Marker, prediction method and system for predicting liver cancer treatment effect of oncolytic virus VG161

Abstract

The invention discloses a marker, a prediction method and a system for predicting the liver cancer treatment effect of oncolytic virus VG161. The marker comprises an expression product of at least one of SEZ6, AKR1C1, LILRA5, NCKAP5 and SETD9 genes or at least one of the SEZ6, AKR1C1, LILRA5, NCKAP5 and SETD9 genes; the prediction method comprises the following steps: quantitatively detecting the RNA expression quantity of the SEZ6 gene, the AKR1C1 gene, the LILRA5 gene, the NCKAP5 gene and the SETD9 gene in the in-vitro biological tissue; inputting the RNA expression quantity of each gene into the prediction model, and judging the obtained prediction value according to a prediction threshold value to obtain a prediction result; the prediction system comprises a detection module, a data processing module and a prediction module. According to the prediction method disclosed by the invention, the liver cancer treatment effect of the oncolytic virus VG161 is predicted by utilizing the marker, the result is stable and has correlation, and the prediction method is widely applied to the fields of curative effect prediction, prognosis evaluation and the like of the liver cancer treatment effect of the oncolytic virus VG161.

Description

Technical Field

[0001] The present invention relates to the fields of medical data analysis and artificial intelligence, and in particular to a marker, prediction method and system for predicting the effect of oncolytic virus VG161 in treating liver cancer. Background Art

[0002] Oncolytic virus (OV) is a virus that can specifically infect and kill tumor cells. It has become an emerging therapy for the treatment of malignant tumors such as liver cancer by directly dissolving cancer cells or stimulating the host immune system to fight tumors. Although oncolytic viruses have shown good tumor cell killing ability and effective inhibition of tumor growth in multiple cancer types, due to factors such as tumor heterogeneity and differences in gene expression between individuals, not all subjects can benefit equally from immunotherapy. Therefore, it is particularly important to predict the response and sensitivity of liver cancer patients to oncolytic viruses in a targeted manner. At present, there is still a lack of effective prediction models for the effect of oncolytic virus VG161 in the treatment of liver cancer.

[0003] Existing tumor immunotherapy prediction models, such as TIDE (Tumor Immune Dysfunction and Exclusion), are used to predict the response of subjects to immune checkpoint inhibitors (such as PD-1 / PD-L1, CTLA-4), which is based on immune dysfunction and immune exclusion; Immunoscore evaluates prognosis and immunotherapy response by analyzing T cell infiltration in the tumor microenvironment; TMB (Tumor Mutational Burden) predicts the potential response of patients to immune checkpoint inhibitors based on the mutation load in the tumor, etc., although they can help doctors screen out patients who are more likely to benefit from immunotherapy in the clinic. However, the current liver cancer prediction model is mainly based on deep learning algorithms, which leads to the training model being heavily dependent on sample size and complete clinical data types, such as imaging, genetic data, and pathological research. Different hospitals have different diagnostic standards and data processing methods, sample imbalance, low availability of high-quality multi-dimensional data, and data missing, which profoundly affect the accuracy of the model. Secondly, although there are complex deep learning algorithms for predicting the effect of specific liver cancer treatment, these algorithms have certain predictions, but lack interpretability due to the difficulty in establishing a connection with actual clinical results. This makes it difficult for users to understand the model's prediction basis, limiting its application in clinical decision-making. Due to the uninterpretability of the model, once the model's prediction results are viewed with skepticism in clinical practice, it will affect its practical application. Many liver cancer prediction models lack verification on independent data sets, resulting in unknown effects in their actual application. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a marker, prediction method and system for predicting the effect of oncolytic virus VG161 in treating liver cancer.

[0005] The technical solution adopted by the present invention is:

[0006] 1. A marker for predicting the efficacy of oncolytic virus VG161 in treating liver cancer

[0007] The marker includes at least one of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene and SETD9 gene, or includes the expression product of at least one of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene and SETD9 gene.

[0008] Furthermore, the markers include at least the NCKAP5 gene and the SETD9 gene, or at least the expression products of the NCKAP5 gene and the SETD9 gene.

[0009] 2. Application of a marker in the preparation of a drug for predicting the efficacy of oncolytic virus VG161 in treating liver cancer.

