Host gene combination for diagnosing acute bacterial and virus infection and application thereof

Abstract

The invention discloses a host gene combination for diagnosing acute bacterial and virus infection and application of the host gene combination. According to the present invention, with 100 related characteristic genes, the rapid diagnosis of acute bacterial and virus infection can be achieved; according to the detection result of the multi-gene combination, acute bacterial and virus infection, risks and types are diagnosed, clinical information and the like of patients are combined, more understanding is provided for clinical infection type cancer progression of the patients, accurate diagnosis and treatment are facilitated, and the clinical diagnosis and treatment efficiency is improved. Meanwhile, detailed data are provided for detection standardization of patients with acute bacterial and virus infection risks.

Description

Technical Field

[0001] The present invention belongs to the field of gene technology and medical health, and in particular to host gene expression profiles for diagnosing patient infections and infection source types. Background Art

[0002] Acute severe infection and sepsis remain one of the leading causes of death worldwide (1,2). Since the symptoms of non-infectious inflammation and infection are extremely similar, it is extremely difficult to accurately diagnose acute infection in the early stages. The timely use of antibiotics in the early stages of acute infection is crucial to saving lives, but the misuse of antibiotics not only leads to higher morbidity, but also increases treatment costs and promotes the rapid spread of antimicrobial resistance (3-5). Current diagnostic methods mainly rely on direct detection of pathogens or evaluation of host biomarkers. However, early use of antibiotics may inhibit the growth of pathogens, thereby affecting the positive rate of culture results and even failing to effectively distinguish infection from colonization. In addition, although molecular detection technologies such as PCR are accurate, they may miss other potential pathogens due to the limitations of their target range (6-8). In this context, metagenomic next-generation sequencing (mNGS) has shown great application prospects as a universal pathogen detection technology, but its specificity still needs to be further optimized (9).

[0003] In addition to direct detection of pathogens, analysis of the host's immune response provides another approach to diagnosing acute infection. However, due to the high complexity of the host's immune response, a single or small number of traditional biomarkers (such as C-reactive protein, procalcitonin, white blood cell count, etc.) are insufficient to accurately diagnose sepsis or effectively stratify patients for personalized treatment (10). At the same time, the infection score (Infect-Score) or bacterial-viral infection score (BV-Score) alone cannot robustly distinguish between all three categories of patients: (1) non-infection, (2) bacterial infection, and (3) viral infection. This is the pain point and difficulty in the current clinical diagnosis of acute infection.

[0004] The host gene expression profile in the peripheral blood produced by different types of inflammation caused by different sources of infection is specific. Studies have found that the host's immune response to inflammation caused by different sources of infection has shared transcriptome characteristics, and different sources of infection will have different effects on these transcriptomes (different expressions). Based on this principle, using genetic means (mNGS and PCR technology, etc.) to accurately diagnose the types of acute inflammation caused by different sources of infection by conducting infection based on gene expression ranking is currently an effective way. Summary of the invention

[0005] The technical problem to be solved by the present invention is to overcome the deficiencies in the prior art and provide a host gene combination and device for diagnosing acute bacterial and viral infections.

[0006] To solve the technical problem, the solution of the present invention is:

[0007] A host gene panel for diagnosing acute bacterial and viral infections, the panel includes the following 100 relevant characteristic genes:

[0008]

[0009] In the present invention, the expression of the 100 related characteristic genes in the gene combination is related to the infection type, and the relative expression values ​​of the 100 related characteristic genes in the whole genome sequencing results of the blood of patients with bacterial infection, viral infection or non-infectious diseases are different.

[0010] Using the gene combination to determine the infection type includes the following steps:

[0011] (1) Extract and preserve whole blood samples from patients;

[0012] (2) Determine the gene expression data of the sample in the transcriptome. The expression of 100 relevant characteristic genes at the RNA level can be obtained by RNA-seq technology: extract total RNA from the sample in (1), sort and build a library, sequence it, and perform quality control and trimming on the sequencing sequence. The obtained sequence is compared with the human reference genome (GRCh38). According to the comparison results, the expression level of each relevant characteristic gene per kilobase per million fragments is obtained by transcript splicing and quantitative software (such as Salmon, HISAT2, etc.), which is the relative expression value FPKM of the 100 relevant characteristic genes. FPKM is obtained by adjusting the original read count (i.e., the comparison result) according to the sequencing depth and transcript length. Therefore, the depth and transcript length will not affect the gene expression level in RNA-seq data.

