A tumor-targeted drug recommendation method and system

Abstract

The present application relates to the field of drug recommendation, in particular to a tumor-targeting drug recommendation method and system, the method comprising the following steps: providing gene expression information of a patient; obtaining a cancer cell ratio by using the gene expression information; determining a target gene by using the cancer cell ratio; providing a database comprising targeting drug information; generating a correlation model of targeting drugs and proto-oncogenes by using the database; optimizing the correlation model; obtaining targeting drug recommendation information by using the optimized correlation model in combination with the target gene; and integrating the targeting drug recommendation information to obtain a targeting drug recommendation table. The method takes the gene expression information of a patient as a basis, obtains a cancer cell ratio by calculation, and then determines a target gene, and further precisely matches the target gene with a current targeting drug, so that the most suitable tumor-targeting drug for the individual condition of a patient can be precisely determined and recommended in the case where a cancerous tissue biopsy sample cannot be obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of drug recommendation, and in particular to a tumor-targeted drug recommendation method and system. BACKGROUND

[0002] Cancer is one of the most serious diseases threatening human health at present, and numerous medical workers are trying their best to conquer this problem. At present, the main treatment methods for tumors in clinical practice are surgery, radiotherapy, chemotherapy, etc. However, due to the complexity of the pathogenesis of tumors and the large individual difference factors, the treatment effect varies from person to person, and the overall effect is not optimistic. In recent years, with the progress of "targeted drug" technology, more and more patients have begun to benefit from it, not only effectively prolonging the overall survival period, but also significantly improving the quality of life. In recent years, with the concept of "precision medicine" being advocated by more and more people, the treatment method of using targeted drugs guided by gene detection technology is being accepted and used by more and more clinical doctors, and is bringing more suitable treatment methods and better treatment effects to patients.

[0003] Renal pelvis carcinoma (RPC) refers to a malignant tumor occurring in the renal pelvis. It has a relatively low incidence in the urinary system tumors, accounting for about 10% of all urothelial tumors. Due to the deep location of the renal pelvis, it is difficult to perform biopsy before surgery, so CT urography (CTU) has become the most important basis for preoperative diagnosis of renal pelvis carcinoma patients by clinical doctors. For patients diagnosed with renal pelvis carcinoma before surgery, radical nephroureterectomy (RNU) will be performed and postoperative adjuvant chemotherapy will be given. On the other hand, many doctors currently advocate the use of neoadjuvant chemotherapy for patients who are physically able, i.e. chemotherapy treatment before surgery, and then decide to maintain treatment or proceed to surgery according to the specific situation. In general, whether it is surgery or chemotherapy, it will cause great trauma or non-specific damage to the patient's body, and due to the deep location of the renal pelvis, it is also impossible to take out tumor samples for biopsy. Therefore, how to accurately determine the tumor-targeted drug suitable for the individualized condition of the patient without obtaining tumor biopsy samples is a difficult problem in the treatment of renal pelvis carcinoma. SUMMARY

[0004] In view of the problems in the prior art, in a first aspect, the present application provides a tumor-targeting drug recommendation method, comprising the following steps: providing gene expression information of a patient, the gene expression information comprising gene expression data and gene expression categories, the gene expression data comprising different proto-oncogenes and expression values thereof; obtaining a cancer cell ratio by using the gene expression information; determining a target gene by using the cancer cell ratio; providing a database, the database comprising targeting drug information; generating a correlation model of targeting drugs and proto-oncogenes by using the database; optimizing the correlation model; obtaining targeting drug recommendation information by using the optimized correlation model in combination with the target gene, the targeting drug recommendation information comprising recommendation strength of the targeting drugs; and integrating the targeting drug recommendation information to obtain a targeting drug recommendation table. This method takes the gene expression information of a renal pelvis cancer patient as a basis, obtains a cancer cell ratio by calculation, and then determines a target gene, and through accurate matching of the target gene and current targeting drugs, can accurately determine and recommend a tumor-targeting drug suitable for individualized conditions of a renal pelvis cancer patient in the case where cancerous tissue biopsy samples cannot be obtained. Meanwhile, this method is not only suitable for non-invasive examination of renal pelvis cancer, but also can provide a reference and basis for exploration of preoperative neoadjuvant therapy for other cancer types that cannot or are difficult to biopsy.