[0010] 3. A method for predicting the efficacy of oncolytic virus VG161 in treating liver cancer

[0011] The prediction method comprises the following steps:

[0012] 1) Quantitatively detect the RNA expression levels of SEZ6, AKR1C1, LILRA5, NCKAP5 and SETD9 genes in ex vivo biological tissues;

[0013] In the step 1), a sequencing instrument and / or a drug are used to quantitatively detect the RNA expression level of each gene in the ex vivo biological tissue. The ex vivo biological tissue is specifically an ex vivo biological liver tissue.

[0014] 2) Inputting the RNA expression amount of each gene detected in step 1) into the prediction model to obtain a predicted value, judging the predicted value according to the prediction threshold, and predicting the effect of oncolytic virus VG161 on the treatment of liver cancer according to the judgment result to obtain a prediction result. The prediction result is divided into low benefit and high benefit.

[0015] In the step 2), the predicted value is judged according to the prediction threshold, and the effect of oncolytic virus VG161 on the treatment of liver cancer is predicted according to the judgment result. The process of obtaining the prediction result is specifically as follows: comparing the size of the prediction threshold and the prediction value. When the prediction value is greater than the prediction threshold, the prediction result is low benefit, and when the prediction value is less than the prediction threshold, the prediction result is high benefit.

[0016] The method for constructing the prediction model in step 2) comprises the following steps:

[0017] D1) Use single-cell RNA-seq sequencing technology to process each in vitro biological tissue sample, obtain the UMI count matrix of each in vitro biological tissue sample, obtain the survival information corresponding to all in vitro biological tissue samples and calculate the median survival time, and divide the UMI count matrix of the in vitro biological tissue samples into a low-benefit group and a high-benefit group according to the median survival time, thereby constructing a sample data set;

[0018] D2) Using data processing software to perform gene ontology analysis and KEGG pathway analysis on the sample data set, multiple significantly differentially expressed genes were obtained, and univariate Cox regression method was used to process the significantly differentially expressed genes to obtain multiple potential prognostic genes;

[0019] D3) Using the LASSOCox regression analysis method, at least one most significantly differentially expressed gene is screened out from the potential prognostic genes as a marker, and the prediction model is constructed based on the marker.

[0020] The prediction model in step 2) is specifically:

[0021] R=y (SEZ6) +y (AKR1C1) +y (LILRA5) +y (NCKAP5) +y (SETD9)

[0022] In the formula, R is the predicted value, y (SEZ6) ,y (AKR1C1) ,y (LILRA5) ,y (NCKAP5) ,y (SETD9) They are the RNA expression indexes of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene, and SETD9 gene;

[0023] Among them, the RNA expression index of each gene is:

[0024] y (SEZ6) =-2.2222×FPKM (SEZ6) +100

[0025] y (AKR1C1) =0.0418×FPKM (AKR1C1) +2×10 -15

[0026] y (LILRA5) =2.4875×FPKM (LILRA5) +7×10 -15

[0027] y(NCKAP5) =1.543×FPKM (NCKAP5) -2×10 -7

[0028] y (SETD9) =0.7337×FPKM (SETD9) -2×10 -7

[0029] In the formula, FPKM (SEZ6) 、FPKM (AKR1C1) 、FPKM (LILRA5) 、FPKM (NCKAP5) 、FPKM (SETD9) They are the RNA expression levels of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene, and SETD9 gene, respectively.

[0030] The prediction threshold is 101.812.

[0031] IV. A prediction system for the efficacy of oncolytic virus VG161 in treating liver cancer

[0032] The prediction system comprises:

[0033] A detection module, used for detecting RNA expression levels of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene and SETD9 gene in ex vivo biological tissues;

[0034] The data processing module is used to calculate the RNA expression level of each gene using the prediction model stored in the prediction module to obtain a predicted value;

[0035] The prediction module is used to judge the predicted value according to the prediction threshold, and predict the effect of oncolytic virus VG161 in treating liver cancer according to the judgment result to obtain the prediction result.

[0036] Optionally, the detection module in the prediction system can be replaced by a visualization module, and the visualization module is used to input the RNA expression levels of the SEZ6 gene, the AKR1C1 gene, the LILRA5 gene, the NCKAP5 gene and the SETD9 gene into the prediction system, and visualize the prediction values ​​and prediction results;

[0037] Optionally, the prediction system further includes a visualization module, and the visualization module is used to visualize the prediction value and the prediction result.

