Lung cancer drug resistance prediction method based on user feature analysis

Abstract

The invention discloses a lung cancer drug resistance prediction method based on user feature analysis, and relates to the technical field of biological information processing, and the method comprises the steps: calling a biological gene pool, analyzing and mining a gene interaction network based on gene variation statistics of targeted drugs, supervising and training a drug resistance detector based on a self-attention mechanism, receiving user inspection data, and predicting the drug resistance of the lung cancer; and controlling a sequencer to carry out gene sequencing detection, determining a user gene sequence, transmitting the user gene sequence to a drug resistance detector, and determining a variation trend of a targeted fragment so as to analyze drug resistance prediction data under a time sequence. The method and the device are used for solving the technical problem that the accuracy of a prediction effect is limited due to insufficient depth and comprehensiveness of drug resistance analysis in the prior art. Drug resistance analysis is carried out based on variation prediction of the target gene, and the accuracy of drug resistance prediction can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of biological information processing, and in particular to a method for predicting lung cancer drug resistance based on user feature analysis. Background Art

[0002] At present, the drug resistance problem in lung cancer drug intervention has become a major challenge. Drug-based targeted therapy intervention has significantly improved the status of some patients. However, with the continuation of drug intervention, drug resistance to targeted drugs is prone to occur, resulting in decreased efficacy and recurrence of the disease, which poses a greater threat to patients' lives.

[0003] In the existing technology, the prediction of drug resistance to intervention drugs mainly relies on the patient's medication response and basic analysis of genetic data, which is usually too dependent on static genetic features, and the direct analysis effect is limited and unpredictable. In addition, the existing genomic analysis methods are mostly traditional single gene mutation detection, and the technical means are relatively limited, which cannot fully reflect the interaction between different genes and the complex relationship between them and drug resistance.

[0004] In summary, the existing technology still has certain technical limitations, and the depth and comprehensiveness of drug resistance analysis are insufficient, resulting in limited accuracy of prediction effects. Summary of the invention

[0005] The present application provides a lung cancer resistance prediction method based on user feature analysis, which is used to solve the technical problem that the depth and comprehensiveness of drug resistance analysis in the prior art are insufficient, resulting in limited accuracy of prediction results.

[0006] In view of the above problems, the present application provides a method for predicting lung cancer resistance based on user feature analysis.

[0007] In the first aspect, the present application provides a method for predicting lung cancer resistance based on user feature analysis, the method comprising: calling a biological gene library, conducting statistical analysis of gene variations based on targeted drugs, and mining a gene interaction network, wherein the gene interaction network is constructed based on the interaction between targeted drugs and gene fragments; using the gene interaction network as a basic database, supervising the training of a drug resistance detector based on a self-attention mechanism, wherein the drug resistance detector comprises an identification unit and a detection unit; receiving user submitted data, performing gene sequencing detection by controlling a sequencer, and determining the user gene sequence, wherein the method comprises multiple parallel detections; transmitting the user gene sequence to the drug resistance detector, performing self-attention recognition based on targeted gene fragments and judgment based on the gene interaction network, determining the variation trend of the targeted fragments, and determining the drug resistance prediction data under the time series according to the variation trend.

[0008] In the second aspect, the present application provides a lung cancer resistance prediction system based on user feature analysis, the system comprising: a mining module, the mining module is used to call a biological gene library, to mine a gene interaction network based on statistical analysis of gene variations of targeted drugs, wherein the gene interaction network is constructed based on the interaction between targeted drugs and gene fragments; a construction module, the construction module is used to supervise the training of a drug resistance detector based on a self-attention mechanism based on the gene interaction network as a basic database, wherein the drug resistance detector comprises an identification unit and a detection unit; a sequencing module, the sequencing module is used to receive user submitted data, perform gene sequencing detection by controlling a sequencer, and determine the user's gene sequence, wherein multiple parallel detections are included; a prediction module, the prediction module is used to transmit the user's gene sequence to the drug resistance detector, perform self-attention recognition based on targeted gene fragments and judgment based on gene interaction networks, determine the variation trend of the targeted fragments, and determine the drug resistance prediction data under the time series according to the variation trend.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: The embodiment of the present application provides a lung cancer resistance prediction method based on user feature analysis, calling a biological gene library, mining a gene interaction network based on the statistical analysis of gene variation of targeted drugs, and constructing the gene interaction network based on the interaction of targeted drugs and gene fragments; using the gene interaction network as the basic database, supervising the training of a drug resistance detector based on a self-attention mechanism, receiving user inspection data, and controlling the sequencer to perform gene sequencing detection, determining the user's gene sequence and transmitting it to the drug resistance detector, performing self-attention recognition based on targeted gene fragments and judgment based on gene interaction networks, determining the variation trend of targeted fragments, and determining the drug resistance prediction data under the time series according to the variation trend. It is used to solve the technical problem that the depth and comprehensiveness of drug resistance analysis in the prior art are insufficient, resulting in limited accuracy of the prediction effect. Drug resistance analysis based on variation prediction of targeted genes can effectively improve the accuracy of drug resistance prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of a lung cancer resistance prediction method based on user feature analysis is provided for this application; Figure 2 A schematic diagram of a process for controlling a sequencer to perform gene sequencing detection in a lung cancer drug resistance prediction method based on user feature analysis is provided for this application; Figure 3 A schematic diagram of the structure of a lung cancer resistance prediction system based on user feature analysis is provided for this application.