[0013] (3) The relative expression value FPKM of each relevant characteristic gene obtained in step (2) is input into a device for predicting the infection type of the patient, and the infection type of the patient can be obtained. The calculation formula of FPKM is:

[0014]

[0015] The present invention also provides a device for predicting the type of infection of a patient, which uses the relative expression value FPKM of each relevant characteristic gene in the above-mentioned gene combination to predict the infection type. The device for predicting the type of infection of a patient includes an RNA level gene expression data calculation module, an infection inference module and an infection type inference module, wherein the RNA level gene expression data is used to obtain the relative expression value FPKM of the 100 relevant characteristic genes; the infection inference module is used to determine whether the sample is infected according to the relative expression value FPKM of the relevant characteristic genes; if the inference result of the infection inference module is not infected, it is directly output; if the inference result of the infection inference module is infection, the infection type inference module and the relative expression value FPKM of the relevant characteristic genes are further used to further judge the infected sample to obtain the infection type of the sample. The input of the device is the relative expression value FPKM of each relevant characteristic gene in the gene combination, and the output result is not infected or a specific infection type, and the infection type includes viral infection and bacterial infection.

[0016] The present invention is based on a weighted integrated machine learning method based on k-TSP. It can obtain the corresponding risk stratification of infection types by calculation based on the sequencing results of patient tissue samples (the relative expression value FPKM of the characteristic gene), and determine the patient's infection type according to the judgment of the risk stratification of the infection type.

[0017] In the present invention, characteristic genes related to bacterial and viral infections are obtained based on the differences in expression levels of confirmed non-infectious samples, bacterial infection samples and viral infection samples, and then the above samples and related characteristic genes are used to train a patient infection type prediction model, which includes an infection classifier and an infection type classifier.

[0018] The training process of the infection classifier is as follows:

[0019] 1) First, RNA-seq technology was used to analyze all samples to obtain the relative expression values ​​(FPKM) of 100 relevant characteristic genes. 2) Then, the Top Scoring Pairs (TSP) algorithm was used to calculate the expression of different gene pairs in infected and uninfected samples.

[0020] For each pair of genes i and j, the relative expression order in infected samples and uninfected samples is calculated. The specific method is:

[0021] Calculate the score Δ for each gene pair ij =|p ij (C1)-p ij (C2)|, that is, the degree of difference in expression order between infected samples and uninfected samples. The gene pair with a higher score has a greater role in judging whether it is infected. Pij (C m )=Prob(R i <R j |Y=C m ). Among them, P ij (C m ) indicates that in category C m The probability that the expression of gene i exceeds that of gene j; C1 represents uninfected samples, and C2 represents infected samples (including bacterial and viral infections); R i and R j are the relative expression order of the gene pair in the i-th and j-th samples, respectively, and i≠j; Y represents the category corresponding to a certain sample. Multiple gene pairs may obtain the same highest score. In order to avoid the situation where multiple gene pairs obtain the same score, the ranking difference of each gene pair in samples of different categories is further calculated, which is called the secondary score. Where n represents the category C m The sample type in |C m |Indicates category C m The total number of samples.

[0022] Set the secondary score γ ij As the basis for eliminating ties, the gene pairs are sorted according to the secondary scores, and the top k highest-scoring and independent gene pairs (i.e., each gene pair does not contain the same gene) are selected, and the weighted score is calculated based on the scores of the selected k gene pairs as the infection score, which is used as the basis for classification of infection or not. The k value is determined by cross-validation and is limited to no more than 10 and an odd number to ensure that there will be no ties in the final voting mechanism of the classifier.

[0023] 3) Finally, the selected k gene pairs are learned through the TSP algorithm to obtain a trained infection classifier.

[0024] The trained infection classifier is used to determine whether the new sample is infected. The specific method is to use the selected k gene pairs to determine the infection type of the new sample. new The infection score is calculated by weighted summing the performance in . When judging whether a gene pair meets a certain category C, if it meets the criteria, the positive ranking difference score Δ is increased. ij If not, reduce Δ ij . , an infection score greater than 0 indicates infection, and an infection score less than 0 indicates non-infection.

[0025] The specific calculation formula of the infection score Infect-Score is as follows:

[0026]

[0027] in

[0028]

[0029] Then, based on the infection classifier, an infection type classifier is further constructed to distinguish between bacterial infection and viral infection. The training process of the infection type classifier is similar to that of the infection classifier. The steps are as follows:

[0030] This process also relies on the TSP (Top Scoring Pairs) algorithm, but the difference is that it focuses on the differences in the expression order of gene pairs in bacterial infection samples and viral infection samples.