[0038] The beneficial effects of the present invention are:

[0039] 1. The prediction method of the present invention establishes a connection between the microscopic genetic changes in in vitro biological tissues and the macroscopic effects of oncolytic virus VG161 in treating liver cancer through a prediction model. The genetic data of five key genes are used to predict the effects of oncolytic virus VG161 in treating liver cancer. The results are interpretable, statistically robust and biologically relevant.

[0040] 2. The prediction method and prediction system of the present invention can be widely used in the fields of efficacy prediction of oncolytic virus VG161 in treating liver cancer, targeted drug screening, and prognosis evaluation.

[0041] 3. The construction of the prediction model in the present invention only requires the single-cell RNA-seq sequencing results of in vitro biological tissues, without relying on complete clinical data. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 Schematic diagram of the construction process of a sample data set in an embodiment of the present invention;

[0043] Figure 2 Schematic diagram of the method for constructing a prediction model for the effect of oncolytic virus treatment of liver cancer in the present invention; wherein, (A) is a schematic diagram of the process of single-cell RNA-seq sequencing; (B) is a KM (Kaplan-Meier) survival curve between the low-benefit group and the high-benefit group; (C) is a volcano plot of differentially expressed genes between the low-benefit group and the high-benefit group; (D) is a result diagram of GO enrichment and KEGG pathway enrichment analysis; (E) is a result diagram of univariate Cox regression analysis; (F) is a schematic diagram of LASSOCox regression analysis for evaluating prognosis and screening risk genes Figure 1 ; (G) Schematic diagram of LASSOCox regression analysis for evaluating prognosis and screening risk genes Figure 2 ; (H) is the result of multivariate Cox regression analysis; (I) is the predicted KM survival curve obtained by the prediction model; (J) is the ROC curve obtained by the prediction model; (K) is the KM survival curve obtained by the prediction model based on the TCGA-LIHC cohort. DETAILED DESCRIPTION

[0044] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] The present invention uses the single-cell RNA-seq sequencing information of in vitro biological tissue to construct a sample data set, and then uses a machine learning method to analyze the sample data set, and obtains five key genes (SEZ6, AKR1C1, LILRA5, NCKAP5, SETD9) as markers for predicting the effect of oncolytic virus VG161 in treating liver cancer, and establishes a prediction system and prediction method based on the marker. The prediction method of the present invention establishes a connection between the genetic changes at the microscopic level and the oncolytic virus VG161 at the macroscopic level for treating liver cancer, and uses the genetic data of five key genes to predict the effect of oncolytic virus VG161 for treating liver cancer, and the results are statistically robust and biologically relevant. The prediction system and prediction method of the present invention can be widely used in clinical screening, efficacy prediction, targeted drug screening, and prognosis evaluation of oncolytic virus VG161.

[0046] The specific embodiments of the present invention are as follows:

[0047] In the following examples, all reagents used are commercial reagents unless otherwise specified.

[0048] Example 1

[0049] This embodiment shows a method for constructing a prediction model for the effect of oncolytic virus treatment of liver cancer based on a machine learning algorithm. In this embodiment, multimodal data is used to train a machine learning model to predict the effect of oncolytic virus VG161 in treating liver cancer. In this embodiment, sequencing and bioinformatics analysis are provided by Shanghai Ouyi Biomedical Co., Ltd. Specifically comprising the following steps:

[0050] D1) Use single-cell RNA-seq sequencing technology to process each ex vivo biological tissue sample, obtain the UMI count matrix of each ex vivo biological tissue sample, obtain the survival information in the corresponding clinical information of all ex vivo biological tissue samples, calculate the median survival time, and divide the UMI count matrix of the ex vivo biological tissue samples into a low-benefit group and a high-benefit group according to the median survival time, and then construct a sample data set. Specifically: ex vivo biological tissue samples with corresponding survival information lower than the median survival time belong to the low-benefit group, and ex vivo biological tissue samples with corresponding survival information higher than the median survival time belong to the high-benefit group.

[0051] like Figure 1 As shown in Figure 2, in this example, clinical data of 44 patients with advanced primary liver cancer who were resistant to standard therapy and received VG161 oncolytic virus treatment were collected. The clinical data included age, gender, tumor stage, AFP level, liver function indicators (ALT, AST, etc.), and previous treatment history. Virus characteristic data collection includes injection dose, injection route, and virus distribution in target tissues and non-target tissues after treatment. During the treatment, patients received 1.0×10 8Up to 5.0×10 8 VG161 was injected intratumorally at a PFU dose. Single-cell RNA-seq sequencing data of ex vivo liver cancer tissues were obtained by RNA-seq sequencing technology, namely the UMI count matrix, focusing on genes related to oncolytic virus response.