[0011] Explanation of the reference numerals: mining module 11, construction module 12, sequencing module 13, prediction module 14. DETAILED DESCRIPTION

[0012] This application provides a lung cancer resistance prediction method based on user feature analysis, calls a biological gene library, conducts statistical analysis of gene variation based on targeted drugs, mines gene interaction networks, supervises the training of drug resistance detectors based on self-attention mechanisms, receives user submitted data, controls sequencers to perform gene sequencing, determines user gene sequences and transmits them to drug resistance detectors, determines the variation trend of targeted fragments, and analyzes drug resistance prediction data under time series. It is used to solve the technical problem that the depth and comprehensiveness of drug resistance analysis in the prior art are insufficient, resulting in limited accuracy of prediction results.

[0013] Embodiment 1: Figure 1 As shown, the present application provides a method for predicting lung cancer drug resistance based on user feature analysis, the method comprising: S1: Call the biological gene library to mine the gene interaction network based on the statistical analysis of gene variation based on targeted drugs, wherein the gene interaction network is constructed based on the interaction between targeted drugs and gene fragments.

[0014] In the embodiment of the present application, the biological gene bank is a database that stores a large amount of genome information, including gene variation data of various populations and different disease states. In the present invention, by calling the gene bank, a large amount of genome data can be obtained, providing the basic data support required for building a gene interaction network.

[0015] During the specific implementation process, there are certain authority restrictions on the call of biological gene banks, which are generally divided into public gene banks and private gene banks, and are called within the scope of authority.

[0016] Then, we conduct a statistical analysis of gene mutations based on targeted drugs. The targeted drugs are drugs related to lung cancer. Therefore, analyzing gene characteristics and gene mutations related to these drugs is the key to revealing the mechanism of drug resistance. By performing statistical analysis on the mutation data of targeted drugs and related genes, we can identify the types of gene mutations related to drug action and their frequency of occurrence. This step helps to determine which gene mutations may cause drug resistance.

[0017] After completing the statistical analysis of gene variation, the gene interaction network is a graphical representation of the interaction between genes and the interactive variation characteristics of genes and drugs. It usually reveals the mutual influence of gene combinations and drug interactions by analyzing the mutual combinations of gene fragments.

[0018] In this application, the construction of the gene interaction network is based on the interactive relationship between targeted drugs and gene fragments. That is, targeted drugs interact with specific gene fragments, thereby affecting the expression or variation of these genes. The variation of gene fragments may affect the efficacy of the drug and thus produce drug resistance. Therefore, the interactive construction of targeted drugs and gene fragments provides a direction for the gene interaction network, ensuring that the network can accurately reflect the variation trend related to targeted drugs.

[0019] Specifically, through data calling and integration, genetic data under various disease states are obtained from the biological gene library; gene variation analysis is performed, and the potential drug resistance mechanism is determined by statistically analyzing the relationship between targeted drugs and related gene variations; interaction network construction is performed, and a gene interaction network is constructed based on the interaction between targeted drugs and gene fragments to reflect the interaction between genes and targeted drugs.

[0020] For example, if a targeted drug (such as EGFR inhibitor) is used to treat lung cancer, the mutation of related genes (such as EGFR gene) may affect the effect of the drug. Through statistical analysis, drug resistance markers related to EGFR gene mutations are identified, and the EGFR gene is connected to other genes that may cause drug resistance (such as KRAS gene) through a gene interaction network, thereby revealing the drug resistance performance in different patients.

[0021] In summary, by calling the biological gene library and conducting statistical analysis of gene variation based on targeted drugs, a gene interaction network that can accurately predict lung cancer resistance is gradually constructed, providing a solid foundation for subsequent drug resistance prediction.

[0022] Furthermore, by using statistical analysis of gene variation based on targeted drugs to mine gene interaction networks, step S1 of this application includes: Based on the biological gene library, extract the first sample genome, wherein the first sample genome includes gene sequences of multiple period time nodes under the medication cycle; traverse the first sample genome for comparison, identify the variant gene fragments, and determine the first targeted gene fragment group; complete the extraction and determination of the Nth sample genome, and determine the Nth targeted gene fragment group; perform statistical analysis on the first targeted gene fragment group to the Nth targeted gene fragment group, and construct the gene interaction network, wherein the gene interaction network identifies a linear relationship between gene variability and drug resistance.

