country.

[0003] Liquid biopsy is a novel detection technology that is gradually gaining widespread attention. Peripheral blood, saliva, urine, or gastric lavage fluid / gastric juice can serve as sources of specific biomarkers, providing important data for the screening and diagnosis of gastric cancer.

[0004] Biomarkers are a class of biochemical indicators used to mark changes or potential changes in the structure, tissues, organs, systems, or functions of cells and subcellular structures. They can be used to determine disease staging, diagnose diseases, or evaluate the safety and efficacy of new drugs or therapies. Protein biomarkers have unique advantages in accurately and sensitively screening for early, low-level damage, providing early warning of tumor development and offering clinicians a basis for auxiliary diagnosis.

[0005] However, there is still much room for improvement in the screening and research of protein biomarkers.

[0006] First, most current proteomics-based studies do not adequately consider the effectiveness of protein biomarkers in subsequent practical applications, which is one of the reasons why protein biomarkers fail in practical applications.

[0007] Secondly, most current proteomics studies do not distinguish between random and non-random missing protein quantification values, applying the same treatment method to both. This may lead to the omission of some protein biomarkers that are highly discriminative for cancer. For example, a protein may be expressed in most healthy individuals or patients with benign diseases, but not expressed in most cancer patients. This protein is excellent for differentiating cancer. If the mean or median is used to fill gaps for such proteins indiscriminately, they are very likely to be filtered out in subsequent protein biomarker screening stages, thus missing an important protein biomarker. Only a few studies have differentiated the sources of missing protein quantification values ​​and used different missing value filling methods, and the methods used in these studies are difficult to implement.

[0008] Furthermore, current research shows that data processing methods also significantly impact the selection of proteomics data. For example, the choice of data standardization method is crucial for subsequent analysis of proteomics data. Different literature or research sources recommend varying optimal standardization methods. Moreover, considering the two most prominent characteristics of proteomics data—"sparseness" and "high dimensionality"—which are also the two most challenging aspects of data processing. Summary of the Invention

[0009] Based on existing technology, the inventors conducted research on the screening of protein biomarkers and found that if the screening process is designed to filter proteins based on the proportion of proteins with quantitative values ​​in the tumor group or control group, and if a correlation coefficient-based method is used to identify non-random missing values ​​in the protein quantification matrix and different methods are used to fill in non-random and random missing values, and if multiple commonly used standardization methods are comprehensively adopted to standardize the proteomics data and used for screening protein biomarkers based on partial least squares regression analysis and machine learning, then due to the excellent performance of machine learning-based feature screening in addressing the "sparseness" and "high dimensionality" of proteomics data, using a random forest model based on genetic algorithms in machine learning for protein biomarker screening will achieve extremely excellent screening results.

[0010] This invention is based on the above-mentioned findings. Therefore, one aspect of this invention relates to serum protein biomarkers for gastric cancer screening, wherein the serum protein biomarkers are selected from P04040, O60814, A0A1S5UZ39, P02750, P00918, A0A3B3IQ51, B7ZKJ8, Q86UD1, A0A5C2GX62, P32119, P68871, V9H1D9, A2NH54, P59665, A0A 5C2G4M5, A0A5C2GQ40, P00915, P02042, P02748, P05155, P08727, Q16610, P01034, Q01973, A0A5C2GIT4, P 55058, Q00610, Q05C46, A0A5C2GX34, A0A5C2G5Z6, Q06033, Q14624, Q8N1N1, P00746, Q6P089, Q76LX8, Q9P2 D6, A0A5C2FYN5, P13473, Q14520, Q9H5P4, A0A075B6I0, A0A5C2FWF0, A0A5C2FXQ6, A0A5C2G1J3, A0A5C2G3 P7, A0A5C2G5F2, A0A5C2G6C5, A0A5C2G731, A0A5C2G9F4, A0A5C2GA02, A0A5C2GAN8, A0A5C2GKW6, A0A5C2G One or more of PK3, A0A5C2GRN7, A0A5C2GRQ1, A0A5C2GTE2, A0A5C2GTG7, A0A5C2GU59, A0A5C2GVJ2, A0A5C2GVL1, A0A5C2H1L6, A0N7I9, B2M1S7, B4DPR2, M0QY62, P00450, P07359, P0DJI8, Q15166, Q6N091, Q9P278 and V9HW95.

[0011] Another aspect of the present invention relates to a method for detecting serum protein biomarkers for gastric cancer screening as described above, characterized by comprising the following steps:

[0012] a) For each sample, the protein content of each serum protein biomarker was measured;

[0013] b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the following 13 proteins, namely Q86UD1, P08727, O60814, A0A3B3IQ51, A0A5C2GQ40, P04040, A0A5C2GX62, P00918, B7ZKJ8, A2NH54, A0A5C2GX34, Q76LX8 and A0A5C2GVJ2, will be filled with a constant of 1.0. The missing content of the remaining proteins will be filled with the K-nearest neighbor method.

[0014] c) The protein content data obtained in step (b) is standardized using one of the eight standardization methods provided by this invention, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the standardization method provided by this invention. When the P-value is ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value is < 0.5, the sample is identified as a benign gastric disease sample. Attached Figure Description

[0015] Figure 1 This is a flowchart for determining whether a sample originates from a gastric cancer patient.

[0016] Figure 2 This is a flowchart for protein biomarker screening.

[0017] Figure 3 This is a graph showing the results of principal component analysis.

[0018] Figure 4 It is a clustering heatmap of differentially expressed proteins based on t-test.

[0019] Figure 5 This is the cross-validation ROC curve of the logistic regression model for four serum protein biomarkers, P04040, A0A1S5UZ39, P02750, and P00918, under eight standardized methods.

[0020] Figure 6 This is the cross-validation ROC curve of the logistic regression model for four serum protein biomarkers, A0A3B3IQ51, P02750, O60814, and P00918, under eight standardized methods.

[0021] Figure 7 This is the cross-validation ROC curve of the logistic regression model for three serum protein biomarkers, A0A3B3IQ51, P02750, and P00918, under eight standardized methods.

[0022] Figure 8This is the cross-validation ROC curve of the logistic regression model for three serum protein biomarkers, A0A1S5UZ39, P02750, and P00918, under eight standardized methods.

[0023] Figure 9 This is the cross-validation ROC curve of the logistic regression model for two serum protein biomarkers, A0A3B3IQ51 and P02750, under eight standardized methods.

[0024] Figure 10 This is the cross-validation ROC curve of the logistic regression model for two serum protein biomarkers, A0A1S5UZ39 and P02750, under eight standardized methods.

[0025] Figure 11 This is the ROC curve of the logistic regression model for four serum protein biomarkers, P04040, A0A1S5UZ39, P02750, and P00918, on the test set under eight standardized methods.

[0026] Figure 12 This is the ROC curve of the logistic regression model for four serum protein biomarkers, A0A3B3IQ51, P02750, O60814, and P00918, on the test set under eight standardized methods.

[0027] Figure 13 This is the ROC curve of the logistic regression model for three serum protein biomarkers, A0A3B3IQ51, P02750, and P00918, on the test set under eight standardized methods.

[0028] Figure 14 This is the ROC curve of the logistic regression model for three serum protein biomarkers, A0A1S5UZ39, P02750, and P00918, on the test set under eight standardized methods.

[0029] Figure 15 This is the ROC curve of the logistic regression model for two serum protein biomarkers, A0A3B3IQ51 and P02750, on the test set under eight standardized methods.

[0030] Figure 16 This is the ROC curve of the logistic regression model for two serum protein biomarkers, A0A1S5UZ39 and P02750, on the test set under eight standardized methods. Detailed Implementation

[0031] The specific embodiments of the present invention will be described below.

[0032] As used herein, the term "cancer" refers to the presence of cells that exhibit typical characteristics of cancer cells, such as uncontrolled proliferation, immortality, metastatic potential, rapid growth and proliferation rates, and certain characteristic morphological features known in the art.

[0033] In one instance, "cancer" can be stomach cancer or gastric cancer. In one embodiment, "cancer" can include both pre-malignant and malignant cancer. Therefore, the term "gastric cancer" encompasses all stages of gastric cancer as described in the 2020 CSCO Guidelines for the Diagnosis and Treatment of Gastric Cancer.

[0034] In one instance, as those skilled in the art will understand, the methods described herein do not involve steps performed by a physician / doctor. Therefore, the results obtained from the methods described herein need to be considered in conjunction with clinical data and other clinical presentations before a final diagnosis by a physician can be provided to the subject. A final diagnosis regarding whether a subject has gastric cancer is within the scope of the physician and is not considered part of this disclosure.

[0035] Therefore, as used herein, the terms “identify,” “detect,” and “diagnose” refer to determining the probability or likelihood that a subject has a disease at any stage of development (such as gastric cancer) or to determining a subject’s susceptibility to developing said disease. In one instance, “diagnose,” “identify,” or “detect” is performed before symptoms appear. In another instance, “diagnose,” “identify,” or “detect” allows a clinician / physician (in conjunction with other clinical presentations) to confirm gastric cancer in a subject suspected of having it.

[0036] As used herein, the term "sample" refers to a sample collected from a subject for the purpose of detecting the types and amounts of proteins contained therein. Subject samples may be from the circulatory system (i.e., from blood) or not from the circulatory system (i.e., not from blood). Subject samples can be any sample containing proteins suitable for detection, and their sources include whole blood, bone marrow, pleural fluid, peritoneal fluid, central cerebrospinal fluid, milk, urine, tears, sweat, saliva, organ secretions, and lavage fluid from the bronchi, nasal cavity, pharynx, etc.

[0037] In one instance, the subject sample is blood, including, for example, whole blood or any part or component thereof. Blood samples suitable for use in this invention can be extracted from any known source including blood cells or components thereof, such as veins, arteries, peripheral tissues, tissues, spinal cord, and the like. For example, the obtained sample can be obtained and processed using known and conventional clinical methods (e.g., procedures for drawing and processing whole blood).