[0005] Optionally, the obtaining of the cancer cell ratio by using the gene expression information comprises the following steps: randomly selecting N data in the gene expression data of different proto-oncogenes as initial center points; comparing the gene expression data with the initial center points; according to the comparison result, combining the gene expression categories, dividing the gene expression data into an overexpression group and a normal expression group; respectively using the gene expression data in the overexpression group and the normal expression group, redefining the initial center by calculating the mean value; using the redefined initial center, classifying the gene expression data again in combination with the gene expression data to obtain an overexpression group N1 and a normal expression group N2 of the proto-oncogene that meet the error threshold value; and obtaining a cancer cell ratio caused by the proto-oncogene by using the overexpression group N1 and the normal expression group N2. This method screens the overexpression group N1 and the normal expression group N2 of different proto-oncogenes under the same termination condition, and improves the calculation efficiency.

[0006] Optionally, the error threshold value satisfies the following formula:

[0007]

[0008] wherein i=1, 2; K=2, K represents the number of classification of classification expression, that is, the overexpression group N1 and the normal expression group N2; μ i represents the mean value of the gene expression data in N i ; and σ represents the standard deviation of the gene expression data.

[0009] Optionally, the cancer cell ratio satisfies the following formula:

[0010]

[0011] wherein, η represents the cancer cell ratio, Num(N2) represents the number of cells in the overexpression group N2, and Num(N1+N2) represents the sum of the number of cells in the normal expression group N1 and the overexpression group N2.

[0012] Optionally, the method for generating a correlation model of a targeted drug and a proto-oncogene using the database comprises the following steps: extracting feature words from the database; defining correlation attributes, wherein the correlation attributes comprise the name of a targeted drug, a target gene, and a pharmacological action mode; classifying and labeling the feature words using the correlation attributes; designing a unification rule; unifying the classified feature words using the unification rule; constructing a correlation structure according to the unified feature words and the correlation attributes; setting a screening condition; and generating a correlation model using the correlation structure in combination with the screening condition. This method uses a database comprising targeted drug information to generate a correlation model by extracting feature values, thereby accurately obtaining a targeted drug corresponding to overexpression of a target gene.

[0013] Optionally, the correlation structure comprises: t +B+C i wherein, A represents a cancer type, t = 1, 2, …, n, different subscripts t correspond to different cancer types, B represents a target gene, C i represents a targeted drug for A t , i = 1, 2, …, n, and different subscripts i correspond to different targeted drugs.

[0014] Optionally, the screening condition comprises: the cancer type A is caused by overexpression of the target gene B; and the targeted drug C is used to inhibit non-physiological activity caused by overexpression of the target gene B. By setting the screening condition, a targeted drug with a mechanism of inhibiting non-physiological activity caused by overexpression of a gene can be obtained.

[0015] Optionally, the correlation model comprises: t +B’+C′ i and A t +B’+NA, wherein, A represents a cancer type, t = 1, 2, …, n, different subscripts t correspond to different cancer types, B’ represents an overexpressed target gene; C′ iThe target drugs represent the inhibition of non-physiological activity caused by overexpression of the target gene B, i=1, 2, …, n, different subscripts i correspond to different target drugs, and NA represents no corresponding target drug for overexpression of the target gene B.

[0016] Optionally, the optimization of the correlation model comprises the following steps: counting the word frequency of the feature words; and designing the unification rule by using the word frequency; and the unification rule comprises: when different feature words represent the same meaning, extracting the feature word with the highest word frequency as the unified feature word; and mapping the remaining feature words to the feature word with the highest word frequency. By normalizing the feature words, the diversity of the expression of the feature words with the same meaning is eliminated, the probability of error matching is reduced, the error of the overall operation is reduced, and the correlation model is optimized, so that the method is more practical.