[0023] First, based on the biological gene library, the first sample genome is extracted. The first sample genome is the genome data of a specific individual extracted from the library, that is, the set of gene sequences under the entire medication cycle, to obtain the gene variation information of the patient in the time zone before and after treatment. In particular, taking before and after treatment as a complete cycle, the first sample genome contains the gene sequences of each cycle time node under different medication cycle nodes. Each cycle time node represents a different stage of the patient in the process of targeted drug treatment. The changes in gene sequences in these stages can reflect the impact of drugs on genes and the gradual change of drug resistance. Therefore, data from multiple time nodes provide dynamic observation of gene evolution, ensuring the accuracy and comprehensiveness of the prediction.

[0024] Next, the first sample genome is traversed for comparison, that is, the same gene fragments of each first sample gene are compared to determine the evolution state of each gene fragment within the cycle. Exemplarily, an efficient comparison algorithm is used to identify gene fragments with mutations in the sample, that is, gene fragments related to drug resistance.

[0025] For example, gene mutations may include single nucleotide polymorphisms (SNPs), insertions and deletions (Indels), etc. These mutations may affect the targeting effect of drugs. Therefore, identifying variant gene fragments is a key step in building a gene interaction network. Through this comparison process, drug-related gene mutation fragments can be obtained to determine whether they are related to the development of drug resistance.

[0026] After identifying the variant gene fragments, the first targeted gene fragment group is further determined. That is, according to the mechanism of action of the targeted drug, the gene fragments related to the drug target in the complete gene sequence are selected and grouped into a targeted gene fragment group.

[0027] In order to ensure the universality and accuracy of the analysis, the sample genome is extracted and analyzed multiple times, and the extraction and judgment methods based on the sample genome are the same each time, with only differences in samples. After the extraction of the first sample genome is completed, the above steps are repeated to extract the genome data of the Nth sample from the biological gene bank, and the gene variation is also compared and identified. The genome of each sample provides information related to targeted drug therapy, ensuring that the increase in sample size can improve the predictive and generalization capabilities of the model.

[0028] Subsequently, statistical analysis is performed on the first targeted gene fragment group to the Nth targeted gene fragment group, that is, by performing statistical analysis on the targeted gene fragments in different samples, the relationship between the variability between gene fragments and drug resistance is identified. Through statistical analysis, the variation frequency of each gene fragment can be evaluated, and gene fragments with high variability and possible drug resistance can be identified. This analysis not only reveals the genetic differences between different individuals, but also provides a reliable basis for drug resistance prediction.

[0029] Finally, the gene interaction network is constructed, wherein the gene interaction network identifies a linear relationship between gene variability and drug resistance. The gene interaction network combines the variation data of the targeted gene fragments with the drug resistance information to construct a multi-dimensional network structure, reflecting the interaction between different gene fragments and how these gene variations affect the effect of drug treatment, and determines the linear relationship between the mutual influence of gene variation and drug resistance based on drug efficacy, so as to provide a scientific basis.

[0030] In summary, by constructing a gene interaction network related to targeted drug resistance, it can help identify gene variants closely related to drug resistance and provide effective support for drug resistance prediction.

[0031] Further, statistical analysis is performed on the first targeted gene segment group to the Nth targeted gene segment group. Step S1 of the present application includes: Traversing the first targeted gene fragment group to the Nth targeted gene fragment group, performing cluster analysis, and determining M gene fragment clusters, wherein M is the number of distinguished targeted gene fragments; traversing the M gene fragment clusters, screening and determining X targeted gene fragments according to the preset number within the cluster, wherein X is a positive integer less than or equal to M, and the X targeted gene fragments are gene fragments with strong correlation with drug resistance; constructing the gene interaction network for the X targeted gene fragments.

[0032] First, the first targeted gene segment group to the Nth targeted gene segment group are traversed to perform cluster analysis. Specifically, the targeted gene segment group includes gene segments related to targeted drugs in different groups of sample sequences. Different variant segments and different quantities may exist in each sample sequence. Multi-sample statistical analysis can effectively improve the completeness and accuracy of mining.

[0033] Among them, these gene fragments may show different mutation patterns. Cluster analysis is to group these gene fragments according to their mutation characteristics, with the aim of discovering clusters of gene fragments with similar mutation trends. Cluster analysis classifies gene fragments with similar characteristics into one category by using common clustering algorithms, such as K-means or hierarchical clustering. The result of cluster analysis will generate M gene fragment clusters, where M is the number of distinct targeted gene fragments. This means that each cluster represents a class of targeted gene fragments that are similar in mutation characteristics. In this way, multiple groups of gene fragments associated with drug resistance can be identified, and possible common features between them can be revealed.