[0038] In one instance, the subject sample is serum. Methods for obtaining serum from blood are well known to those skilled in the art.

[0039] This invention discovers that by monitoring whether a sample contains a set of gastric cancer biomarkers, gastric cancer can be diagnosed with high specificity and sensitivity. Especially for early-stage gastric cancer, which was previously difficult to diagnose, the biomarkers of this invention exhibit extremely high specificity and sensitivity.

[0040] As used herein, a “biomarker” or “marker” is a biological molecule that is objectively measured to serve as a characteristic indicator of the physiological state of a biological system. For the purposes of this disclosure, biological molecules include ions, small molecules, peptides, peptide chains, proteins, and peptides and proteins with post-translational modifications, nucleosides, nucleotides, and polynucleotides including RNA and DNA, glycoproteins, lipoproteins, and various covalent or non-covalent modifications of these types of molecules. Biological molecules include any kind of entities that are native to, characteristic of, and / or essential to the function of a biological system. Most biomarkers are polypeptides, although they may also be pre-translational mRNA or modified mRNA representing a gene product expressed as a polypeptide, or may include post-translational modifications of said polypeptide.

[0041] As used herein, "protein biomarker" means the biomarker that contains protein information. In one instance, it refers to the biomarker that contains a protein sequence. Further, in one instance, it refers to a full-length protein, a single peptide chain, a characteristic peptide segment of a peptide chain, and a stable isotopic protein or a stable isotopic characteristic peptide segment thereof.

[0042] This invention utilizes a protein expression matrix based on clinical samples. Through a series of steps including abnormal feature processing, feature filtering based on the effectiveness of subsequent practical applications, identification of non-random and random missing values, missing value imputation, identification and processing of abnormal samples, data standardization, screening of protein biomarkers, training of logistic regression models, evaluation of the effectiveness of logistic regression models, and evaluation of the predictive performance of logistic regression models, a set of serum protein biomarkers for gastric cancer screening was identified. Further screening yielded six core serum protein biomarkers. Based on these six protein biomarkers, 48 ​​logistic regression prediction models were constructed that can effectively distinguish gastric cancer samples from benign gastric disease samples and possess predictive capabilities.

[0043] The following lists some terms used in the embodiments of the present invention. Within the scope of the specification and claims of this invention, the relevant terms are defined as follows. Other terms not listed are defined using common definitions in the art, and their meanings are well known to those skilled in the art.

[0044] Non-random missing data: This refers to the situation where, due to problems with the sample itself, some protein quantification data cannot be detected, i.e., non-random missing protein quantification data occurs.

[0045] Random missing data: This refers to the situation where, due to random perturbations in the instrument during the actual protein quantification process, some samples cannot be detected in the protein quantification data, i.e., random missing protein quantification data occurs.

[0046] Principal component analysis (PCA) is a commonly used preprocessing method for linearly reducing the dimensionality of data. Its goal is to use variance to measure the variability of the data and project high-dimensional data with significant variability into a low-dimensional space for representation, thus making it applicable to anomaly detection.

[0047] K-nearest neighbor imputation is a well-known method for imputing missing values. It uses the combined information of multiple nearest neighbors of a sample with missing values ​​to impute the missing values.

[0048] Log2 normalization, also known as Log2 transformation, is a logarithmic transformation of the expression values ​​in the expression matrix, with the base 2.

[0049] Log2+Median Standardization: For the expression values ​​in the expression matrix, first perform Log2 transformation. For each protein expression value in each sample in the transformed matrix, divide the expression value by the median value of all protein expression values ​​in that sample, and then multiply by the median value of all protein expression values ​​in all samples.

[0050] Log2+CycLoess standardization: For the expression values ​​in the expression matrix, first perform Log2 transformation, and then use local weighted regression to standardize the transformed expression matrix.

[0051] Log2+Mean standardization: For the expression values ​​in the expression matrix, first perform Log2 transformation. For each protein expression value in each sample in the transformed matrix, divide it by the mean of all protein expression values ​​in that sample, and then multiply it by the mean of all protein expression values ​​in all samples.

[0052] VSN Standardization: Since VSN standardization is similar to log2 transformation, there is no need to perform log2 transformation first. The expression values ​​in the expression matrix can be directly standardized by variance stabilization.

[0053] Log2+RLR standardization: For the expression values ​​in the expression matrix, first perform Log2 transformation, and then use robust linear regression to standardize the transformed expression matrix.

[0054] Log2+GI standardization: For the expression values ​​in the expression matrix, first perform Log2 transformation, then divide the expression value of each protein in each sample in the transformed matrix by the sum of the expression values ​​of all proteins in that sample, and then multiply by the median value of the sum of the expression values ​​of all proteins in all samples.

[0055] Log2+Quantile normalization: For the expression values ​​in the expression matrix, first perform Log2 transformation, then sort each column separately, calculate the average of the sorted matrix to obtain the average vector, and then replace the corresponding average according to the original matrix sorting.

[0056] PLS: Partial Least Squares Regression Analysis, which combines principal component analysis, canonical correlation, and multiple linear regression into one, is a mapping dimensionality reduction method, and therefore can be used for feature screening of small samples.

[0057] GA-RF: A random forest model based on genetic algorithms, which uses a genetic algorithm (GA) to optimize and tune the parameters of a random forest (RF) model.

[0058] True Yang: Correctly predicts the number of positive samples, which is actually a positive sample, and the prediction is also a positive sample.

[0059] True negative: The number of negative samples is correctly predicted, and the actual number of negative samples is also predicted to be negative.

[0060] False positive: The number of positive samples is incorrectly predicted when they are actually negative samples.

[0061] False negative: The number of negative samples is incorrectly predicted when they are actually positive samples.

[0062] Sensitivity: also known as recall or true positive rate, is the number of correctly predicted positive samples / the total number of actual positive samples, or (true positive) / (true positive + false negative).

[0063] Specificity: also known as the true negative rate, which is the number of correctly predicted negative samples / the total number of actual negative samples, or (true negative) / (true negative + false positive).

[0064] Accuracy: also known as positive predictive value, is the number of correctly predicted positive samples / the total number of predicted positive samples, or (true positives) / (true positives + false positives).

[0065] Accuracy: The number of correctly predicted positive and negative samples / the total number of samples, that is, (true positive + true negative) / (true positive + true negative + false positive + false negative).

[0066] F1 score: F1 = 2 * (precision * recall) / (precision + recall).

[0067] False positive rate: the number of incorrectly predicted positive samples / the total number of actual negative samples, equal to (1 - specificity), or (false positive) / (true negative + false positive).

[0068] ROC curve and AUC value: Both are standards used to measure the performance of a classifier. The ROC (Receiver Operating Characteristic) curve has the false positive rate on the horizontal axis and the true positive rate on the vertical axis. The plotted curve should lie above the y=x line. The area under the ROC curve is the AUC value. The larger the AUC, the better the classification performance of the classifier (such as the logistic regression model).

[0069] In one aspect, the present invention provides a set of serum protein biomarkers for gastric cancer screening, which are particularly suitable for screening early gastric cancer, and provides a logistic regression prediction model based on the above-mentioned serum protein biomarkers, which can be used to identify whether a sample originates from a gastric cancer patient.

[0070] The serum protein biomarkers used for gastric cancer screening are P04040, O60814, A0A1S5UZ39, P02750, P00918, A0A3B3IQ51, B7ZKJ8, Q86UD1, A0A5C2GX62, P32119, P68871, V9H1D9, A2NH54, P59665, A0A5C2G4M5, A0A5C2GQ40, P00915, and P02. 042, P02748, P05155, P08727, Q16610, P01034, Q01973, A0A5C2GIT4, P55058, Q00610, Q05C46, A0A 5C2GX34, A0A5C2G5Z6, Q06033, Q14624, Q8N1N1, P00746, Q6P089, Q76LX8, Q9P2D6, A0A5C2FYN5, P13 473. Q14520, Q9H5P4, A0A075B6I0, A0A5C2FWF0, A0A5C2FXQ6, A0A5C2G1J3, A0A5C2G3P7, A0A5C2G5 F2, A0A5C2G6C5, A0A5C2G731, A0A5C2G9F4, A0A5C2GA02, A0A5C2GAN8, A0A5C2GKW6, A0A5C2GPK3, A0 A5C2GRN7, A0A5C2GRQ1, A0A5C2GTE2, A0A5C2GTG7, A0A5C2GU59, A0A5C2GVJ2, A0A5C2GVL1, A0A5C2H1L6, A0N7I9, B2M1S7, B4DPR2, M0QY62, P00450, P07359, P0DJI8, Q15166, Q6N091, Q9P278 and V9HW95.

[0071] Preferably, the serum protein biomarkers for gastric cancer screening include at least one or more serum protein biomarkers selected from P04040, O60814, A0A1S5UZ39, P02750, P00918 and A0A3B3IQ51.

[0072] In one aspect, the present invention also provides a serum protein biomarker-based detection method for gastric cancer screening, which combines serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 for gastric cancer screening, comprising the following steps:

[0073] (1) For the samples, the protein content of serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 were measured respectively.

[0074] (2) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 and P04040 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method.