[0017] Optionally, the target drug recommendation information is obtained by using the optimized correlation model in combination with the target gene, comprising the following steps: obtaining the feature word similarity by using the unification rule; unifying the target gene in combination with the feature word similarity; obtaining the name of the recommended target drug by matching the unified target gene by using the correlation model; and extracting the target drug recommendation information in the database according to the name of the recommended target drug. The use of feature word similarity eliminates the probability of matching errors caused by the diversity of the target gene name, and the use of the unified target gene in combination with the correlation model for matching improves the matching efficiency and accuracy.

[0018] Optionally, the target drug recommendation information is obtained by using the optimized correlation model in combination with the target gene, comprising the following steps: obtaining the feature word similarity by using the unification rule; unifying the target gene in combination with the feature word similarity; obtaining the name of the recommended target drug by matching the unified target gene by using the correlation model; and extracting the target drug recommendation information in the database according to the name of the recommended target drug. The use of feature word similarity eliminates the probability of matching errors caused by the diversity of the target gene name, and the use of the unified target gene in combination with the correlation model for matching improves the matching efficiency and accuracy.

[0019] In a second aspect, the present application further provides a tumor-targeted drug recommendation system, which is suitable for the tumor-targeted drug recommendation method and comprises a first unit, a second unit, a third unit and a fourth unit. The first unit is used to provide gene expression information of a patient, obtain a cancer cell ratio by using the gene expression information, and determine a target gene by using the cancer cell ratio. The second unit is used to provide a database comprising targeted drug information, generate an association model of a targeted drug and a proto-oncogene by using the database, and optimize the association model. The third unit is used to obtain targeted drug recommendation information by using the optimized association model in combination with the target gene, wherein the targeted drug recommendation information comprises a recommendation strength of the targeted drug. The fourth unit is used to obtain a targeted drug recommendation table by integrating the targeted drug recommendation information. The system can quickly and accurately determine and recommend a tumor-targeted drug suitable for individualized conditions of a patient by the interaction of the four functional units in combination with the tumor-targeted drug recommendation method.

[0020] In a third aspect, the present application further provides a tumor-targeted drug recommendation system, which comprises an input device, a processor, a memory and an output device, wherein the input device, the processor, the memory and the output device are connected to each other, the memory is used to store a computer program comprising program instructions, and the processor is configured to invoke the program instructions to execute the tumor-targeted drug recommendation method. The system has a compact structure, strong applicability, and greatly improved operation efficiency, and provides an entity system model for realizing the tumor-targeted drug recommendation suitable for individualized conditions of a patient in combination with the tumor-targeted drug recommendation method. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 Flow chart of the tumor-targeted drug recommendation method of the present application;

[0022] Figure 2 Scatter plot of proto-oncogene expression of mixed urothelial cell subpopulations of the present application;

[0023] Figure 3 Schematic diagram of classification results of different proto-oncogenes in mixed urothelial cell subpopulations of the present application;

[0024] Figure 4 Schematic diagram of cancer cell ratio in mixed urothelial cell subpopulations of the present application. DETAILED DESCRIPTION

[0025] Specific embodiments of the present invention will now be described in detail. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the invention.

[0026] Throughout this specification, references to "an embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, the phrases "in an embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. Moreover, those skilled in the art will understand that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale.

[0027] Please see Figure 1 In an optional embodiment, a tumor-targeted drug recommendation method of the present invention includes the following steps: S1, providing gene expression information of the patient; specifically, the gene expression information includes gene expression data and gene expression categories, wherein the gene expression data includes different proto-oncogenes and their expression values.

[0028] In this embodiment, taking a patient with renal pelvis cancer as an example, the provision of the patient's gene expression information includes the following steps: collecting a morning urine sample from the patient with renal pelvis cancer; classifying the cells in the morning urine sample using a UF-5000 urine analyzer; statistically analyzing the cell populations to obtain the number of cell populations and the total number of cells in each cell population; comparing the number of cell populations and the total number of cells in each cell population with detection standards, and obtaining a test sample when the detection standards are met, wherein the detection standards are determined according to the actual situation; sequencing the test sample to obtain the cell expression profile of the test sample, specifically using single-cell RNA sequencing technology to obtain the cell expression profile of the test sample; using the cell expression profile to obtain the cell expression profile data of a mixed urothelial cell subpopulation, which includes normal urothelial cells and cancerous urothelial cells; extracting the gene expression data of different proto-oncogenes from the cell expression profile data of the mixed urothelial cell subpopulation, please refer to [link to relevant documentation]. Figure 2 , Figure 2A scatter plot of proto-oncogene expression of mixed urothelial cell subpopulations, wherein the proto-oncogenes include ALK, EGFR, FGFR2, FLT3, HER2, INSR, KIT, NTRK1, PDGFRA, PD-L1, PIK3CA, and SRC, and the ordinate represents the expression value of the corresponding gene, and the abscissa represents the number of cells.