[0034] Next, the M gene fragment clusters are traversed to preset the number of clusters and X targeted gene fragments are screened and determined. After the cluster analysis is completed, each gene fragment cluster may contain multiple gene fragments. In this step, the preset number of clusters refers to the number of gene fragments in each cluster, and the gene fragments with the most significant variation characteristics and the strongest correlation with drug resistance are screened out. These gene fragments are the required X targeted gene fragments. Among them, X is a positive integer less than or equal to M, indicating that the number of gene fragments screened will not exceed the total number of clusters. The X targeted gene fragments are gene fragments with strong correlation with drug resistance. These fragments usually show a high degree of drug resistance correlation in multiple analyses, and therefore can effectively help build a drug resistance prediction model.

[0035] Finally, for the X targeted gene fragments, that is, after screening out the gene fragments with strong correlation with drug resistance, they are used as key nodes to construct an interaction network. By revealing the interaction relationship between these gene fragments, the gene interaction network can help better understand how drug resistance spreads and evolves at the genetic level, and can reveal the synergistic effect of resistance-related genes and how they jointly affect the effect of drugs. Therefore, the constructed gene interaction network can not only show how gene variation plays a role in drug resistance, but also help identify key targets.

[0036] In summary, by analyzing and screening gene fragments with strong correlation with drug resistance, a gene interaction network containing drug resistance information is gradually constructed, which helps provide a reference for the subsequent accurate prediction of drug resistance.

[0037] Furthermore, for the X targeted gene fragments, the gene interaction network is constructed, and step S1 of the present application includes: The X targeted gene fragments are traversed, and for a single targeted gene fragment, the sample genome is traversed to perform variation evolution trend mining under periodic time sequence to determine a first interaction network; for a combined targeted gene fragment, the sample genome is traversed to perform variation evolution trend mining under periodic time sequence to determine a second interaction network; the first interaction network and the second interaction network are integrated to determine the gene interaction network.

[0038] In the embodiment of the present application, first, the X targeted gene fragments are traversed, and for a single targeted gene fragment, that is, each targeted gene fragment is independently mined for variation evolution trend under periodic time series, the periodic variation evolution trend of each targeted gene fragment for the targeted drug is determined, and the first interaction network is determined by integration.

[0039] Specifically, each targeted gene fragment, that is, the individual gene fragment with significant drug resistance correlation determined in the above screening step, is analyzed separately. The time series data analysis method is adopted, that is, each cycle time node is used as the cycle time series, and the variation evolution trend of the individual gene fragment is mined to observe the variation of the gene fragment at different time points.

[0040] Among them, the variation evolution trend reveals the change pattern of gene fragments over time. For example, some gene fragments may show a certain variation trend in the early stage, and further intensify or slow down in the later stage. Through time series analysis, this process can reveal how drug resistance gradually changes over time, thereby affecting the effect of the drug. By performing this analysis on single-targeted gene fragments separately, the first interactive network can be determined, that is, the evolutionary state of variation and resistance trends under the periodic time series under the independent action of single-targeted gene fragments.

[0041] Furthermore, the combined targeted gene fragment refers to a joint analysis of multiple targeted gene fragments, that is, a combination of at least two targeted gene fragments, the purpose of which is to study the interaction between these gene fragments and their combined effects.

[0042] Similarly, similar to the analysis of single-targeted gene fragments, the combined targeted gene fragments also use the periodic time series analysis method to explore their possible synergistic or interdependent changing trends in periodic treatment by analyzing the variation of the combined gene fragments at different time nodes. For example, the variation trend of some combined gene fragments may alternately enhance or inhibit each other at a certain stage, further affecting the generation of drug resistance. By analyzing each combined gene fragment, the second interaction network is determined, that is, the joint evolution pattern of multiple targeted gene fragments and their mutual relationship.

[0043] Finally, in the above steps, the interaction network of a single targeted gene fragment and the interaction network of a combination of multiple targeted gene fragments were constructed respectively. In order to more comprehensively reflect the complex relationship between gene fragments and drug resistance, these two interaction networks need to be fused. The fusion process combines the evolutionary trend of a single gene fragment in the first interaction network with the evolutionary trend of the gene combination in the second interaction network to form a global gene interaction network. This network not only contains the variation evolution information of a single gene fragment, but also integrates the interactions and synergistic effects between multiple gene fragments. Through the fused gene interaction network, the variation trend of each targeted gene fragment and its impact on drug resistance can be comprehensively evaluated, further revealing the complex mechanism of drug resistance development.

[0044] In summary, the gene interaction network reflects the variation of targeted gene fragments in different treatment cycles and their interactions, providing an important theoretical basis and data support for drug resistance prediction.

[0045] S2: Using the gene interaction network as a basic database, supervise the training of a drug resistance detector based on a self-attention mechanism, wherein the drug resistance detector includes a recognition unit and a detection unit.

[0046] Specifically, the gene interaction network is composed of the gene fragments obtained in the aforementioned analysis and clustering steps and the interaction relationships between them, and contains a large amount of gene variation information and its relationship with drug resistance. It is used as the basic database, that is, the basis for drug resistance decision-making, for supervising the training of drug resistance detectors.