[0031] 1) For samples infected with bacteria and infected with viruses, the relative expression order of each pair of genes i and j in the two types of samples was calculated respectively.

[0032] 2) By comparing the expression order differences of the gene pairs in the two types of samples, the scores of the gene pairs are calculated. The gene pairs with higher scores have greater contributions in distinguishing bacterial infection from viral infection.

[0033] 3) To further ensure the accuracy of the infection type classifier, calculate the secondary score of each pair of genes, that is, the average ranking difference in bacterial infection and viral infection samples. Sort the gene pairs according to the secondary score, and select the top k independent and high-scoring gene pairs. Calculate the weighted score based on the scores of the selected k gene pairs as the bacterial-viral infection score, which is used as the basis for classifying the infection type. A bacterial-viral infection score BV-Score less than 0 indicates bacterial infection, and a bacterial-viral infection score greater than 0 indicates viral infection;

[0034] 4) The selected k gene pairs are learned through the TSP algorithm to obtain a trained infection type classifier.

[0035] The infection type classifier is used to determine the infection type of the sample to be tested. The specific method is: use the selected k gene pairs to determine the infection type of the new sample. The bacterial-viral infection score is calculated using the scores of the selected k gene pairs. A bacterial-viral infection score less than 0 is a bacterial infection, and a bacterial-viral infection score greater than 0 is a viral infection. In the present invention, an infection score (Infect-Score) is set for the infection classifier. When the Infect-Score is greater than 0, it is determined to have an infection risk, and when the Infect-Score is less than 0, it is determined to have no infection risk. A bacterial-viral infection score (BV-Score) is set for the infection type classifier. When the BV-Score is greater than 0, it is determined to have a viral infection risk, and when the BV-Score is less than 0, it is determined to have a bacterial infection risk.

[0036] The present invention also provides a memory for storing a computer program.

[0037] The present invention also provides a processor for executing a computer program, which can implement the above technical content.

[0038] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the content of the above technology can be implemented.

[0039] The present invention provides a host gene combination for diagnosing acute bacterial and viral infections. The present invention utilizes a weighted integrated machine learning method based on k-TSP and uses the data of gene expression levels of patient sample tissues to achieve the diagnosis of acute bacterial and viral infections. The present invention can calculate the expression data of multiple gene combinations obtained by different sequencing methods to obtain the infection risk and type. According to the method of the present invention, the high infection risk stratification results are predicted, the risk and type of acute bacterial and viral infection in patients can be diagnosed, and a basis for timely intervention and treatment of patients with clinical risks of acute bacterial and viral infection is provided.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] The present invention provides a detection gene combination for diagnosing acute bacterial and viral infections. According to the detection results of the multi-gene combination, the acute bacterial and viral infections and their risks and types are diagnosed. Combined with the patient's clinical information, a better understanding of the patient's clinical infection type and cancer progression is provided, which is conducive to accurate diagnosis and treatment. At the same time, detailed data is provided for the standardization of detection of patients at risk of acute bacterial and viral infections. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is the prediction result of the method of the present invention for clinical samples. DETAILED DESCRIPTION

[0043] The present invention provides a host gene combination for diagnosing acute bacterial and viral infections, the gene combination comprising

[0044] The following 100 related characteristic genes:

[0045] TDRD9, JAK1, EMR3, STAT5B, YPEL1, GALM, P2RY10, C17orf56, RNF125, SLC12A9, C3AR1, ITGA4, CBX7, PNPT1, RTN1, OAS1, RASSF5, MX1, PI3, IFIT1, GNA15, PTGS2, LFNG, ZBP1, RBL2, ARPC 3. PLEKHO1, PRDX5, PLEKHG3, RAB37, FCER1G, TBC1D7, GPR84, TIMM10, SULF2, PPM1M, KIAA0355, NDST2, FGL2, TAGAP, MAFG, ATOX1, ETF1, IFIT3, UPP1, LST1, ASCC2, PGLYRP1, LY6E, HK3, F AM89A, TP53I3, SIL1, IFI44L, RAB13, VPS37C, TPST2, IRF7, IFIH1, EMR1, CKAP4, C9orf89, G CC2, JUP, EXOSC4, PPM1F, GNG5, ACAA1, NDUFAF1, FXC1, LRPAP1, RPGRIP1, GPR160, ARL6IP5, S AMSN1, OAS2, PECAM1, TRIM5, S100A12, UGCG, NOV, GNA12, GADD45A, RAB5B, MTF1, BST2, DUSP6, IMPDH1, PARVG, ARPC4, OGFRL1, IFI27, RABAC1, FLOT1, CHCHD1, OASL, HP, HCK, S100P, PGD.