[0075] (3) The protein content data obtained in step (2) is standardized using one of the eight standardization methods provided by this invention, and the predicted probability P value is calculated using the probability prediction formula corresponding to the standardization method provided by this invention. When the P value ≥ 0.5, it is determined to be a gastric cancer sample; when the P value < 0.5, it is determined to be a benign gastric disease sample. The following is the probability prediction formula for this group of protein biomarkers, where the name of the protein biomarker in the formula refers to the standardized content value of the protein biomarker:

[0076] a) For Log2+CycLoess standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-56.171-0.692*P04040+0.228*A0A1S5UZ39+5.009*P02750-2.540*P00918)));

[0077] b) For Log2+GI standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-77.717-0.478*P04040-0.341*A0A1S5UZ39+5.392*P02750-1.027*P00918)));

[0078] c) For Log2 standardization, the probability prediction formula for this group of markers is P = 1 / (1 + Exp(-(-33.799-0.473*P04040-0.440*A0A1S5UZ39+4.019*P02750-1.976*P00918)));

[0079] d) For Log2+Mean standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-77.536-0.478*P04040-0.337*A0A1S5UZ39+5.378*P02750-1.035*P00918)));

[0080] e) For Log2+Median standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-55.721-0.441*P04040-0.473*A0A1S5UZ39+4.591*P02750-1.270*P00918)));

[0081] f) For Log2+Quantile standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-67.461-0.647*P04040+0.550*A0A1S5UZ39+5.213*P02750-2.583*P00918)));

[0082] g) For Log2+RLR standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-59.715-0.680*P04040+0.382*A0A1S5UZ39+4.723*P02750-2.191*P00918)));

[0083] h) For VSN standardization, the probability prediction formula for this set of markers is P = 1 / (1 + Exp(-(-44.111-0.269*P04040-1.489*A0A1S5UZ39+5.961*P02750-2.395*P00918))).

[0084] A method for detecting serum protein biomarkers, using a combination of four serum protein biomarkers—A0A3B3IQ51, P02750, O60814, and P00918—for gastric cancer screening, includes the following steps:

[0085] (1) For the sample, the protein content of four serum protein biomarkers, namely A0A3B3IQ51, P02750, O60814 and P00918, was determined respectively.

[0086] (2) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the three proteins A0A3B3IQ51, P00918 and O60814 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method.

[0087] (3) The protein content data obtained in step (2) is standardized using one of the eight standardization methods provided by this invention, and the predicted probability P value is calculated using the probability prediction formula corresponding to the standardization method provided by this invention. When the P value ≥ 0.5, it is determined to be a gastric cancer sample; when the P value < 0.5, it is determined to be a benign gastric disease sample. The following is the probability prediction formula for this group of protein biomarkers, where the name of the protein biomarker in the formula refers to the standardized content value of the protein biomarker:

[0088] a) For Log2+CycLoess standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-329.023+3.922*A0A3B3IQ51+11.630*P02750+4.905*O60814-2.136*P00918)));

[0089] b) For Log2+GI standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-229.243+2.236*A0A3B3IQ51+10.504*P02750+2.630*O60814-3.348*P00918)));

[0090] c) For Log2 standardization, the probability prediction formula for this group of markers is P = 1 / (1 + Exp(-(-262.585 + 3.917 * A0A3B3IQ51 + 7.231 * P02750 + 4.551 * O60814 - 0.432 * P00918)));

[0091] d) For Log2+Mean standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-224.074+2.196*A0A3B3IQ51+10.256*P02750+2.566*O60814-3.293*P00918)));

[0092] e) For Log2+Median standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-118.175+1.487*A0A3B3IQ51+5.076*P02750+1.751*O60814-2.002*P00918)));

[0093] f) For Log2+Quantile standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-108.393+0.910*A0A3B3IQ51+5.493*P02750+1.464*O60814-2.315*P00918)));

[0094] g) For Log2+RLR standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-192.053+2.244*A0A3B3IQ51+8.082*P02750+2.851*O60814-2.860*P00918)));

[0095] h) For VSN standardization, the probability prediction formula for this set of markers is P = 1 / (1 + Exp(-(-195.306 + 3.888 * A0A3B3IQ51 + 6.315 * P02750 + 3.570 * O60814 - 2.862 * P00918))).

[0096] The detection method for serum protein biomarkers, using a combination of three serum protein biomarkers—A0A3B3IQ51, P02750, and P00918—for gastric cancer screening includes the following steps:

[0097] (1) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750 and P00918 was measured respectively;

[0098] (2) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 and P00918 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method.

[0099] (3) The protein content data obtained in step (2) is standardized using one of the eight standardization methods provided by this invention, and the predicted probability P value is calculated using the probability prediction formula corresponding to the standardization method provided by this invention. When the P value ≥ 0.5, it is determined to be a gastric cancer sample; when the P value < 0.5, it is determined to be a benign gastric disease sample. The following is the probability prediction formula for this group of protein biomarkers, where the name of the protein biomarker in the formula refers to the standardized content value of the protein biomarker:

[0100] a) For Log2+CycLoess standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-85.208+0.976*A0A3B3IQ51+4.524*P02750-1.236*P00918)));

[0101] b) For Log2+GI standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-109.138+1.314*A0A3B3IQ51+4.946*P02750-0.566*P00918)));

[0102] c) For Log2 standardization, the probability prediction formula for this group of markers is P = 1 / (1 + Exp(-(-75.531 + 1.118 * A0A3B3IQ51 + 3.545 * P02750 - 0.749 * P00918)));

[0103] d) For Log2+Mean standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-108.872+1.309*A0A3B3IQ51+4.927*P02750-0.569*P00918)));

[0104] e) For Log2+Median standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-84.630+1.328*A0A3B3IQ51+3.794*P02750-0.696*P00918)));

[0105] f) For Log2+Quantile standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-82.513+0.526*A0A3B3IQ51+5.121*P02750-1.750*P00918)));

[0106] g) For Log2+RLR standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-87.794+1.146*A0A3B3IQ51+4.335*P02750-1.002*P00918)));

[0107] h) For VSN standardization, the probability prediction formula for this set of markers is P = 1 / (1 + Exp(-(-86.244 + 1.988 * A0A3B3IQ51 + 4.499 * P02750 - 2.146 * P00918))).

[0108] A method for detecting serum protein biomarkers, using a combination of three serum protein biomarkers—A0A1S5UZ39, P02750, and P00918—for gastric cancer screening, includes the following steps:

[0109] (1) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39, P02750 and P00918 was measured respectively.

[0110] (2) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method.

[0111] (3) The protein content data obtained in step (1) is standardized using one of the eight standardization methods provided by this invention, and the predicted probability P value is calculated using the probability prediction formula corresponding to the standardization method provided by this invention. When the P value ≥ 0.5, it is determined to be a gastric cancer sample; when the P value < 0.5, it is determined to be a benign gastric disease sample. The following is the probability prediction formula for this group of protein biomarkers, where the name of the protein biomarker in the formula refers to the standardized content value of the protein biomarker:

[0112] a) For Log2+CycLoess standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-63.622-1.625*A0A1S5UZ39+6.022*P02750-1.335*P00918)));

[0113] b) For Log2+GI standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-83.752-1.821*A0A1S5UZ39+6.673*P02750-0.584*P00918)));

[0114] c) For Log2 standardization, the probability prediction formula for this group of markers is P = 1 / (1 + Exp(-(-34.719-1.630*A0A1S5UZ39+4.617*P02750-1.397*P00918)));

[0115] d) For Log2+Mean standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-83.501-1.818*A0A1S5UZ39+6.652*P02750-0.588*P00918)));

[0116] e) For Log2+Median standardization, the probability prediction formula for this set of markers is P=1 / (1+Exp(-(-66.466-1.714*A0A1S5UZ39+5.801*P02750-0.741*P00918)));

[0117] f) For Log2+Quantile standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-67.479-1.352*A0A1S5UZ39+6.143*P02750-1.629*P00918)));

[0118] g) For Log2+RLR standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-66.818-1.468*A0A1S5UZ39+5.875*P02750-1.157*P00918)));

[0119] h) For VSN standardization, the probability prediction formula for this set of markers is P = 1 / (1 + Exp(-(-54.044-1.640*A0A1S5UZ39+6.054*P02750-1.930*P00918))).

[0120] In one example, the present invention also provides a serum protein biomarker-based detection method for gastric cancer screening that combines serum protein biomarkers A0A3B3IQ51 and P02750, comprising the following steps:

[0121] (1) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51 and P02750 was measured respectively;

[0122] (2) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method.

[0123] (3) The protein content data obtained in step (2) is standardized using one of the eight standardization methods provided by this invention, and the predicted probability P value is calculated using the probability prediction formula corresponding to the standardization method provided by this invention. When the P value ≥ 0.5, it is determined to be a gastric cancer sample; when the P value < 0.5, it is determined to be a benign gastric disease sample. The following is the probability prediction formula for this group of protein biomarkers, where the name of the protein biomarker in the formula refers to the standardized content value of the protein biomarker:

[0124] a) For Log2+CycLoess standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-88.204+1.621*A0A3B3IQ51+3.263*P02750)));

[0125] b) For Log2+GI standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-100.532+1.676*A0A3B3IQ51+3.841*P02750)));

[0126] c) For Log2 standardization, the probability prediction formula for this group of markers is P = 1 / (1 + Exp(-(-78.997 + 1.513 * A0A3B3IQ51 + 2.872 * P02750)));

[0127] d) For Log2+Mean standardization, the probability prediction formula for this set of markers is P=1 / (1+Exp(-(-100.271+1.672*A0A3B3IQ51+3.820*P02750)));

[0128] e) For Log2+Median standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-79.809+1.648*A0A3B3IQ51+2.808*P02750)));

[0129] f) For Log2+Quantile standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-84.715+1.341*A0A3B3IQ51+3.293*P02750)));

[0130] g) For Log2+RLR standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-88.211+1.571*A0A3B3IQ51+3.296*P02750)));

[0131] h) For VSN standardization, the probability prediction formula for this set of markers is P = 1 / (1 + Exp(-(-103.612 + 2.559 * A0A3B3IQ51 + 3.265 * P02750))).

[0132] In one example, the present invention also provides a serum protein biomarker-based detection method for gastric cancer screening that combines serum protein biomarkers A0A1S5UZ39 and P02750, comprising the following steps:

[0133] (1) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39 and P02750 was measured respectively;

[0134] (2) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the K-nearest neighbor method is used to imput the missing content of the two proteins A0A1S5UZ39 and P02750.