[0047] Based on the sample training method, the model parameters are optimized by using labeled training data, that is, known gene mutations and drug resistance results, to achieve accurate prediction. In this process, the self-attention mechanism is introduced as a training method. In the embodiment of the present application, the goal of self-attention recognition is to mine the determined targeted gene fragments, so that the model can better focus on the gene fragments that have a significant impact on drug resistance.

[0048] Preferably, in the self-attention mechanism, the model not only pays attention to the currently input gene fragment information, but can also adaptively learn the relationships and dependencies between the various gene fragments to improve the accuracy of drug resistance prediction.

[0049] Among them, the drug resistance detector includes an identification unit and a detection unit. The main function of the identification unit is to extract features from the input genetic data, especially to identify key gene fragments related to drug resistance, that is, targeted gene fragments that have been identified in the aforementioned gene interaction network. They may mutate during drug treatment and thus affect the development of drug resistance.

[0050] The recognition unit then performs further analysis based on the features extracted by the recognition unit, evaluates the relationship between these gene variations and drug resistance, and gives a prediction result. Through this process, the drug resistance detector can automatically identify gene fragments related to drug resistance based on the user's gene data, and analyze them in combination with the relationship in the gene interaction network, and finally obtain a drug resistance prediction result.

[0051] Further, supervised training of a drug resistance detector based on a self-attention mechanism, step S2 of this application includes: The self-attention mechanism is configured based on the identification of the X targeted gene fragments; a recognition unit is constructed based on the self-attention mechanism, a detection unit is constructed based on the gene interaction network, and the detection unit is post-connected to the recognition unit to generate a drug resistance detector; sample data is called to perform sample-driven training on the drug resistance detector until convergence.

[0052] In an embodiment of the present application, the X targeted gene fragments refer to a set of gene fragments that have significant drug resistance correlation after screening in the aforementioned gene interaction network analysis. The X targeted gene fragments are used as the target for self-attention identification, and the weights are determined according to the strong correlation with drug resistance, and the self-attention mechanism is configured. As a deep learning technology, the self-attention mechanism can automatically adjust the identification targets and weights according to the importance of different parts of the input data, thereby enhancing the model's attention to important features. In the implementation of this application, the self-attention mechanism will help automatically identify which gene fragments are more critical to drug resistance prediction, thereby improving the pertinence and accuracy of subsequent analysis.

[0053] Next, a recognition unit is constructed based on the self-attention mechanism. Through the self-attention mechanism, features are extracted from the input gene data, especially the extraction of targeted gene fragments in the gene data and adaptive weighted processing, focusing on the variation information related to drug resistance. The detection unit judges and analyzes drug resistance based on the features extracted by the aforementioned recognition unit and the interaction relationship between each gene fragment in the gene interaction network. The structure of the gene interaction network provides information on how gene fragments affect each other, and the detection unit uses this information to further improve the analysis process of drug resistance prediction.

[0054] The detection unit is post-connected to the recognition unit to realize step-by-step processing and analysis of information, so that the recognition unit can first extract the characteristics of the gene fragments, and the detection unit makes a comprehensive judgment based on these characteristics and the gene interaction network, thereby forming a complete architecture of the drug resistance detector, and further training the convergence degree of the detector through sample training to improve the detection accuracy and generalization ability of the detector.

[0055] Specifically, sample data is called to perform sample-driven training on the drug resistance detector until convergence. By using actual sample data, i.e. known gene variation data and corresponding drug resistance data, sample-driven training refers to inputting a large amount of labeled sample data so that the model learns how to identify and analyze variation information related to drug resistance from gene data. The goal of the training process is to continuously adjust the parameters of the model so that the model can achieve the optimal prediction result on the given test data until the training process converges, i.e. the prediction accuracy of the model reaches the expected level or no longer improves significantly.

[0056] In summary, the drug resistance detector based on the self-attention mechanism can accurately identify gene fragments related to drug resistance, and conduct comprehensive analysis by combining gene interaction network information, ultimately providing accurate drug resistance prediction results.

[0057] S3: Receive the data submitted by the user, and perform gene sequencing by controlling the sequencer to determine the user's gene sequence, which includes multiple parallel tests.

[0058] The user submitted data is sample data from the user, usually blood, tissue or other biological samples. These samples contain the user's genomic information, and the quality and type of the samples directly affect the accuracy of subsequent gene sequencing. In this solution, the submitted data refers to biological samples that have been properly collected and processed, and these samples contain the genetic information to be analyzed.

[0059] Next, the gene sequencing test is performed by controlling the sequencer. In this step, the sequencer is used to perform genome sequencing on the sample submitted by the user. Specifically, gene sequencing refers to reading the gene sequence in the sample through modern high-throughput sequencing technology to obtain the user's genome information. Gene sequencing detection generates corresponding genome data by cutting the gene sample into small fragments and reading the base sequences of these fragments one by one. This process can be whole genome sequencing or targeted gene sequencing, and the specific sequencing strategy depends on the analysis requirements. In this process, precise control of the sequencer is very important to ensure high-quality data collection and provide accurate data input for subsequent genetic analysis.