[0052] like Figure 2 As shown in (B), in this embodiment, the UMI count matrix of the ex vivo liver cancer tissue is divided into a low benefit group and a high benefit group according to the median survival (median OS).

[0053] D2) Gene ontology (GO) analysis and KEGG pathway analysis were performed on the UMI count matrix using data processing software, and differentially expressed cellular pathways were obtained, including significantly enriched signaling pathways such as motor proteins, protein digestion, absorption, and calcium, as well as significantly differentially expressed genes including 49 significantly upregulated and 745 significantly downregulated genes. Finally, the univariate Cox regression method was used to process the 794 significantly differentially expressed genes obtained above, and 299 potential prognostic genes were obtained.

[0054] The specific implementation method of this step is:

[0055] D2.1) Figure 2 As shown in (A), ex vivo liver cancer tissue samples were dissociated into single cells. After resuspension in 1× phosphate-buffered saline, all single cell suspensions were subjected to single-cell sequencing (10× Genomics) with a capture target of 7000 cells, and all libraries were sequenced on Illumina NovaSeq 6000 using paired-end 150bp;

[0056] D2.2) The FASTQ files were processed using 10× Genomics’ Cell Ranger software (version 7.0.1) and aligned to the GRCh38 human reference genome to obtain a UMI count matrix.

[0057] D2.3) The UMI count matrix was analyzed using the Seurat47 (version 4.0.0) R package to obtain significantly differentially expressed genes (DEGs). In this embodiment, the significantly differentially expressed genes obtained included 49 significantly upregulated and 745 significantly downregulated genes.

[0058] The specific process is as follows:

[0059] c1) To remove low quality cells and possible multiple capture, filter the cells by the following conditions:

[0060] ① Number of genes (number of genes < 200);

[0061] ②UMI (UMI<1000);

[0062] ③log10GenesPerUMI(log10GenesPerUMI<0.7);

[0063] ④ Mitochondrial RNA UMI percentage (the proportion of UMIs mapped to mitochondrial genes>10%);

[0064] ⑤ Hemoglobin RNA UMI percentage (the proportion of UMIs mapped to the hemoglobin gene > 5%).

[0065] c2) Identify potential doublets using the DoubletFinder48 package (version 2.0.3):

[0066] First, to obtain normalized gene expression data, the library size was normalized using the NormalizeData function. Specifically, the global scaling normalization method "LogNormalize" normalizes the gene expression measurements of each cell to the total expression, multiplies by a scaling factor (default 10000), and then logarithmically transforms the results.

[0067] Subsequently, the top 2000 highly variable genes (HVGs) were calculated using the Seurat function FindVariableGenes(mean.function=FastExpMean, dispersion.function=FastLogVMR). Principal component analysis (PCA) was performed using the RunPCA function to reduce dimensionality.

[0068] Use the FindClusters function to perform graph-based clustering of cells based on their gene expression profiles.

[0069] The cells were visualized using the 2D Uniform Manifold Approximation (UMAP) algorithm and the RunUMAP function.

[0070] Use the FindAllMarkers function (test.use=presto) to identify the marker genes for each cluster. Use the FindMarkers function (test.use=presto) to select differentially expressed genes (DEGs), such as Figure 2 As shown in (C).

[0071] Based on the hypergeometric distribution method, P value < 0.05 and the absolute value of logarithmic fold change |log2foldchange|> 0.58 were set as the thresholds for significant differential expression. GO enrichment and KEGG pathway enrichment analysis were performed on the differentially expressed genes using R (version 4.0.3), and the GO enrichment and KEGG pathway enrichment analysis results were obtained (such as Figure 2(D) By analyzing the results of GO enrichment analysis and KEGG pathway enrichment analysis, significantly differentially expressed genes (DEGs) including 49 significantly up-regulated and 745 significantly down-regulated genes were obtained.

[0072] From the significantly differentially expressed genes, 299 potential prognostic genes were identified using univariate Cox regression analysis (e.g. Figure 2 (as shown in (E)).

[0073] D3) Figure 2 (F) and Figure 2 As shown in (G), the LASSOCox regression model was used to screen out the five most significantly differentially expressed genes from potential prognostic genes as markers, and a prediction model was constructed based on all markers.

[0074] In this embodiment, the markers include SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene and SETD9 gene.