[0029] In an alternative embodiment, the tumor-targeting drug recommendation method of the present application further comprises the following step: S2, obtaining a cancer cell ratio using the gene expression information.

[0030] In detail, in the present embodiment, the step of obtaining a cancer cell ratio using the gene expression information comprises the following steps: randomly selecting N data as initial center points from different gene expression data of the proto-oncogenes; comparing the gene expression data with the initial center points; according to the comparison results, combining the gene expression categories, classifying the gene expression data into an overexpression group and a normal expression group; respectively using the gene expression data in the overexpression group and the normal expression group to redefine the initial center by calculating the mean value; using the redefined initial center, combining the gene expression categories to classify the gene expression data again to obtain an overexpression group N1 and a normal expression group N2 of the proto-oncogenes that satisfy the error threshold, in detail, the error threshold satisfies the following formula:

[0031]

[0032] wherein i = 1, 2; K = 2, K represents the number of classification expressions, i.e. the overexpression group N1 and the normal expression group N2; μ i represents the mean value of the gene expression data in N i ; σ represents the standard deviation of the gene expression data; and then obtaining a cancer cell ratio caused by the proto-oncogenes through the overexpression group N1 and the normal expression group N2, wherein the cancer cell ratio satisfies the following formula:

[0033]

[0034] wherein η represents the cancer cell ratio, Num(N2) represents the number of cells in the group of the overexpression group N2, and Num(N1+N2) represents the total number of cells in the normal expression group N1 and the overexpression group N2. The present method screens the overexpression group N1 and the normal expression group N2 of different proto-oncogenes under the same termination condition, thereby improving the calculation efficiency.

[0035] See Figure 3 and Figure 4In another optional embodiment, the oncogene expression information of the mixed uroepithelial cell subpopulation is used in combination with the S2 step to obtain the cancer cell ratio of different oncogenes in the mixed uroepithelial cell subpopulation, wherein Figure 3 FIG. 6 is a schematic diagram of the classification results of different oncogenes in the mixed uroepithelial cell subpopulation, wherein the title is the gene name, the horizontal axis represents the number of cells, and the vertical axis represents the expression value. The a line is the highest expression value, the b line is the lowest expression value, and the c line is the overall intercept line. The N1 category is above the c line, and the N2 category is below the c line. When the c line is in the optimal position, the overall has the global "smallest sum of squared errors", that is wherein |N1| and |N2| are the number of samples in the category, σ 2 N1, σ 2 N2 is the square of the standard deviation of the samples in the category. Figure 4 FIG. 7 is a schematic diagram of the cancer cell ratio of INSR, MET, BRAF, SRC, IGF1R, EGFR, HER2, PIK3CA, EZH2, CDK6, BCL2, MST1R, PD-L1, and FGFR1 in the mixed uroepithelial cell subpopulation, wherein the horizontal axis represents the type of oncogene, and the vertical axis represents the cancer cell ratio corresponding to the oncogene.

[0036] In an optional embodiment, the tumor-targeted drug recommendation method of the present application further comprises the following step: S3, determining the target gene by the cancer cell ratio. In detail, in this embodiment, a threshold range is set according to the cancer cell ratio caused by the oncogene, and the oncogene whose cancer cell ratio exceeds the threshold is set as the target gene. Wherein, according to the actual situation, the setting of the threshold range is related to the type of cancer and the cell mutation rate caused by the oncogene, and can be set according to the specific test value of the related experiment.