[0060] Then, the user's gene sequence is determined. The raw data obtained through gene sequencing needs to be further processed and analyzed to obtain an accurate gene sequence. Preferably, the data provided by the sequencer is spliced, aligned and corrected to obtain a complete gene sequence that conforms to the user's genome characteristics, which is the basis for subsequent analysis.

[0061] Since errors may exist in the actual gene sequencing process, and the embodiment of the present application has high requirements for the accuracy of gene sequencing, multiple parallel detections are used to perform sequence proofreading and splicing and recombination on multiple sequencing results to ensure the accuracy of the sequencing data.

[0062] Furthermore, the user gene sequence finally obtained will become the basic data for subsequent drug resistance prediction and gene variation analysis.

[0063] Further, such as Figure 2 As shown, by controlling the sequencer to perform gene sequencing detection to determine the user's gene sequence, step S3 of this application includes: The sequencer is controlled to perform parallel detection on the data submitted by the user to determine multiple gene detection sequences; the multiple gene detection sequences are mapped and proofread, and sequence splicing and reconstruction are performed to determine the user gene sequence; wherein the maximum proportion of mapped gene fragments is used for screening and splicing, and the mapped gene fragments are the same gene fragments in each gene detection sequence.

[0064] First, the sequencer is controlled to perform multiple parallel gene tests on the data submitted by the user. Parallel testing refers to multiple gene sequence determinations on the data submitted by the user to improve the coverage and accuracy of the sequencing data. Through parallel sequencing, the error of a single sequencing can be reduced and the reliability of the gene data can be enhanced.

[0065] Multiple sequencing results generated by parallel detection will generate multiple gene detection sequences. Each gene detection sequence represents the complete gene sequence read by the sequencer from the user's test data.

[0066] The multiple gene detection sequences are further proofread to determine whether the parallel detection results are consistent. If they are consistent, it indicates that the detection is accurate, and any gene detection sequence is taken as the user gene sequence. If they are inconsistent, it indicates that there may be an error in the gene detection result, and the accurate gene fragments are further determined through mapping proofreading of gene fragments, and the determined accurate gene fragments are spliced ​​based on the gene sequence order as the user gene sequence.

[0067] During the specific implementation, the multiple gene detection sequences are traversed, and the starting gene fragments of each gene detection sequence are extracted as a group of mapping gene fragments. They are checked against each other to see if they are consistent. If there is any inconsistency, the gene fragment with the largest number in the group of gene fragments, i.e., the largest proportion item, is taken as the final accurate starting gene fragment. Similarly, each detection fragment in the sequence is proofread and analyzed to determine multiple accurate gene fragments, and the fragments are spliced ​​and reconstructed according to the fragment combination order of the gene detection sequence as the user gene sequence.

[0068] Through the above steps, the user's gene sequence finally obtained can more accurately reflect the user's genetic characteristics and provide reliable genetic information support for subsequent drug resistance prediction.

[0069] S4: The user gene sequence is transmitted to the drug resistance detector, self-attention recognition based on the targeted gene fragment and judgment based on the gene interaction network are performed to determine the variation trend of the targeted fragment, and the drug resistance prediction data under the time series is determined according to the variation trend.

[0070] Specifically, the user gene sequence is transmitted to the drug resistance detector for self-attention recognition based on targeted gene fragments and judgment based on gene interaction networks. Specifically, the drug resistance detector first uses the self-attention mechanism to identify the input user gene sequence, automatically identifies the targeted gene fragments that are most critical for drug resistance prediction, focuses on the important parts of the gene sequence related to drug resistance, and can handle complex dependencies in long sequences, identify the synergistic effects between different gene fragments, and thus help judge the trend of gene variation.

[0071] Specifically, according to the periodic variation trend in the gene interaction network, matching analysis is performed on the determined targeted gene fragments to determine the variation trend analysis under the periodic time sequence. The progress of gene variation may directly affect the efficacy of the drug.

[0072] Then, the drug resistance prediction data in the time series is determined according to the variation trend. According to the variation trend of the targeted gene fragment, the drug resistance detector displays the prediction results in the time series. Specifically, the linear relationship between gene variation and drug resistance is used to predict the drug resistance trend that may appear in the future.

[0073] Through the above steps, the drug resistance prediction data finally generated can reflect the temporal gene variation trend based on the user's submitted data and the future change trend of drug resistance.

[0074] Further, to determine the variation trend of the targeted fragment, step S4 of the present application includes: The user gene sequence is transmitted to the identification unit to identify the targeted gene fragment set; the targeted gene fragment set is transmitted to the detection unit to perform gene interaction variation analysis based on targeted drugs to determine the variation trend; wherein, if the number of the targeted gene fragments is 1, the first interaction network is used as the analysis standard; if the number of the targeted gene fragments is greater than 1, the second interaction network is used as the analysis standard.