[0046] The device for diagnosing acute bacterial and viral infections of the present invention is used to determine the infection type of a sample, and the device for diagnosing acute bacterial and viral infections includes an RNA level gene expression data calculation module, an infection inference module, and an infection type inference module. The device is used to determine the infection type of the sample to be tested, and the specific method is as follows:

[0047] 1) Use the RNA level gene expression data calculation module to obtain the relative expression value FPKM of 100 relevant characteristic genes of the sample to be tested;

[0048] 2) Using the infection inference module and the relative expression value FPKM of the relevant characteristic genes to determine whether the sample is infected, if the inference result of the infection inference module is not infected, it is directly output; if the inference result of the infection inference module is infected, proceed to step 3) for further judgment;

[0049] 3) The infection type inference module and the relative expression value FPKM of the relevant characteristic genes are used to further judge the infected sample to obtain the infection type of the sample; the infection type includes viral infection and bacterial infection.

[0050] Wherein, in step 1), the method for obtaining the relative expression value FPKM of 100 relevant characteristic genes is:

[0051] First, a whole blood sample from the patient is extracted and preserved;

[0052] Then, the expression levels of 100 relevant characteristic genes at the RNA level were obtained by RNA-seq technology: total RNA was extracted from the patient's whole blood sample, sorted and library built, and then sequenced. The sequencing sequence was quality controlled and trimmed, and the obtained sequence was compared with the human reference genome (GRCh38). According to the comparison results, the expression level of each relevant characteristic gene per kilobase per million fragments was obtained through transcript splicing and quantitative software (such as Salmon, HISAT2, etc.), which is the relative expression value FPKM of the 100 relevant characteristic genes.

[0053] To verify the effectiveness of the method of the present invention, the present application uses an RNA-seq dataset from blood samples in a clinical validation cohort to evaluate its performance. The prediction accuracy of the above method for infection types is verified. The clinical diagnosis of the samples in the dataset was 59 cases of bacterial infection, 43 cases of viral infection, and 23 cases of non-infectious inflammation.

[0054] The method of the present invention is used to determine the infection type of the sample, and the infection type of the sample is obtained based on the Infect-Score and BV-Score of the sample (such as Figure 1 ). In the above data set, the classification accuracy of the non-infected samples determined by the present invention reached 100%, indicating that the present invention can perfectly identify non-infected samples. The classification accuracy of bacterial infection samples was 97.8%, demonstrating the high reliability of the present invention in identifying bacterial infections. The classification accuracy of virus infection samples was 88.5%, also achieving a relatively high classification accuracy rate.

[0055] The method of the present invention has good accuracy and can provide a basis for timely intervention and treatment for patients with clinical risks of acute bacterial and viral infections.

Claims

1. A host gene combination for diagnosing acute bacterial and viral infections, characterized in that: The gene combination includes the following 100 related characteristic genes: TDRD9, JAK1, EMR3, STAT5B, YPEL1, GALM, P2RY10, C17orf56, RNF125, SLC12A9, C3AR1, ITGA4, CBX7, PNPT1, RTN1, OAS1, RASSF5, MX1, PI3, IFIT1, GNA15, PTGS2, LFNG, ZBP1, RBL2, ARPC 3. PLEKHO1, PRDX5, PLEKHG3, RAB37, FCER1G, TBC1D7, GPR84, TIMM10, SULF2, PPM1M, KIAA0355, NDST2, FGL2, TAGAP, MAFG, ATOX1, ETF1, IFIT3, UPP1, LST1, ASCC2, PGLYRP1, LY6E, HK3, F AM89A, TP53I3, SIL1, IFI44L, RAB13, VPS37C, TPST2, IRF7, IFIH1, EMR1, CKAP4, C9orf89, G CC2, JUP, EXOSC4, PPM1F, GNG5, ACAA1, NDUFAF1, FXC1, LRPAP1, RPGRIP1, GPR160, ARL6IP5, S AMSN1, OAS2, PECAM1, TRIM5, S100A12, UGCG, NOV, GNA12, GADD45A, RAB5B, MTF1, BST2, DUSP6, IMPDH1, PARVG, ARPC4, OGFRL1, IFI27, RABAC1, FLOT1, CHCHD1, OASL, HP, HCK, S100P, PGD.