[0135] (3) The protein content data obtained in step (2) is standardized using one of the eight standardization methods provided by this invention, and the predicted probability P value is calculated using the probability prediction formula corresponding to the standardization method provided by this invention. When the P value ≥ 0.5, it is determined to be a gastric cancer sample; when the P value < 0.5, it is determined to be a benign gastric disease sample. The following is the probability prediction formula for this group of protein biomarkers, where the name of the protein biomarker in the formula refers to the standardized content value of the protein biomarker:

[0136] a) For Log2+CycLoess standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-53.231-2.793*A0A1S5UZ39+5.764*P02750)));

[0137] b) For Log2+GI standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-75.499-3.016*A0A1S5UZ39+7.149*P02750)));

[0138] c) For Log2 standardization, the probability prediction formula for this group of markers is P = 1 / (1 + Exp(-(-36.635-2.676*A0A1S5UZ39+4.801*P02750)));

[0139] d) For Log2+Mean standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-75.257-3.011*A0A1S5UZ39+7.124*P02750)));

[0140] e) For Log2+Median standardization, the probability prediction formula for this set of markers is P=1 / (1+Exp(-(-55.018-2.803*A0A1S5UZ39+5.875*P02750)));

[0141] f) For Log2+Quantile standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-56.416-2.719*A0A1S5UZ39+5.852*P02750)));

[0142] g) For Log2+RLR standardization, the probability prediction formula for this group of markers is P=1 / (1+Exp(-(-61.631-2.649*A0A1S5UZ39+6.042*P02750)));

[0143] h) For VSN standardization, the probability prediction formula for this set of markers is P = 1 / (1 + Exp(-(-60.780-2.778*A0A1S5UZ39+6.136*P02750))).

[0144] In one aspect, the present invention provides a method for determining whether a sample is derived from a patient with gastric cancer.

[0145] For ease of description, a schematic diagram of the method for determining whether a sample originates from a gastric cancer patient proposed in this invention is shown below. Figure 1 .like Figure 1 As shown, in one example, the method includes:

[0146] (1) Determination of serum protein biomarker content: The relevant protein content of the sample is determined in order to obtain information on the expression level of relevant protein biomarkers in the sample;

[0147] (2) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the 13 proteins Q86UD1, P08727, O60814, A0A3B3IQ51, A0A5C2GQ40, P04040, A0A5C2GX62, P00918, B7ZKJ8, A2NH54, A0A5C2GX34, Q76LX8 and A0A5C2GVJ2 will be filled with a constant of 1.0, and the missing protein content of other proteins will be filled with the K nearest neighbor method.

[0148] (3) Data standardization: Protein expression data are standardized using one of eight methods (Log2, Log2+Median, Log2+Mean, VSN, Log2+RLR, Log2+GI, Log2+Quantile and Log2+CycLoess);

[0149] (4) Calculation of the probability value of the sample originating from a gastric cancer patient: Based on the standardized data obtained in (3) and the probability prediction formula of the corresponding standardization method mentioned in this paper, the probability value of the sample to be tested originating from a gastric cancer patient is determined. According to the method of the present invention, without relying on the patient's sample information, it is possible to determine whether the sample to be tested originates from a gastric cancer patient based solely on the content of relevant protein biomarkers in the sample to be tested, and solely through the prediction model provided by the present invention, with high specificity and high sensitivity, thereby indicating the result of gastric cancer screening.

[0150] The present invention will be further described below with reference to the embodiments.

[0151] Example

[0152] The embodiments described in this invention should not be construed as limiting the invention. All other embodiments obtained by those skilled in the art or by analysis without inventive effort are within the scope of protection of this invention. The reference figures are merely illustrative and intended to explain the invention, and should not be construed as limiting the invention.

[0153] Example 1: Screening of protein biomarkers

[0154] For ease of description, a schematic diagram of the protein biomarker screening process described in this invention is shown below. Figure 2 :

[0155] (I) Mass spectrometry analysis and UniProt database search:

[0156] Mass spectrometry analysis and UniProt database searches were performed on samples from known sources to obtain protein expression matrices for these samples. The known sources consisted of a known number of benign gastric disease samples and a known number of gastric cancer samples. The specific steps are as follows:

[0157] Sample source

[0158] Forty serum samples from non-gastric cancer patients and forty serum samples from stage I-IV gastric cancer patients were obtained from Shanghai Zhongshan Hospital and used as samples for this invention. The collection of these samples followed the ethical standards established by the Ethics Committee of Shanghai Zhongshan Hospital, and informed consent forms were signed.

[0159] Sample processing

[0160] The above 80 samples were used as high-abundance, high-abundance mixed samples, and original samples, respectively. Appropriate amounts of protein from each sample were subjected to SDS-PAGE electrophoresis to assess the consistency between samples. The purpose was to exclude potentially contaminated samples. After evaluation, all samples were considered uncontaminated.

[0161] Protein reduction and alkylation, enzyme digestion: Add 35 μL UA buffer (8M Urea, 150mM Tris-HCl, pH 8.0) and mix well. Add DTT to a final concentration of 20 mM, react at 37℃ for 2 h, then return to room temperature. Add IAA to a final concentration of 25 mM (50 mM IAA in UA), shake at 600 rpm for 1 min, and incubate at room temperature in the dark for 30 min. Add 150 μL NH4HCO3 buffer (50 mM), then add 2 μg Lys-C to the sample and react for 4 h. Finally, add 4 μg Trypsin and incubate at 37℃ for 16 h.

[0162] The peptides were desalted using a C18 column, and their concentration was determined at OD280. Then, 2 μg of peptide was extracted from each sample, incorporated with an appropriate amount of iRT standard peptide, and analyzed by LC-MS / MS DDA and LC-MS / MS DIA methods.

[0163] Mass spectrometry analysis

[0164] The proteins contained in the sample were analyzed using nano-liquid chromatography-Q-Exactive HF mass spectrometry.

[0165] First, chromatographic separation was performed using an Easy nLC-1200 nano-flow HPLC system. Buffer solutions: Solution A was a 0.1% formic acid aqueous solution, and Solution B was a 0.1% formic acid-acetonitrile aqueous solution. The column was equilibrated with 95% of Solution A. After the sample was injected into the trap column, it underwent gradient separation through a 50 cm tip column at a flow rate of 250 nL / min. The HPLC separation gradient was as follows: 0 min–80 min, linear gradient of Solution B from 8% to 30%; 70 min–100 min, linear gradient of Solution B from 30% to 100%; 80 min–120 min, linear gradient of Solution B increased to 100% and remained thereafter.

[0166] Then, the chromatographically separated samples were analyzed by DDA scanning using a Q-Exactive HF mass spectrometer (Thermo Scientific). Ion mode: ESI positive ion. Primary mass spectrometry scan range: 300-1800 m / z, mass resolution: 60,000 (@m / z 200), AGC target: 3e6, Maximum IT: 50 ms. After each primary MS scan (full MS scan), 20 ddMS2 scans were acquired according to the inclusion list. Isolation window: 1.6Th, mass resolution: 30,000 (@m / z 200), AGC target: 3e6, Maximum IT: 120 ms, MS2 Activation Type: HCD, Normalized collision energy: 27.

[0167] The chromatographically separated samples were analyzed by DIA mass spectrometry. Ionization mode: positive ion. Primary mass spectrometry scan range: 350-1650 m / z, mass resolution: 120,000 (@m / z 200), AGC target: 3e6, Maximum IT: 50 ms. MS2 used DIA data acquisition mode, with 30 DIA acquisition windows set, mass resolution: 30,000 (@m / z 200), AGC target: 3e6, Maximum IT: auto, MS2 Activation Type: HCD, Normalized collision energy: 30, Spectral data type: profile.

[0168] UniProt database search

[0169] The obtained mass spectrometry data were used to search the UniProt database using Maxquant software (Maxquant_1.5.3.17), with other parameters set to the software defaults, to identify proteins and obtain a protein expression matrix with 80 samples and 1806 proteins.

[0170] (II) Protein filtering based on application effectiveness: Due to the limitations of the technology itself, the quantitative values ​​of protein biomarkers may be randomly missing. In order to improve the effectiveness of the application, proteins that have quantitative values ​​in more than 90% of benign gastric disease samples or more than 90% of gastric cancer samples are selected for subsequent protein biomarker screening. The number of proteins that meet the requirements is 1123.

[0171] (III) Identifying and processing anomalous samples: such as Figure 3 As shown, principal component analysis was used to identify an anomalous sample (Case 12), and this anomalous sample was removed from the population, resulting in a protein intensity information matrix with 79 samples and 1123 proteins.

[0172] (iv) Identification of missing values: Identify whether the missing values ​​in the protein expression matrix are due to random or non-random deletion. This step specifically includes: (1) Constructing a missing value matrix: Fill the positions in the protein expression matrix containing qualitative and quantitative information of proteins with 1, and fill the positions where the qualitative and quantitative information of proteins is missing with 0; (2) Constructing phenotypic data sequences: Define all benign disease samples as 0, and all gastric cancer samples as 1; (3) Calculate the Pearson correlation coefficient between the data sequence formed by the value of each protein in the missing value matrix in all samples and the phenotypic data sequence formed by all samples. Proteins with a correlation coefficient >= 0.3 are considered to have missing quantitative values. Whether a protein's quantitative value is missing depends on the sample's phenotype. Thirteen proteins meet this criterion: Q86UD1, P08727, O60814, A0A3B3IQ51, A0A5C2GQ40, P04040, A0A5C2GX62, P00918, B7ZKJ8, A2NH54, A0A5C2GX34, Q76LX8, and A0A5C2GVJ2. These proteins form set ①. The absence of these 13 proteins is considered a non-random missing quantitative value. (4) The remaining 1110 proteins, excluding those with non-random missing quantitative values ​​from step (3), are considered randomly missing quantitative values. For these proteins, there is no distinction between the control group and the tumor group; any sample with a missing quantitative signal intensity is considered to have a randomly missing quantitative value.