[0075] Specifically, the user gene sequence is transmitted to the identification unit, and the targeted gene fragments in the gene sequence of the user's test data are analyzed and identified, that is, the gene fragments related to the targeted drugs suitable for lung cancer patients. These gene fragments are the key part of drug resistance analysis.

[0076] Once the recognition unit completes the extraction of the targeted gene fragments, the resulting set of targeted gene fragments is transmitted to the detection unit to analyze their variation trends, analyze the pattern of gene variation through the gene interaction network, and then infer the potential impact of the targeted gene fragments in drug treatment.

[0077] In this process, a gene interaction variation analysis based on targeted drugs is performed to determine the variation trend. The detection unit evaluates the variation trend of the targeted gene fragment through the gene interaction network. The gene variation trend refers to the change of the gene fragment at different time nodes or different treatment cycles.

[0078] Specifically, for the identified targeted gene fragments, matching analysis is performed in combination with the gene interaction network. Among them, if the number of the targeted gene fragments is 1, and only one gene fragment in the user's gene sequence will undergo mutation based on the targeted drug, then the detection unit will choose to use the first interaction network for analysis. The first interaction network is usually an interactive relationship network between a single targeted gene and its related gene fragments, which is used to analyze the variation trend of a single gene fragment and its relationship with drug tolerance.

[0079] If the number of the targeted gene fragments is greater than 1, that is, when the number of identified targeted gene fragments is greater than 1, the detection unit will perform analysis based on the second interaction network. The second interaction network involves multiple targeted gene fragments and the interaction relationship between them, and is suitable for processing complex interactions between multiple gene fragments. This network can provide a more comprehensive perspective and help more accurately predict the comprehensive impact of multiple gene variations on drug tolerance.

[0080] Through the above steps, the user's personalized genetic information is used to determine the appropriate interaction network for detailed analysis, and ultimately provide prediction data on the trend of targeted gene variation and its impact on drug tolerance.

[0081] Further, according to the variation trend, the drug resistance prediction data in the time series is determined, and step S4 of the present application includes: Identify the linear relationship between gene variability and drug resistance; based on the linear relationship, measure the drug resistance trend of the variation trend to determine the drug resistance prediction data.

[0082] In the embodiments of the present application, the linear relationship between the gene variability and drug resistance is the relationship between the gene variability (i.e., the degree of variation of the gene sequence or gene fragment) and drug resistance. The linear relationship is determined when the gene interaction network mining is performed as described above. According to the variation trend, the gene variability of each node in the time series is determined, which reflects the variation of the gene such as mutation, insertion, deletion, etc. Drug resistance refers to the resistance of cells or microorganisms to drug treatment. Based on the linear relationship, the variation trend is measured for drug resistance trend.

[0083] Specifically, according to the specific degree of gene variation (such as the number, type, frequency, etc. of gene variation), combined with the known linear relationship between gene variation and drug resistance, the variation trend of each sample is predicted. The specific value of gene variation is mapped to the corresponding drug resistance value, that is, the future drug resistance trend is quantitatively analyzed through the established linear relationship to determine the drug resistance prediction data.

[0084] Through the above steps, the system can quantitatively evaluate and predict the development trend of drug resistance based on the linear relationship between gene variability and drug resistance, provide strong forward-looking data support for follow-up, and guide the formulation of personalized treatment plans for patients.

[0085] The present application provides a lung cancer resistance prediction method based on user feature analysis, which has the following technical effects: 1. Combining gene variation with drug resistance analysis, by constructing a gene interaction network and combining targeted drug information, exploring the variation trend of gene fragments, and then analyzing the root cause of drug resistance. By performing multiple periodic analyses of the genomic data of multiple samples, it is possible to track the evolution trend of variant genes, identify the correlation between variant genes and drug resistance, and provide accurate drug resistance assessment. At the same time, by combining the linear relationship between gene variation and drug resistance, it is possible to accurately predict the development of drug resistance that patients may encounter during medication.

[0086] 2. In traditional drug resistance detection methods, there is often a lack of in-depth analysis of complex gene interactions. By introducing a drug resistance detector based on a self-attention mechanism, it can automatically learn the variation rules of gene sequences and accurately identify the complex relationship between gene fragments and drug resistance in combination with gene interaction networks. Through the self-attention mechanism, the system can focus on important gene fragments that affect drug resistance and improve the accuracy and reliability of predictions. At the same time, through cluster analysis and targeted gene fragment screening, the number of gene fragments that need to be paid attention to is reduced, the computational efficiency is improved, and the complexity of processing is reduced.

[0087] 3. Taking into account the detection error of the sequencer, multiple parallel sequencing methods are used to proofread and reorganize the sequencing results to ensure the accuracy of the genetic testing submitted by the user, to accurately measure the user's personalized characteristics, and to ensure the accuracy of subsequent drug resistance predictions.

[0088] In summary, it provides a precise scientific basis for drug resistance prediction, which can effectively improve the accuracy of drug resistance prediction based on user personalized characteristics.