[0037] In an alternative embodiment, the method for recommending a tumor-targeting drug of the present application further comprises the following step: S4, providing a database comprising targeting drug information. In detail, the database is constructed by a collection of marketed targeting drug information, which comprises target gene of the targeting drug, drug description and pharmacological action, the drug description comprises treating overexpression of the target gene, and the pharmacological action comprises inhibiting non-physiological activity caused by overexpression of the target gene. Specifically, the protein molecular weight of HER2 is about 185 kD, which is a transmembrane protein located in the cell membrane. HER2 belongs to tyrosine kinase coupled receptors, which consists of three parts: extracellular ligand P binding domain, single-chain transmembrane domain and intracellular tyrosine kinase domain. Under normal physiological conditions, the extracellular ligand P binding domain of HER2 can bind to the ligand P (growth factor) in the external environment, which in turn causes a change in the conformation of the intracellular tyrosine kinase domain. In this case, the tyrosine kinase domain binds to ATP in the cell, and then uses ATP as a phosphate group donor to phosphorylate the tyrosine site of the downstream substrate protein (RAS), which activates the function of the substrate protein through protein interaction, so that the extracellular growth signal is transmitted to the cell interior, and then transmitted step by step through different downstream signal factors, and finally regulates the cell proliferation process. However, under pathological conditions, the cell overexpresses HER2 for unknown reasons, and these proteins will over-transmit growth signals under the action of ligand P, causing uncontrolled proliferation of cells, which ultimately leads to the occurrence and development of tumors. Trastuzumab Q is a humanized monoclonal antibody against HER2, which has a very strong binding capacity to HER2. It can competitively bind to the extracellular ligand binding domain of HER2 in the presence of HER2 ligand P, i.e. blocking the ligand binding site of HER2. The intracellular tyrosine kinase domain of HER2 bound to trastuzumab Q does not change in conformation, and thus cannot complete the growth signal transmission process, and the proliferation of cells is inhibited, ultimately benefiting the tumor patients.

[0038] In an alternative embodiment, the method for recommending a tumor-targeting drug of the present application further comprises the following step: S5, generating a correlation model of the targeting drug and the proto-oncogene using the database; in detail, the proto-oncogene includes but is not limited to the target gene.

[0039] In detail, in this embodiment, the step of generating a correlation model of the targeting drug and the proto-oncogene using the database comprises the following step: extracting feature words from the database. Specifically, extracting feature words from the database is a prior art, which is not specifically explained here.

[0040] In another embodiment, the method further comprises the following steps: defining the association attributes, wherein the association attributes comprise the name of the targeted drug, the target gene, and the pharmacological action mode. In this embodiment, the association attributes are used to classify the feature words, and each feature word corresponds to one association attribute. For example, the name of the targeted drug comprises the feature words of afatinib, osimertinib, cetuximab, panitumumab, and necitumumab; the target gene comprises the feature words of INSR, MET, BRAF, SRC, IGF1R, EGFR, HER2, PIK3CA, EZH2, CDK6, and BCL2; and the pharmacological action mode comprises the feature words of inhibiting the expression of the target gene and inhibiting the activity of INSR.

[0041] In another embodiment, the method further comprises the following steps: classifying and labeling the feature words by using the association attributes. In this embodiment, each feature word corresponds to one association attribute. For example, the association attribute of tasigna is the name of the targeted drug, and the association attributes of renal pelvis cancer, breast cancer, prostate cancer, and colorectal cancer are the cancer types.

[0042] In another embodiment, the method further comprises the following steps: designing a unification rule, and unifying the classified feature words by using the unification rule. In this embodiment, FGFR1, FGFR2, FGFR3, and FGFR4 are unified as FGFR, which eliminates the expression diversity of feature words with the same meaning, reduces the probability of false matching, reduces the error of the overall operation, optimizes the association model, and makes the method more practical.

[0043] In another embodiment, the method further comprises the following steps: constructing an association structure according to the unified feature words and the association attributes, wherein the association structure comprises: t A + B + C i wherein A represents the cancer type, t = 1, 2, …, n, different subscripts t correspond to different cancer types, B represents the target gene, C i represents the targeted drug for A, i = 1, 2, …, n, and different subscripts i correspond to different targeted drugs. For example, the association structure can be gastric cancer + EGFR + afatinib, gastric cancer + EGFR + osimertinib, gastric cancer + EGFR + cetuximab, lung cancer + EGFR + afatinib, lung cancer + EGFR + panitumumab, and the like, which corresponds to the random matching of feature words.