[0173] (v) Imputation of missing values: Different strategies are used to impute non-randomly missing protein quantification values ​​and randomly missing protein quantification values. For non-randomly missing protein quantification values, a constant of 1.0 is used for imputation, while for randomly missing protein quantification values, the K-nearest neighbor method is used for imputation.

[0174] The imputation of missing values ​​is detailed below:

[0175] (1) Imputation strategy for proteins with non-randomly missing quantitative values: As shown in Table 1, the quantitative value missing values ​​of the five proteins Q86UD1, P08727, O60814, A0A3B3IQ51, and A0A5C2GQ40 mainly occurred in the control group. Therefore, the constant 1.0 was used to impute the quantitative value missing values ​​of these five proteins in the control group samples, while the K-nearest neighbor method was used to calculate and impute the quantitative value missing values ​​of these proteins in the tumor group samples. Conversely, the quantitative value missing values ​​of the eight proteins P00918, B7ZKJ8, A2NH54, P04040, A0A5C2GX62, A0A5C2GX34, Q76LX8, and A0A5C2GVJ2 mainly occurred in the tumor group. Therefore, the constant 1.0 was used to impute the quantitative value missing values ​​of these eight proteins in the tumor group samples, while the K-nearest neighbor method was used to calculate and impute the quantitative value missing values ​​of these proteins in the control group samples.

[0176] (2) Strategy for filling the quantitative value of proteins with random missing quantitative values: For these proteins, regardless of whether they are in the control group or the tumor group, the K-nearest neighbor method is used to calculate and fill the missing quantitative signal intensity in all samples.

[0177] Table 1. Quantitative value loss of non-randomly deleted proteins in the control and tumor groups.

[0178] ProteinID The proportion of samples with missing quantitative values ​​in the control group Proportion of samples with missing quantitative values ​​in the tumor group Q86UD1 37.50％ 5.00％ P08727 37.50％ 10.00％ O60814 35.00％ 7.50％ A0A3B3IQ51 35.00％ 7.50％ A0A5C2GQ40 27.50％ 2.50％ P00918 7.50％ 47.50％ B7ZKJ8 7.50％ 37.50％ A2NH54 7.50％ 35.00％ P04040 10.00％ 35.00％ A0A5C2GX62 10.00％ 35.00％ A0A5C2GX34 5.00％ 30.00％ Q76LX8 0.00％ 22.50％ A0A5C2GVJ2 0.00％ 20.00％

[0179] (vi) Differential expression protein analysis: The expression levels of the protein biomarkers to be screened in the samples of benign gastric diseases after the above five steps were compared with the expression levels of the protein biomarkers to be screened in the samples of gastric cancer based on t-test (the results are shown in Table 2), thus obtaining the protein biomarker set ②. This collection contains a total of 54 proteins, namely Q86UD1, V9H1D9, A0A1S5UZ39, P55058, Q9P2D6, A0A5C2GTE2, P02042, P05155, P68871, A0A5C2GIT4, P00918, A0A5C2G4M5, A0A5C2GRN7, B7ZKJ8, A0A5C2G731, V9HW95, P02750, P00915, A0A5C2GKW6, A2NH54, P59665, A0A5C2H1L6, A0A075B6I0, A0A5C2G9F4, O60814, Q8N1N1, Q6P089, A 0A5C2G5F2, A0N7I9, A0A5C2FWF0, A0A5C2GTG7, M0QY62, Q16610, A0A5C2FXQ6, Q6N091, P00746, P32119, Q15166, P0DJI8, A0A5C2GU59, A0A5C2GVL1, B2M1S7, A0A5C2GPK3, A0A5C2G3P7, B4DPR2, A0A5C2GRQ1, A0A5C2GX62, Q9P278, A0A5C2G6C5, A0A5C2GA02, A0A5C2G1J3, P04040, A0A5C2GAN8 and Q76LX8. Hierarchical clustering analysis was performed on all samples using these differentially expressed proteins, and the results are as follows: Figure 4 The clustering heatmap based on differentially expressed proteins using t-tests shows that these differentially expressed proteins can correctly cluster most gastric cancer samples and benign gastric disease samples.

[0180] Table 2. Results of differentially expressed proteins based on t-test

[0181] Protein name ProteinID abs(logFC) P-value Out at first protein homolog Q86UD1 1.08 6.94E-09 Alpha globin V9H1D9 1.31 2.21E-07 Hemoglobin subunit alpha A0A1S5UZ39 1.03 8.19E-07 Phospholipid transfer protein P55058 1.55 2.72E-06 Protein FAM135A Q9P2D6 0.67 1.16E-05 IG c617_light_IGKV3-15_IGKJ5 A0A5C2GTE2 0.65 1.51E-05 HBD P02042 1.01 1.67E-05 C1Inh P05155 1.01 3.35E-05 HBB P68871 1.10 3.73E-05 IG c17_heavy_IGHV5-51_IGHD5-12_IGHJ6(Fragment) A0A5C2GIT4 0.72 5.14E-05 CA-II P00918 0.93 6.93E-05 IGL c1223_light_IGLV8-61_IGLJ3 A0A5C2G4M5 0.76 7.93E-05 IG c1553_heavy_IGHV3-7_IGHD3-16_IGHJ5(Fragment) A0A5C2GRN7 0.74 1.91E-04 ITIH4 protein B7ZKJ8 0.75 1.96E-04 IGL c3507_light_IGKV4-1_IGKJ3 A0A5C2G731 0.63 2.05E-04 Reticulocalbin-1; Epididymis secretory protein Li 84 V9HW95 0.73 2.25E-04 LRG P02750 0.63 2.56E-04 CA-I P00915 0.80 5.33E-04 IG c195_heavy_IGHV3-33_IGHD3-22_IGHJ6(Fragment) A0A5C2GKW6 0.60 5.94E-04 Immunogobulin kappa, VJ region A2NH54 0.78 7.56E-04 HP2 P59665 0.76 7.99E-04 IGL c1474_light_IGLV1-40_IGLJ3(Fragment) A0A5C2H1L6 0.59 8.33E-04 IGLV8-61 A0A075B6I0 0.67 9.79E-04 IGH c584_heavy__IGHV5-51_IGHD1-26_IGHJ4(Fragment) A0A5C2G9F4 0.65 1.07E-03 H2B K O60814 0.91 1.75E-03 cDNA FLJ39107fis,clone NTONG2005062 Q8N1N1 0.99 2.00E-03 IGH@protein Q6P089 0.75 2.13E-03 IGL c1553_light_IGKV4-1_IGKJ2(Fragment) A0A5C2G5F2 0.73 2.51E-03 F5-20(Fragment) A0N7I9 0.75 5.03E-03 IGL c1361_light_IGKV3-11_IGKJ5(Fragment) A0A5C2FWF0 0.70 5.42E-03 IG c1800_heavy_IGHV3-23_IGHD1-26_IGHJ3(Fragment) A0A5C2GTG7 0.67 5.44E-03 Zinc finger protein 587B M0QY62 0.81 5.46E-03 Testicular tissue protein Li 61;Extracellular matrix protein 1 Q16610 0.61 5.48E-03 IGL c405_light_IGKV3-11_IGKJ4(Fragment) A0A5C2FXQ6 0.59 5.76E-03 Uncharacterized protein DKFZp686C02220(Fragment) Q6N091 0.66 6.06E-03 Complement factor D P00746 0.75 7.84E-03 Peroxiredoxin-2 P32119 1.22 7.96E-03 Serum paraoxonase / lactonase 3 Q15166 0.65 8.86E-03 SAA P0DJI8 1.80 9.59E-03 IG c1054_light_IGKV1-27_IGKJ1(Fragment) A0A5C2GU59 0.64 1.18E-02 IG c1347_light_IGKV2-30_IGKJ4(Fragment) A0A5C2GVL1 0.68 1.26E-02 Beta-globin Showa Yakushiji variant B2M1S7 1.72 1.50E-02 IG c1040_light_IGKV1-6_IGKJ1 A0A5C2GPK3 1.09 1.54E-02 IGL c3116_light_IGLV2-18_IGLJ3(Fragment) A0A5C2G3P7 0.59 1.62E-02 cDNA FLJ50830,highly similar to Serum albumin B4DPR2 0.65 2.65E-02 IG c559_heavy_IGHV3-53_IGHD3-9_IGHJ2(Fragment) A0A5C2GRQ1 0.59 3.20E-02 IG c288_light_IGLV1-40_IGLJ1 A0A5C2GX62 0.62 3.20E-02 Folliculin-interacting protein 2 Q9P278 0.67 3.52E-02 IGL c3197_light_IGKV2D-29_IGKJ4 A0A5C2G6C5 1.11 3.78E-02 IGH+IGL c553_heavy_IGHV4-39_IGHD6-19_IGHJ3(Fragment) A0A5C2GA02 0.82 3.79E-02 IGL c3029_light_IGLV3-21_IGLJ2(Fragment) A0A5C2G1J3 0.80 3.99E-02 CAT P04040 3.33 4.02E-02 IGL c492_light_IGLV3-21_IGLJ1(Fragment) A0A5C2GAN8 0.90 4.18E-02 ADAM-TS 13 Q76LX8 0.82 4.56E-02

[0182] (vii) Data standardization: Based on preliminary analysis, the inventors found that the combinations of protein biomarkers obtained by different standardization methods may vary greatly. Considering the effectiveness of practical application, the inventors of this invention used eight standardization methods (Log2, Log2+Median, Log2+Mean, VSN, Log2+RLR, Log2+GI, Log2+Quantile, and Log2+CycLoess) to standardize the protein expression matrices processed in steps (i) to (v). All eight standardized protein expression matrices were then subjected to partial least squares regression analysis and protein biomarker screening based on a random forest model using a genetic algorithm.