[0075] The prediction model obtained in this embodiment is specifically:

[0076] R=y (SEZ6) +y (AKR1C1) +y (LILRA5) +y (NCKAP5) +y (SETD9)

[0077] In the formula, R is the predicted value, y (SEZ6) ,y (AKR1C1) ,y (LILRA5) ,y (NCKAP5) ,y (SETD9) They are the RNA expression indexes of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene, and SETD9 gene;

[0078] Among them, the RNA expression index of each gene is:

[0079] y (SEZ6) =-2.2222×FPKM (SEZ6) +100

[0080] y (AKR1C1) =0.0418×FPKM (AKR1C1) +2×10 -15

[0081] y (LILRA5) =2.4875×FPKM (LILRA5) +7×10 -15

[0082] y (NCKAP5)=1.543×FPKM (NCKAP5) -2×10 -7

[0083] y (SETD9) =0.7337×FPKM (SETD9) -2×10 -7

[0084] In the formula, FPKM (SEZ6) 、FPKM (AKR1C1) 、FPKM (LILRA5) 、FPKM (NCKAP5) 、FPKM (SETD9) They are the RNA expression levels of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene and SETD9 gene respectively, and the unit of RNA expression level is fragments per kilobase of transcript per million mapped reads (FPKM).

[0085] In the prediction model obtained in this embodiment, the prediction value is negatively correlated with the beneficial effect.

[0086] D4) Use the fitted linear model to process the prediction model, obtain the relationship between the risk value and the prediction value, set the risk threshold and input it into the relationship to obtain the prediction threshold.

[0087] The relationship between the risk value and the predicted value obtained in this embodiment is: F = 0.0543 × R-5.0284, where F is the risk value and R is the predicted value. The risk threshold is F = 0.5, and the predicted threshold is 101.812. When the predicted value is greater than the predicted threshold 101.812, the predicted result is low benefit, and when the predicted value is less than the predicted threshold 101.812, the predicted result is high benefit.

[0088] Example 2

[0089] This example uses the multivariate Cox regression analysis method, Kaplan-Meier (KM) curve and receiver operating characteristic curve (ROC) to test the performance of the prediction model obtained in Example 1.

[0090] Comparative Example 1

[0091] In this comparative example, the actual benefit effect is evaluated by combining the previously acquired immune data and the revised solid tumor clinical efficacy evaluation criteria (mRECIST) to support the accuracy of the correlation of this prediction model. The immune data is obtained by immunohistochemistry or flow cytometry to detect in vitro biological blood samples or in vitro biological tissue samples to obtain the immune cell infiltration in the blood and tumor microenvironment (such as the distribution and density of CD4+, CD8+T cells, NK cells, Tregs and other immune cells) and the concentration of key immune-related markers related to liver cancer prognosis (such as CD69, PD-1, PD-L1, TNF-α, IFN-γ, etc.).

[0092] In Example 2 of the present invention, the prediction model obtained in Example 1 was tested using a multivariate Cox regression analysis method. The results are shown in Figure 2 (H). It can be seen that NCKAP5 and SETD9 are important independent predictors.

[0093] In Example 2 of the present invention, the prediction model obtained in Example 1 was used to predict the Kaplan-Meier (KM) curve. Figure 2 (I).

[0094] In Example 2 of the present invention, the receiver operating characteristic (ROC) curve was predicted using the prediction model obtained in Example 1. Figure 2 (J). The Kaplan-Meier (KM) curve and the receiver operating characteristic curve show that the prediction model can accurately predict the survival rates of 6, 9, 12, 15, 18 and 21 months.

[0095] In Example 2 of the present invention, the survival data of 418 advanced HCC patients who did not receive such treatment in Cancer Genome Atlas (TCGA) were input into the prediction model obtained in Example 1. The prediction results are shown in Figure 2 It can be seen that there is no significant difference in the survival results between the high-income group and the low-income group obtained by the model.

[0096] In summary, the method of the present invention introduces new markers and an innovative prediction system, which uses the genetic data of five key genes (SEZ6, AKR1C1, LILRA5, NCKAP5, SETD9) to predict the therapeutic effect of oncolytic virus VG161 on liver cancer. The results are statistically robust and biologically relevant.

Claims

1. A marker for predicting the effect of oncolytic virus VG161 in treating liver cancer, characterized in that: The marker includes at least one of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene and SETD9 gene, or includes the expression product of at least one of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene and SETD9 gene.