[0044] In another embodiment, the step of generating a correlation model between targeted drugs and proto-oncogenes using the database further includes the following step: setting screening conditions. In this embodiment, the screening conditions include: the cancer type A is caused by the overexpression of the target gene B; and the targeted drug C is used to inhibit the non-physiological activity caused by the overexpression of the target gene B. By setting screening conditions, i.e., setting matching rules, targeted drugs whose pharmacological mechanism is to inhibit the non-physiological activity caused by gene overexpression can be obtained in a targeted manner.

[0045] In yet another embodiment, the step of generating an association model between targeted drugs and proto-oncogenes using the database further includes the following step: generating an association model using the association structure and the screening conditions; specifically, the association model includes: A t +B'+C' i and A t +B'+NA, where A represents the cancer type, t = 1, 2, ..., n, different subscripts t correspond to different cancer types, B' represents the overexpressed target gene; C' i The term "targeted drug" represents the drug that inhibits the non-physiological activity caused by the overexpression of the target gene B, where i = 1, 2, ..., n, and different subscripts i correspond to different targeted drugs. NA indicates that there is no corresponding targeted drug for the overexpression of the target gene B. Specifically, in this embodiment, the association model can be: renal pelvis cancer + EGFR + afatinib, renal pelvis cancer + AEG1 + NA, etc., which is equivalent to obtaining an association model that conforms to the matching rules.

[0046] In summary, step S5 of this method utilizes a database containing targeted drug information, extracts feature words to generate an association model, and sets screening conditions to specifically screen for targeted drugs whose mechanism of action is to inhibit gene overexpression and thus non-physiological activity.

[0047] In an optional embodiment, a tumor-targeted drug recommendation method of the present invention further includes the following step: S6, optimizing the association model.

[0048] In detail, optimizing the association model includes the following steps: calculating the word frequency of the feature words; designing the unification rule using the word frequency; the unification rule includes: when different feature words represent the same meaning, extracting the feature word with the highest word frequency as the unified feature word; mapping the remaining feature words to the feature word with the highest word frequency. Specifically, in this embodiment, the word frequency represents the frequency of feature words appearing in the database; by regularizing the feature words, the diversity of expression of feature words with the same meaning is eliminated, reducing the probability of false matching, thereby reducing the overall calculation error and optimizing the association model, making this method more practical.

[0049] In an alternative embodiment, the tumor-targeted drug recommendation method of the present application further comprises the following step: S7, obtaining targeted drug recommendation information by using the optimized correlation model in combination with the target gene.

[0050] In detail, the step of obtaining targeted drug recommendation information by using the optimized correlation model in combination with the target gene comprises the following steps: obtaining a feature word similarity by using the unification rule, i.e., the similarity between the remaining feature words and the feature word with the highest frequency in step S6. Specifically, the similarity can be represented by the consistency of characters and / or the consistency of character sequences. For example, FGFR1, FGFR2, FGFR3 and FGFR4 are collectively referred to as FGFR, where the similarity can be determined by the consistency of FGFR1 and FGFR characters and / or the consistency of character sequences. The specific representation parameters are set according to the actual situation. The target gene is unified in combination with the feature word similarity. The name of the recommended targeted drug is obtained by matching the unified target gene using the correlation model. The targeted drug recommendation information is extracted from the database according to the name of the recommended targeted drug. In detail, the targeted drug recommendation information includes the recommendation strength of the targeted drug, which is positively correlated with the size of the cancer cell ratio. The use of feature word similarity eliminates the probability of matching errors caused by the diversity of target gene names. Meanwhile, the use of the unified target gene in combination with the correlation model for matching improves the matching efficiency and accuracy.