[0183] (viii) Partial Least Squares Regression Analysis (PLS): Partial least squares regression analysis was used to screen protein markers for the protein expression matrices obtained by the eight standardization methods, and the top 30 markers for each method were selected. The results are shown in Table 3. The union of all proteins in Table 3 was used to obtain protein set ③, and the frequency information of all proteins in this set is shown in Table 4.

[0184] Table 3 lists the protein biomarkers obtained by partial least squares regression analysis after standardization using eight different methods.

[0185]

[0186] Table 4 shows the frequency information of proteins in the union of the protein biomarker lists obtained by partial least squares regression analysis after standardization using eight methods.

[0187]

[0188] (ix) Random Forest Model Based on Genetic Algorithm (GA-RF) for Scoring and Screening Protein Biomarkers: The genetic algorithm has the characteristics of a search heuristic algorithm, meaning that protein biomarkers with strong classification effects have more opportunities to be evaluated, enhancing the stability of the random forest evaluation results; at the same time, the mutation process of the genetic algorithm also allows protein biomarkers with weak classification effects to have the opportunity to be evaluated. To ensure the stability of the selected protein biomarkers, protein biomarkers were screened through 150 random resampling and extraction of different protein subsets, and each resampling process was validated using 10-fold cross-validation repeated 5 times. Finally, the 5 proteins with the highest frequency in all resampling processes were selected. The protein expression matrix was normalized by Log2+CycLoess and then screened by GA-RF to obtain the 5 proteins P04040, A0A1S5UZ39, A0A3B3IQ51, P02750, and P32119. The protein expression matrix was normalized using Log2+GI, and then screened using GA-RF to obtain five proteins: O60814, P02750, P00918, P02748, and P04040. The protein expression matrix was normalized using Log2, and then screened using GA-RF to obtain five proteins: A0A1S5UZ39, O60814, P04040, P00918, and P68871. The protein expression matrix was normalized using Log2+Mean, and then screened using GA-RF to obtain five proteins: O60814, P02750, P00918, P04040, and A0A1S5UZ39. The protein expression matrix was normalized using Log2+Median, and then screened using GA-RF to obtain five proteins: P04040, P02750, O60814, A0A1S5UZ39, and A0A5C2GX62. The protein expression matrix was normalized using Log2+Quantile and then screened using GA-RF to obtain five proteins: P04040, A0A3B3IQ51, P59665, P02750, and V9H1D9. The protein expression matrix was normalized using Log2+RLR and then screened using GA-RF to obtain five proteins: Q14624, P04040, P59665, A0A1S5UZ39, and P02750. The protein expression matrix was normalized using VSN and then screened using GA-RF to obtain five proteins: A0A1S5UZ39, O60814, B7ZKJ8, A0A3B3IQ51, and Q86UD1.The proteins obtained from the screening results of the eight standardized methods were combined to form set ④, which includes proteins P04040, P02750, A0A1S5UZ39, O60814, A0A3B3IQ51, P00918, P59665, Q14624, B7ZKJ8, P02748, P32119, P68871, A0A5C2GX62, V9H1D9, and Q86UD1. The frequencies of these proteins in set ④ in the screening results of the eight standardized methods were counted and sorted from high to low frequency. The sorting results are shown in Table 5.

[0189] Table 5 shows the frequency information of proteins in the union of the protein biomarker lists obtained by the random forest model based on genetic algorithms after standardization using eight methods.

[0190]

[0191] (X) Table 6 summarizes the protein biomarkers obtained by the four methods in Tables 1, 2, 4, and 5, and their frequency of occurrence in the results of all 18 methods. These 18 methods specifically refer to: (1) Identifying non-randomly missing proteins by identifying missing values; (2) Identifying differentially expressed proteins by differential expression analysis; (3) Standardizing the protein expression matrix using Log2+CycLoess, followed by PLS screening; (4) Standardizing the protein expression matrix using Log2+GI, followed by PLS screening; (5) Standardizing the protein expression matrix using Log2, followed by PLS screening; (6) Standardizing the protein expression matrix using Log2+Mean, followed by PLS screening; (7) Standardizing the protein expression matrix using Log2+Median, followed by PLS screening; (8) Standardizing the protein expression matrix using Log2+Quantile, followed by PLS screening; (9) Standardizing the protein expression matrix using Log2+RLR, followed by PLS screening; (10) Protein expression... The protein expression matrix was normalized by VSN and then screened by PLS; (11) The protein expression matrix was normalized by Log2+CycLoess and then screened by GA-RF; (12) The protein expression matrix was normalized by Log2+GI and then screened by GA-RF; (13) The protein expression matrix was normalized by Log2 and then screened by GA-RF; (14) The protein expression matrix was normalized by Log2+Mean and then screened by GA-RF; (15) The protein expression matrix was normalized by Log2+Median and then screened by GA-RF; (16) The protein expression matrix was normalized by Log2+Quantile and then screened by GA-RF; (17) The protein expression matrix was normalized by Log2+RLR and then screened by GA-RF; (18) The protein expression matrix was normalized by VSN and then screened by GA-RF. The six proteins with the highest frequency from Table 6, namely P04040, O60814, A0A1S5UZ39, P02750, P00918 and A0A3B3IQ51, were selected for subsequent logistic regression modeling.

[0192] Table 6 shows the union of protein biomarker sets ①, ②, ③, and ④, and the frequency of protein biomarkers within these sets across all 18 methods.

[0193] Example 2: Establishment and Validation of a Logistic Regression Prediction Model

[0194] (1) The proteomic content of samples from known sources was determined by mass spectrometry and UniProt was searched to obtain the protein expression matrix of the samples from known sources, which consisted of 40 benign gastric disease samples and 39 gastric cancer samples.

[0195] (2) The protein expression matrix above was standardized using eight standardization methods (Log2, Log2+Median, Log2+Mean, VSN, Log2+RLR, Log2+GI, Log2+Quantile and Log2+CycLoess);

[0196] (3) All samples were divided into training and test sets in a stratified sampling ratio of 8:2.

[0197] (4) For the protein expression matrices obtained by the eight standardization methods, the training set sample data was used to train the logistic regression model, perform 5-fold cross-validation, and perform internal validation using the bootstrap method, as detailed below:

[0198] [1]. Four protein biomarkers, P04040, A0A1S5UZ39, P02750, and P00918, were used to build models on the training set, resulting in eight logistic regression prediction models with four factors. The stability and effectiveness of the trained models were tested by performing 1000 random samplings with replacement on the training set using the bootstrap method. The mean and standard deviation of the AUC for the eight standardization methods are shown in Table 7. The results of model cross-validation are shown in... Figure 5 The probability prediction formulas obtained by logistic regression modeling for protein biomarkers P04040, A0A1S5UZ39, P02750, and P00918 after processing with different standardization methods are shown in Table 8.

[0199] Table 7 shows the mean and standard deviation of the AUC of the logistic regression models obtained after processing protein biomarkers P04040, A0A1S5UZ39, P02750, and P00918 using different standardization methods, validated internally using bootstrap.

[0200]

[0201] Table 8. Probability prediction formulas for protein biomarkers P04040, A0A1S5UZ39, P02750, and P00918 obtained by logistic regression modeling after treatment with different standardization methods.

[0202]

[0203] [2]. Four protein biomarkers, A0A3B3IQ51, P02750, O60814, and P00918, were used to build models on the training set, resulting in eight logistic regression prediction models with four factors. The stability and effectiveness of the trained models were tested by performing 1000 random samplings with replacement on the training set using the bootstrap method. The mean and standard deviation of the AUC for the eight standardization methods are shown in Table 9. The results of model cross-validation are shown in... Figure 6The probability prediction formulas obtained by logistic regression modeling for protein biomarkers A0A3B3IQ51, P02750, O60814, and P00918 after processing with different standardization methods are shown in Table 10.

[0204] Table 9 shows the mean and standard deviation of the AUC of the logistic regression models obtained after processing protein biomarkers A0A3B3IQ51, P02750, O60814, and P00918 using different standardization methods, validated internally using bootstrap methods.

[0205]

[0206] Table 10. Probability prediction formulas for protein biomarkers A0A3B3IQ51, P02750, O60814, and P00918 obtained by logistic regression modeling after treatment with different standardization methods.

[0207]

[0208] [3]. Using three protein biomarkers, A0A3B3IQ51, P02750, and P00918, models were built on the training set, resulting in eight logistic regression prediction models with three factors. The stability and effectiveness of the trained models were tested by performing 1000 random samplings with replacement on the training set using the bootstrap method. The mean and standard deviation of the AUC for the eight standardization methods are shown in Table 11. The results of model cross-validation are shown in... Figure 7 The probability prediction formulas obtained by logistic regression modeling for protein biomarkers A0A3B3IQ51, P02750, and P00918 after processing with different standardization methods are shown in Table 12.

[0209] Table 11 shows the mean and standard deviation of the AUC of the logistic regression models obtained after processing protein biomarkers A0A3B3IQ51, P02750, and P00918 using different standardization methods, validated internally using bootstrap.

[0210]

[0211] Table 12 Probability prediction formulas for protein biomarkers A0A3B3IQ51, P02750, and P00918 obtained by logistic regression modeling after processing with different standardization methods.

[0212]

[0213] [4]. Using the three protein biomarkers A0A1S5UZ39, P02750, and P00918, models were built on the training set, resulting in eight logistic regression prediction models containing three factors. The stability and effectiveness of the trained models were tested by performing 1000 random samplings on the training set using the bootstrap method. The mean and standard deviation of the AUC for the eight standardization methods are shown in Table 13. The results of model cross-validation are shown in... Figure 8 The probability prediction formulas obtained by logistic regression modeling for protein biomarkers A0A1S5UZ39, P02750, and P00918 after processing with different standardization methods are shown in Table 14.