2. The marker for predicting the effect of oncolytic virus VG161 in treating liver cancer according to claim 1, characterized in that: The markers include at least the NCKAP5 gene and the SETD9 gene, or at least the expression products of the NCKAP5 gene and the SETD9 gene.

3. Use of a marker as described in any one of claims 1 or 2 in the preparation of a drug for predicting the effect of oncolytic virus VG161 in treating liver cancer.

4. A method for predicting the effect of oncolytic virus VG161 in treating liver cancer, characterized in that: The prediction method comprises the following steps: 1) Quantitatively detect the RNA expression levels of SEZ6, AKR1C1, LILRA5, NCKAP5 and SETD9 genes in ex vivo biological tissues; 2) Inputting the RNA expression level of each gene detected in step 1) into the prediction model to obtain a predicted value, judging the predicted value according to the prediction threshold, predicting the effect of oncolytic virus VG161 in treating liver cancer according to the judgment result, and obtaining a predicted result, wherein the predicted result is divided into low benefit and high benefit.

5. The prediction method according to claim 4, characterized in that: In the step 2), the process of obtaining the prediction result according to the prediction threshold is specifically as follows: comparing the prediction threshold and the prediction value. When the prediction value is greater than the prediction threshold, the prediction result is a low benefit; when the prediction value is less than the prediction threshold, the prediction result is a high benefit.

6. The prediction method according to claim 4, characterized in that: The method for constructing the prediction model comprises the following steps: D1) Using single-cell sequencing technology to process each ex vivo biological tissue sample, obtain the UMI count matrix of each ex vivo biological tissue sample, obtain the survival information corresponding to all ex vivo biological tissue samples and calculate the median survival time, and divide the UMI count matrix of the ex vivo biological tissue samples into a low-benefit group and a high-benefit group according to the median survival time, thereby constructing a sample data set; D2) Using data processing software to perform gene ontology analysis and KEGG pathway analysis on the sample data set, multiple significantly differentially expressed genes were obtained, and univariate Cox regression method was used to process the significantly differentially expressed genes to obtain multiple potential prognostic genes; D3) Using the LASSOCox regression analysis method, at least one most significantly differentially expressed gene is screened out from the potential prognostic genes as a marker, and the prediction model is constructed based on the marker.

7. The prediction method according to claim 4, characterized in that: The prediction model in step 2) is specifically: R=y (SEZ6) +y (AKR1C1) +y (LILRA5) +y (NCKAP5) +y (SETD9) In the formula, R is the predicted value, y (SEZ6) ,y (AKR1C1) ,y (LILRA5) ,y (NCKAP5) ,y (SETD9) They are the RNA expression indexes of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene, and SETD9 gene; Among them, the RNA expression index of each gene is: and (SEZ6) -2.2222×FPKM (SEZ6) +100 and (AKR1C1) =0.0418×FPKM (AKR1C1) +2×10 -15 and (LILRA5) =2.4875×FPKM (LILRA5) +7×10 -15 and (NCKAP5) =1.543×FPKM (NCKAP5) -2×10 -7 and (SETD9) =0.7337×FPKM (SETD9) -2×10 -7 In the formula, FPKM (SEZ6) 、FPKM (AKR1C1) 、FPKM (LILRA5) 、FPKM (NCKAP5) 、FPKM (SETD9) They are the RNA expression levels of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene, and SETD9 gene, respectively.

8. The prediction method according to claim 7, characterized in that: The prediction threshold is 101.

812.

9. A prediction system for the effect of oncolytic virus VG161 in treating liver cancer, characterized by: The prediction system comprises: A detection module, used for detecting RNA expression levels of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene and SETD9 gene in ex vivo biological tissues; The data processing module is used to calculate the RNA expression level of each gene using the prediction model stored in the prediction module to obtain a predicted value; The prediction module is used to judge the predicted value according to the prediction threshold, and predict the effect of oncolytic virus VG161 in treating liver cancer according to the judgment result to obtain the prediction result.

10. The prediction system according to claim 7, characterized in that: The detection module is replaced by a visualization module, wherein the visualization module is used to input RNA expression levels of SEZ6 gene, AKR1C1 gene, LILRA5 gene, NCKAP5 gene and SETD9 gene into the prediction system, and to visualize the prediction values ​​and prediction results; Alternatively, the prediction system further includes a visualization module, and the visualization module is used to visualize the prediction value and the prediction result.

Images