2. A device for diagnosing acute bacterial and viral infections, characterized in that The device includes an RNA level gene expression data calculation module, an infection inference module and an infection type inference module; the RNA level gene expression data calculation module is used to obtain the relative expression value FPKM of the 100 relevant characteristic genes as described in claim 1; the infection inference module is used to judge whether the sample is infected according to the relative expression value FPKM of the relevant characteristic genes, and if the inference result of the infection inference module is not infected, it is directly output; if the inference result of the infection inference module is infection, the infection type inference module and the relative expression value FPKM of the relevant characteristic genes are further used to further judge the infected sample to obtain the infection type of the sample; the infection type includes viral infection and bacterial infection.

3. A device for diagnosing acute bacterial and viral infections according to claim 2, characterized in that: The method for obtaining the relative expression value FPKM of the relevant characteristic gene is as follows: total RNA is extracted from the sample, and after sorting and library construction, sequencing is performed; then the sequencing sequence is quality controlled and trimmed, and the obtained sequence is compared with the human reference genome GRCh38. According to the comparison result, the expression level of each relevant characteristic gene per kilobase per million fragments is obtained by transcript splicing and quantitative software, which is the relative expression value FPKM of the relevant characteristic gene.

4. A device for diagnosing acute bacterial and viral infections according to claim 2, characterized in that: The infection inference module is used to determine whether a sample is infected according to the expression order difference of different gene pairs, and is specifically implemented based on an infection classifier. The infection classifier takes the relative expression value FPKM of the relevant characteristic gene of the sample as input and outputs whether it is infected, wherein the sample is divided into an uninfected sample and an infected sample; the training process of the infection classifier is specifically as follows: 1) Perform RNA-seq technical analysis on all samples to obtain the relative expression value FPKM of 100 relevant characteristic genes; 2) Using the TSP algorithm, calculate the expression order of different gene pairs in infected samples and uninfected samples; the specific method is: For each pair of genes i and j, the relative expression order in infected and uninfected samples was calculated; Calculate the score Δ for each gene pair ij =|p ij (C1)-p ij (C2)|, used to determine the degree of difference in the expression order of the corresponding gene pairs in infected and uninfected samples; Among them, P ij (C m ) indicates that in category C m The probability that gene i is expressed more than gene j in ij (C m )=Prob(R i <R j |Y=C m ); C1 represents uninfected samples, C2 represents infected samples; R i and R j are the relative expression order of the gene pair in the i-th and j-th samples, respectively, and i≠j; Y represents the category corresponding to a certain sample; In order to avoid the situation where multiple gene pairs obtain the same score, the ranking difference of each pair of genes in samples of different categories is further calculated, that is, the secondary score Where n represents the category C m The sample type in |C m |Indicates category C m The total number of samples; Set the secondary score γ ij As the basis for eliminating ties, the gene pairs are sorted according to the secondary scores, and the top k independent gene pairs with the highest scores are selected; and the weighted scores are calculated based on the scores of the selected k gene pairs as the infection score, which is used as the basis for classification of infection or not. The infection score greater than 0 indicates infection, and the infection score less than 0 indicates non-infection; 3) The selected k gene pairs are learned through the TSP algorithm to obtain a trained infection classifier.

5. A device for diagnosing acute bacterial and viral infections according to claim 4, characterized in that: The infection type inference module determines the infection type of the sample based on an infection type classifier. The infection type classifier takes the relative expression value FPKM of the relevant characteristic gene of the sample as input and the infection type as output, wherein the sample is divided into bacterial infection samples and viral infection samples; the training process of the infection type classifier is similar to the training process of the infection inference module.

6. A method for diagnosing acute bacterial and viral infections, characterized in that The specific method is: 1) Determine the RNA level gene expression data of the sample to obtain the relative expression value FPKM of the 100 relevant characteristic genes as claimed in claim 1; 2) Determine whether the sample is infected based on the relative expression value FPKM of each relevant characteristic gene; if the judgment result is no, the diagnosis ends; if the judgment result is yes, proceed to step 3); 3) The relative expression value FPKM of each relevant characteristic gene is used to further judge the infected sample and obtain the infection type of the sample.

7. An electronic device, characterized in that: include: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

8. A computer-readable storage medium having computer instructions stored thereon, characterized in that: The computer instructions are used to enable a computer to execute the steps of the method according to any one of claims 1 to 6.

Images