[0214] Table 13 shows the mean and standard deviation of the AUC of the logistic regression models obtained after processing protein biomarkers A0A1S5UZ39, P02750, and P00918 using different standardization methods, validated internally using bootstrap.

[0215]

[0216] Table 14 Probability prediction formulas for protein biomarkers A0A1S5UZ39, P02750, and P00918 obtained by logistic regression modeling after processing with different standardization methods.

[0217]

[0218] [5]. Models were constructed on the training set using the protein biomarkers A0A3B3IQ51 and P02750, resulting in eight logistic regression prediction models with two factors. The stability and effectiveness of the trained models were tested by performing 1000 random samplings with replacement on the training set using the bootstrap method. The mean and standard deviation of the AUC for the eight standardization methods are shown in Table 15. The results of model cross-validation are shown in... Figure 9 The probability prediction formulas for protein biomarkers A0A3B3IQ51 and P02750 obtained by logistic regression modeling after processing with different standardization methods are shown in Table 16.

[0219] Table 15 shows the mean and standard deviation of the AUC of the logistic regression models obtained after processing protein biomarkers A0A3B3IQ51 and P02750 using different standardization methods, validated internally using bootstrap.

[0220]

[0221] Table 16 Probability prediction formulas for protein biomarkers A0A3B3IQ51 and P02750 obtained by logistic regression modeling after treatment with different standardization methods.

[0222]

[0223] [6]. Using the protein biomarkers A0A1S5UZ39 and P02750, eight logistic regression prediction models with two factors were constructed on the training set. The stability and effectiveness of the trained models were tested by performing 1000 random samplings with replacement on the training set using the bootstrap method. The mean and standard deviation of the AUC for the eight standardization methods are shown in Table 17. The results of model cross-validation are shown in... Figure 10 The probability prediction formulas for protein biomarkers A0A1S5UZ39 and P02750 obtained by logistic regression modeling after processing with different standardization methods are shown in Table 18.

[0224] Table 17 shows the mean and standard deviation of the AUC of the logistic regression models obtained after different standardization methods for protein biomarkers A0A1S5UZ39 and P02750, validated internally using bootstrap methods.

[0225]

[0226] Table 18 Probability prediction formulas for protein biomarkers A0A1S5UZ39 and P02750 obtained by logistic regression modeling after treatment with different standardization methods.

[0227]

[0228] (5) The predicted performance of the trained model is evaluated using test set samples, as follows:

[0229] [1]. The prediction results of eight logistic regression prediction models constructed using four protein biomarkers, P04040, A0A1S5UZ39, P02750, and P00918, under eight normalization methods on the test set are shown in Table 19. The ROC curves are shown in [Table 19]. Figure 11 .

[0230] Table 19 Predictive performance of protein biomarkers P04040, A0A1S5UZ39, P02750, and P00918 on the test set after logistic regression modeling following different standardization methods.

[0231]

[0232] [2]. The prediction results of eight logistic regression prediction models constructed using four protein biomarkers, A0A3B3IQ51, P02750, O60814, and P00918, under eight normalization methods on the test set are shown in Table 20. The ROC curves are shown in [Table 20]. Figure 12 .

[0233] Table 20 shows the predictive performance of protein biomarkers A0A3B3IQ51, P02750, O60814, and P00918 on the test set after logistic regression modeling following different standardization methods.

[0234]

[0235] [3]. The prediction results of eight logistic regression prediction models constructed using the three protein biomarkers A0A3B3IQ51, P02750, and P00918 under eight normalization methods on the test set are shown in Table 21. The ROC curves are shown in [Table 21]. Figure 13 .

[0236] Table 21 Predictive performance of protein biomarkers A0A3B3IQ51, P02750, and P00918 on the test set after logistic regression modeling following different standardization methods.

[0237]

[0238] [4]. The prediction results of eight logistic regression prediction models constructed using the three protein biomarkers A0A1S5UZ39, P02750, and P00918 under eight normalization methods on the test set are shown in Table 22. The ROC curves are shown in [Table 22]. Figure 14 .

[0239] Table 22 Predictive performance of protein biomarkers A0A1S5UZ39, P02750, and P00918 on the test set after logistic regression modeling following different standardization methods.

[0240]

[0241] [5]. The prediction results of eight logistic regression prediction models constructed using the two protein biomarkers A0A3B3IQ51 and P02750 under eight normalization methods on the test set are shown in Table 23. The ROC curves are shown in [Table 23]. Figure 15 .

[0242] Table 23 Predictive performance of protein biomarkers A0A3B3IQ51 and P02750 on the test set after logistic regression modeling following different standardization methods.

[0243]

[0244] [6]. The prediction results of eight logistic regression prediction models constructed using the two protein biomarkers A0A1S5UZ39 and P02750 under eight normalization methods on the test set are shown in Table 24. The ROC curves are shown in […]. Figure 16 .

[0245] Table 24 Predictive performance of protein biomarkers A0A1S5UZ39 and P02750 on the test set after logistic regression modeling following different standardization methods.

[0246]

Claims

1. The application of reagents for detecting serum protein biomarkers in the preparation of kits for gastric cancer screening, characterized in that, The serum protein biomarkers are P04040, A0A1S5UZ39, P02750 and P00918.

2. The application as described in claim 1, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 were measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 and P04040 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+CycLoess standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+CycLoess standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1+Exp(-(-56.171-0.692)). P04040+0.228 A0A1S5UZ39+5.009 P02750-2.540 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

3. The application as described in claim 1, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 were measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 and P04040 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+GI standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+GI standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-77.717-0.478)). P04040-0.341 A0A1S5UZ39+5.392 P02750-1.027 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

4. The application as described in claim 1, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 were measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 and P04040 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2 standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2 standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-33.799-0.473)). P04040-0.440 A0A1S5UZ39+4.019 P02750-1.976 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

5. The application as described in claim 1, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 were measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 and P04040 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Mean standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Mean standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-77.536 - 0.478)). P04040-0.337 A0A1S5UZ39+5.378 P02750-1.035 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

6. The application as described in claim 1, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 were measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 and P04040 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Median standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Median standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-55.721-0.441)). P04040-0.473 A0A1S5UZ39+4.591 P02750-1.270 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

7. The application as described in claim 1, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 were measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 and P04040 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Quantile standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Quantile standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1+Exp(-(-67.461-0.647)). P04040+0.550 A0A1S5UZ39+5.213 P02750-2.583 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

8. The application as described in claim 1, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 were measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 and P04040 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+RLR standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+RLR standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-59.715-0.680)). P04040+0.382 A0A1S5UZ39+4.723 P02750-2.191 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

9. The application as described in claim 1, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers P04040, A0A1S5UZ39, P02750 and P00918 were measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of P00918 and P04040 is filled with a constant of 1.0, and the missing protein content of A0A1S5UZ39 and P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the VSN normalization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the VSN normalization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-44.111-0.269)). P04040-1.489 A0A1S5UZ39+5.961 P02750-2.395 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

10. The application of reagents for detecting serum protein biomarkers in the preparation of kits for gastric cancer screening, characterized in that, The serum protein biomarkers are A0A3B3IQ51, P02750, O60814, and P00918.

11. The application as described in claim 10, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750, O60814 and P00918 were measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the three proteins A0A3B3IQ51, P00918 and O60814 will be filled with a constant of 1.0, and the missing protein content of P02750 will be filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+CycLoess standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+CycLoess standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1+Exp(-(-329.023+3.922)). A0A3B3IQ51+11.630 P02750+4.905 O60814-2.136 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

12. The application as described in claim 10, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750, O60814 and P00918 were measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the three proteins A0A3B3IQ51, P00918 and O60814 will be filled with a constant of 1.0, and the missing protein content of P02750 will be filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+GI standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+GI standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-229.243 + 2.236)). A0A3B3IQ51+10.504 P02750+2.630 O60814-3.348 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

13. The application as described in claim 10, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750, O60814 and P00918 were measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the three proteins A0A3B3IQ51, P00918 and O60814 will be filled with a constant of 1.0, and the missing protein content of P02750 will be filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2 standardization method, and the predicted probability P-value is calculated using the corresponding probability prediction formula. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-262.585 + 3.917)). A0A3B3IQ51+7.231 P02750+4.551 O60814-0.432 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

14. The application as described in claim 10, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750, O60814 and P00918 were measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the three proteins A0A3B3IQ51, P00918 and O60814 will be filled with a constant of 1.0, and the missing protein content of P02750 will be filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Mean standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Mean standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-224.074 + 2.196)). A0A3B3IQ51+10.256 P02750+2.566 O60814-3.293 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

15. The application as described in claim 10, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750, O60814 and P00918 were measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the three proteins A0A3B3IQ51, P00918 and O60814 will be filled with a constant of 1.0, and the missing protein content of P02750 will be filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Median standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Median standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-118.175 + 1.487)). A0A3B3IQ51+5.076 P02750+1.751 O60814-2.002 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

16. The application as described in claim 10, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750, O60814 and P00918 were measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the three proteins A0A3B3IQ51, P00918 and O60814 will be filled with a constant of 1.0, and the missing protein content of P02750 will be filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Quantile standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Quantile standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-108.393 + 0.910)). A0A3B3IQ51+5.493 P02750+1.464 O60814-2.315 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

17. The application as described in claim 10, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750, O60814 and P00918 were measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the three proteins A0A3B3IQ51, P00918 and O60814 will be filled with a constant of 1.0, and the missing protein content of P02750 will be filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+RLR standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+RLR standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-192.053 + 2.244)). A0A3B3IQ51+8.082 P02750+2.851 O60814-2.860 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

18. The application as described in claim 10, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750, O60814 and P00918 were measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of the three proteins A0A3B3IQ51, P00918 and O60814 will be filled with a constant of 1.0, and the missing protein content of P02750 will be filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the VSN normalization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the VSN normalization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-195.306 + 3.888)). A0A3B3IQ51+6.315 P02750+3.570 O60814-2.862 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

19. The application of reagents for detecting serum protein biomarkers in the preparation of kits for gastric cancer screening, characterized in that, The serum protein biomarkers are A0A3B3IQ51, P02750, and P00918.