[0051] In an alternative embodiment, the tumor-targeted drug recommendation method of the present application further comprises the following step: S8, integrating the targeted drug recommendation information to obtain a targeted drug recommendation table. In detail, the step of integrating the targeted drug recommendation information to obtain a targeted drug recommendation table comprises the following steps: sorting the target genes according to the cancer cell ratio to obtain a first sorting result; sorting the recommended targeted drugs according to the recommendation strength to obtain a second sorting result; and generating a targeted drug recommendation table in combination with the first sorting result and the second sorting result. The targeted drug recommendation table obtained by integrating the targeted drug recommendation information is helpful for the visual display of the targeted drug recommendation information.

[0052] Referring to Table 1, in an alternative embodiment, the personalized drug use reference list for the patient with renal pelvis cancer is obtained by using the proto-oncogene expression information of the mixed uroepithelial cell subpopulation in combination with steps S2-S8, and the specific content is shown in Table 1:

[0053]

[0054] According to the ranking results of the proto-oncogenes in Table 1, the proto-oncogenes and the corresponding targeted drugs are listed again in sequence as a whole, and the targeted drug with the highest recommended intensity is determined as the first choice of targeted drug for the patient with renal pelvis cancer to take neoadjuvant therapy strategy, and Table 1 shows the top 6 recommended targeted drugs for the patient with renal pelvis cancer.

[0055] In summary, the present method calculates the cancer cell ratio to determine the target gene through the gene expression information of the patient, and realizes the accurate determination and recommendation of the tumor targeted drug suitable for the individualized condition of the patient by the accurate matching of the target gene and the current targeted drug in the case of being unable to obtain the cancer tissue biopsy sample. On the other hand, the present method is not only suitable for the non-invasive examination of renal pelvis cancer, but also can provide reference and basis for the preoperative neoadjuvant therapy strategy of other cancers caused by overexpression of proto-oncogenes, thereby reducing the risk of causing great trauma or non-specific damage to the patient's body.

[0056] The present application also provides a tumor targeted drug recommendation system, which is suitable for the tumor targeted drug recommendation method and comprises a first unit, a second unit, a third unit and a fourth unit; the first unit is used for providing gene expression information of a patient, obtaining a cancer cell ratio by using the gene expression information, and determining a target gene by using the cancer cell ratio; the second unit is used for providing a database, the database comprising targeted drug information, generating an association model of targeted drugs and proto-oncogenes by using the database, and optimizing the association model; the third unit is used for obtaining targeted drug recommendation information by using the optimized association model in combination with the target gene, the targeted drug recommendation information comprising recommended intensity of the targeted drugs; and the fourth unit is used for obtaining a targeted drug recommendation table by integrating the targeted drug recommendation information. The present system quickly and accurately determines and recommends the tumor targeted drug suitable for the individualized condition of the patient by the interaction of the four functional units in combination with the tumor targeted drug recommendation method.

[0057] The present application also provides a tumor targeted drug recommendation system, which comprises an input device, a processor, a memory and an output device, and the input device, the processor, the memory and the output device are connected with each other, wherein the memory is used for storing a computer program, the computer program comprises program instructions, and the processor is configured to call the program instructions to execute the tumor targeted drug recommendation method. The present system has compact structure and strong applicability, and not only provides an entity system model for realizing the recommendation of the tumor targeted drug suitable for the individualized condition of the patient in combination with the tumor targeted drug recommendation method, but also improves the practicability and expandability of the present application while ensuring the running speed and efficiency.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; the present application has been described in detail with reference to the foregoing embodiments, and those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and they should be covered in the scope of the claims and the description of the present application.