20. The application as described in claim 19, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750 and P00918 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 and P00918 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+CycLoess standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+CycLoess standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1+Exp(-(-85.208+0.976)). A0A3B3IQ51+4.524 P02750-1.236 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

21. The application as described in claim 19, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750 and P00918 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 and P00918 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+GI standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+GI standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-109.138 + 1.314)). A0A3B3IQ51+4.946 P02750-0.566 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

22. The application as described in claim 19, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750 and P00918 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 and P00918 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2 standardization method, and the predicted probability P value is calculated using the probability prediction formula corresponding to the Log2 standardization method. When the P value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-75.531 + 1.118)). A0A3B3IQ51+3.545 P02750-0.749 P00918)));In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

23. The application as described in claim 19, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750 and P00918 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 and P00918 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Mean standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Mean standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-108.872 + 1.309)). A0A3B3IQ51+4.927 P02750-0.569 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

24. The application as described in claim 19, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750 and P00918 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 and P00918 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Median standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Median standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-84.630 + 1.328)). A0A3B3IQ51+3.794 P02750-0.696 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

25. The application as described in claim 19, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750 and P00918 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 and P00918 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Quantile standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Quantile standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-82.513 + 0.526)). A0A3B3IQ51+5.121 P02750-1.750 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

26. The application as described in claim 19, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750 and P00918 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 and P00918 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+RLR standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+RLR standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-87.794 + 1.146)). A0A3B3IQ51+4.335 P02750-1.002 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

27. The application as described in claim 19, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51, P02750 and P00918 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 and P00918 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the VSN normalization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the VSN normalization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-86.244 + 1.988)). A0A3B3IQ51+4.499 P02750-2.146 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

28. The application of reagents for detecting serum protein biomarkers in the preparation of kits for gastric cancer screening, characterized in that, The serum protein biomarkers are A0A1S5UZ39, P02750, and P00918.

29. The application as described in claim 28, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39, P02750 and P00918 was measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing content of P00918 protein is filled with a constant of 1.0, and the missing content of A0A1S5UZ39 and P02750 proteins is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+CycLoess standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+CycLoess standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1+Exp(-(-63.622-1.625)). A0A1S5UZ39+6.022 P02750-1.335 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

30. The application as described in claim 28, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39, P02750 and P00918 was measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing content of P00918 protein is filled with a constant of 1.0, and the missing content of A0A1S5UZ39 and P02750 proteins is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+GI standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+GI standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-83.752-1.821)). A0A1S5UZ39+6.673 P02750-0.584 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

31. The application as described in claim 28, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39, P02750 and P00918 was measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing content of P00918 protein is filled with a constant of 1.0, and the missing content of A0A1S5UZ39 and P02750 proteins is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2 standardization method, and the predicted probability P value is calculated using the probability prediction formula corresponding to the Log2 standardization method. When the P value is ≥ 0.5, the sample is identified as a gastric cancer sample; when the P value is < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-34.719-1.630)). A0A1S5UZ39+4.617 P02750-1.397 P00918)));In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

32. The application as described in claim 28, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39, P02750 and P00918 was measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing content of P00918 protein is filled with a constant of 1.0, and the missing content of A0A1S5UZ39 and P02750 proteins is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Mean standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Mean standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-83.501-1.818)). A0A1S5UZ39+6.652 P02750-0.588 P00918)));In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

33. The application as described in claim 28, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39, P02750 and P00918 was measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing content of P00918 protein is filled with a constant of 1.0, and the missing content of A0A1S5UZ39 and P02750 proteins is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Median standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Median standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-66.466-1.714)). A0A1S5UZ39+5.801 P02750-0.741 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

34. The application as described in claim 28, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39, P02750 and P00918 was measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing content of P00918 protein is filled with a constant of 1.0, and the missing content of A0A1S5UZ39 and P02750 proteins is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Quantile standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Quantile standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-67.479-1.352)). A0A1S5UZ39+6.143 P02750-1.629 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

35. The application as described in claim 28, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39, P02750 and P00918 was measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing content of P00918 protein is filled with a constant of 1.0, and the missing content of A0A1S5UZ39 and P02750 proteins is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+RLR standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+RLR standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-66.818-1.468)). A0A1S5UZ39+5.875 P02750-1.157 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

36. The application as described in claim 28, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39, P02750 and P00918 was measured respectively. b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing content of P00918 protein is filled with a constant of 1.0, and the missing content of A0A1S5UZ39 and P02750 proteins is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the VSN normalization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the VSN normalization method. When the P-value is ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value is < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-54.044-1.640)). A0A1S5UZ39+6.054 P02750-1.930 P00918))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

37. The application of reagents for detecting serum protein biomarkers in the preparation of kits for gastric cancer screening, characterized in that, The serum protein biomarkers are A0A3B3IQ51 and P02750.

38. The application as described in claim 37, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+CycLoess standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+CycLoess standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1+Exp(-(-88.204+1.621)). A0A3B3IQ51+3.263 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

39. The application as described in claim 23, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+GI standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+GI standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-100.532 + 1.676)). A0A3B3IQ51+3.841 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

40. The application as described in claim 37, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2 standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2 standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-78.997 + 1.513)). A0A3B3IQ51+2.872 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

41. The application as described in claim 37, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Mean standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Mean standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-100.271 + 1.672)). A0A3B3IQ51+3.820 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

42. The application as described in claim 37, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Median standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Median standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-79.809 + 1.648)). A0A3B3IQ51+2.808 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

43. The application as described in claim 37, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+Quantile standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Quantile standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-84.715 + 1.341)). A0A3B3IQ51+3.293 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

44. The application as described in claim 37, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the Log2+RLR standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+RLR standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-88.211 + 1.571)). A0A3B3IQ51+3.296 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

45. The application as described in claim 37, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A3B3IQ51 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the missing protein content of A0A3B3IQ51 is filled with a constant of 1.0, and the missing protein content of P02750 is filled with the K nearest neighbor method. c) The protein content data obtained in step (b) is standardized using the VSN normalization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the VSN normalization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-103.612 + 2.559)). A0A3B3IQ51+3.265 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

46. ​​The application of reagents for detecting serum protein biomarkers in the preparation of kits for gastric cancer screening, characterized in that, The serum protein biomarkers are A0A1S5UZ39 and P02750.

47. The application as described in claim 46, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the K-nearest neighbor method is used to imput the missing content of the two proteins A0A1S5UZ39 and P02750. c) The protein content data obtained in step (b) is standardized using the Log2+CycLoess standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+CycLoess standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1+Exp(-(-53.231-2.793)). A0A1S5UZ39+5.764 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

48. The application as described in claim 46, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the K-nearest neighbor method is used to imput the missing content of the two proteins A0A1S5UZ39 and P02750. c) The protein content data obtained in step (b) is standardized using the Log2+GI standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+GI standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-75.499-3.016)). A0A1S5UZ39+7.149 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

49. The application as described in claim 46, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the K-nearest neighbor method is used to imput the missing content of the two proteins A0A1S5UZ39 and P02750. c) The protein content data obtained in step (b) is standardized using the Log2 standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2 standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-36.635-2.676)). A0A1S5UZ39+4.801 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

50. The application as described in claim 46, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the K-nearest neighbor method is used to imput the missing content of the two proteins A0A1S5UZ39 and P02750. c) The protein content data obtained in step (b) is standardized using the Log2+Mean standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Mean standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-75.257-3.011)). A0A1S5UZ39+7.124 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

51. The application as described in claim 46, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the K-nearest neighbor method is used to imput the missing content of the two proteins A0A1S5UZ39 and P02750. c) The protein content data obtained in step (b) is standardized using the Log2+Median standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Median standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-55.018-2.803)). A0A1S5UZ39+5.875 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

52. The application as described in claim 46, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the K-nearest neighbor method is used to imput the missing content of the two proteins A0A1S5UZ39 and P02750. c) The protein content data obtained in step (b) is standardized using the Log2+Quantile standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+Quantile standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-56.416-2.719)). A0A1S5UZ39+5.852 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

53. The application as described in claim 46, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the K-nearest neighbor method is used to imput the missing content of the two proteins A0A1S5UZ39 and P02750. c) The protein content data obtained in step (b) is standardized using the Log2+RLR standardization method, and the predicted probability P-value is calculated using the probability prediction formula corresponding to the Log2+RLR standardization method. When the P-value ≥ 0.5, the sample is identified as a gastric cancer sample; when the P-value < 0.5, the sample is identified as a benign gastric disease sample. The probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-61.631-2.649)). A0A1S5UZ39+6.042 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

54. The application as described in claim 46, characterized in that, The detection method of the kit includes: a) For the samples, the protein content of serum protein biomarkers A0A1S5UZ39 and P02750 was measured respectively; b) Missing value imputation: If the content of one or more serum protein biomarkers is missing in the test results, the K-nearest neighbor method is used to imput the missing content of the two proteins A0A1S5UZ39 and P02750. c) The protein content data obtained in step (b) is standardized using the VSN normalization method, and the predicted probability P value is calculated using the probability prediction formula corresponding to the VSN normalization method; when the P value ≥ 0.5, it is determined to be a gastric cancer sample; when the P value < 0.5, it is determined to be a benign gastric disease sample; the probability prediction formula for the serum protein biomarker is P = 1 / (1 + Exp(-(-60.780-2.778)). A0A1S5UZ39+6.136 P02750))); In the formula, the name of the protein biomarker refers to the standardized content value of the protein biomarker.

Images