Claims

1. A method for recommending tumor-targeted drugs, characterized in that, The method comprises the following steps: providing gene expression information of a patient, the gene expression information comprising gene expression data and gene expression categories, the gene expression data comprising different proto-oncogenes and expression values thereof; obtaining a cancer cell ratio by using the gene expression information, comprising the following steps: randomly selecting N data from the gene expression data of different proto-oncogenes as initial center points; comparing the gene expression data with the initial center points; according to the comparison results, combining the gene expression categories, and dividing the gene expression data into an overexpression group and a normal expression group; redefining the initial center points by calculating the mean values of the gene expression data in the overexpression group and the normal expression group, respectively; re-classifying the gene expression data using the redefined initial center in combination with the gene expression data to obtain an overexpression set of the proto-oncogene that uniquely satisfies an error threshold and a normal expression set ; by said overexpression group and said normal expression group obtaining a cancer cell ratio caused by said proto-oncogene; the cancer cell ratio satisfies the following formula: ； wherein, represents the ratio of the cancer cells, represents the number of cells in the overexpression group of the group, represents the total number of cells in the normal expression group and the overexpression group . determining a target gene by using the cancer cell ratio; providing a database, the database comprising information of targeted drugs; generating a correlation model of targeted drugs and proto-oncogenes by using the database, the proto-oncogenes comprising the target gene; The correlation model comprises: and wherein A represents a cancer type, different subscripts t correspond to different cancer types, B represents an overexpressed target gene; C represents a targeted drug for inhibiting non-physiological activity caused by overexpression of the target gene B, different subscripts t correspond to different cancer types, different subscripts t correspond to different targeted drugs, D represents that there is no corresponding targeted drug for overexpression of the target gene B. optimizing the correlation model; obtaining targeted drug recommendation information by using the optimized correlation model and combining the target gene, the targeted drug recommendation information comprising a recommendation strength of the targeted drug, comprising the following steps: obtaining a feature word similarity by using a unification rule; unifying the target gene by combining the feature word similarity; obtaining the name of a recommended targeted drug by matching the unified target gene by using the correlation model; extracting targeted drug recommendation information from the database according to the name of the recommended targeted drug; integrating the targeted drug recommendation information to obtain a targeted drug recommendation table.

2. The method of claim 1, wherein the method further comprises: the error threshold satisfies the following formula: ， wherein, K=2, K represents the number of categories of the categorical expression, i.e. overexpression group and normal expression group ; represents the mean of the gene expression data described in the represents the standard deviation of the gene expression data. 3.The method of claim 1, wherein, the step of generating a correlation model of targeted drugs and proto-oncogenes by using the database comprises the following steps: extracting feature words from the database; defining correlation attributes, the correlation attributes comprising the name of a targeted drug, a target gene and a pharmacological action mode; classifying and labeling the feature words by using the correlation attributes; designing a unification rule; unifying the classified feature words by using the unification rule; constructing a correlation structure according to the unified feature words and the correlation attributes; setting a screening condition; generating a correlation model by using the correlation structure and combining the screening condition.

4. The method of claim 3, wherein the tumor targeting drug recommendation method is characterized by, the step of optimizing the correlation model comprises the following steps: counting the word frequency of the feature words; designing the unification rule by using the word frequency; the unification rule comprises: when different feature words represent the same meaning, extracting the feature word with the highest word frequency as the unified feature word; mapping the remaining feature words to the feature word with the highest word frequency.

5. The method of claim 1, wherein the method further comprises: determining a tumor type of the tumor based on the tumor image; and determining the recommended drug based on the tumor type. the step of integrating the targeted drug recommendation information to obtain a targeted drug recommendation table comprises the following steps: sorting the target gene by using the cancer cell ratio to obtain a first sorting result; sorting the recommended targeted drug by using the recommendation strength to obtain a second sorting result; generating a targeted drug recommendation table by combining the first sorting result and the second sorting result.

6. A tumor-targeted drug recommendation system, the system being adapted for use in the tumor-targeted drug recommendation method according to any one of claims 1 to 5, characterized in that, comprise: a first unit, a second unit, a third unit and a fourth unit; The first unit is configured to provide gene expression information of a patient, obtain a cancer cell ratio by using the gene expression information, and determine a target gene by using the cancer cell ratio; The second unit is configured to provide a database including targeted drug information, generate a correlation model of a targeted drug and a proto-oncogene by using the database, and optimize the correlation model; The third unit is configured to obtain targeted drug recommendation information by using the optimized correlation model and the target gene, the targeted drug recommendation information including a recommended strength of the targeted drug; The fourth unit is configured to integrate the targeted drug recommendation information to obtain a targeted drug recommendation table. 7.A tumor-targeted drug recommendation system, characterized in that, An apparatus includes an input device, a processor, a memory, and an output device, which are connected to each other, wherein the memory is configured to store a computer program including program instructions, and the processor is configured to invoke the program instructions to execute a method according to any one of claims 1 to 5.

Images