[0089] Embodiment 2: Based on the same inventive concept as the lung cancer drug resistance prediction method based on user feature analysis in the above embodiment, Figure 3 As shown, the present application provides a lung cancer resistance prediction system based on user feature analysis, the system comprising: A mining module 11, the mining module 11 is used to call the biological gene library to mine the gene interaction network based on the statistical analysis of gene variation of the targeted drug, wherein the gene interaction network is constructed by the interaction between the targeted drug and the gene fragment; A construction module 12, wherein the construction module 12 is used to supervise the training of a drug resistance detector based on a self-attention mechanism using the gene interaction network as a basic database, wherein the drug resistance detector includes a recognition unit and a detection unit; A sequencing module 13, which is used to receive the data submitted by the user, and to control the sequencer to perform gene sequencing to determine the user's gene sequence, including multiple parallel tests; Prediction module 14, the prediction module 14 is used to transmit the user gene sequence to the drug resistance detector, perform self-attention recognition based on the targeted gene fragment and judgment based on the gene interaction network, determine the variation trend of the targeted fragment, and determine the drug resistance prediction data under the time series according to the variation trend.

[0090] Among them, the mining module 11 is also used to perform the following steps: based on the biological gene library, extract the first sample genome, wherein the first sample genome contains gene sequences of multiple period time nodes under the medication cycle; traverse the first sample genome for comparison, identify variant gene fragments, and determine the first targeted gene fragment group; complete the extraction and determination of the Nth sample genome, and determine the Nth targeted gene fragment group; perform statistical analysis on the first targeted gene fragment group to the Nth targeted gene fragment group, and construct the gene interaction network, wherein the gene interaction network identifies a linear relationship between gene variability and drug resistance.

[0091] The mining module 11 is further used to perform the following steps: traverse the first targeted gene fragment group to the Nth targeted gene fragment group, perform cluster analysis, and determine M gene fragment clusters, where M is the number of distinguished targeted gene fragments; traverse the M gene fragment clusters, and screen and determine X targeted gene fragments based on the preset number of clusters, where X is a positive integer less than or equal to M, and the X targeted gene fragments are gene fragments with strong correlation with drug resistance; and construct the gene interaction network for the X targeted gene fragments.

[0092] The mining module 11 is further used to perform the following steps: traverse the X targeted gene fragments, and for a single targeted gene fragment, traverse the sample genome to mine the variation evolution trend under periodic time sequence to determine a first interaction network; for a combined targeted gene fragment, traverse the sample genome to mine the variation evolution trend under periodic time sequence to determine a second interaction network; and fuse the first interaction network and the second interaction network to determine the gene interaction network.

[0093] The sequencing module 13 is further used to execute the following steps: control the sequencer to perform parallel detection on the user submitted data to determine multiple gene detection sequences; map and proofread the multiple gene detection sequences, perform sequence splicing and reconstruction, and determine the user gene sequence; wherein the maximum proportion of mapped gene fragments is used for screening and splicing, and the mapped gene fragments are the same gene fragments in each gene detection sequence.

[0094] The construction module 12 is further used to perform the following steps: configuring the self-attention mechanism based on the identification of the X targeted gene fragments; constructing a recognition unit based on the self-attention mechanism, constructing a detection unit based on the gene interaction network, and post-connecting the detection unit to the recognition unit to generate a drug resistance detector; calling sample data to perform sample-driven training on the drug resistance detector until convergence.

[0095] The prediction module 14 is further used to perform the following steps: transmitting the user gene sequence to the identification unit to identify the targeted gene fragment set; transmitting the targeted gene fragment set to the detection unit to perform gene interaction variation analysis based on targeted drugs to determine the variation trend; if the number of the targeted gene fragments is 1, the first interaction network is used as the analysis standard; if the number of the targeted gene fragments is greater than 1, the second interaction network is used as the analysis standard.

[0096] The prediction module 14 is further configured to perform the following steps: identifying the linear relationship between gene variability and drug resistance; and based on the linear relationship, performing drug resistance trend measurement on the variation trend to determine the drug resistance prediction data.

[0097] Through the above detailed description of a method for predicting lung cancer resistance based on user feature analysis in this specification, those skilled in the art can clearly understand a method for predicting lung cancer resistance based on user feature analysis in this embodiment. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0098] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting lung cancer drug resistance based on user feature analysis, characterized in that: The method comprises: Calling a biological gene library to mine a gene interaction network based on statistical analysis of gene variation based on targeted drugs, wherein the gene interaction network is constructed based on the interaction between targeted drugs and gene fragments; Using the gene interaction network as a basic database, supervised training of a drug resistance detector based on a self-attention mechanism, wherein the drug resistance detector includes a recognition unit and a detection unit; Receive the data submitted by the user, perform gene sequencing test by controlling the sequencer, and determine the user's gene sequence, which includes multiple parallel tests; The user gene sequence is transmitted to the drug resistance detector, self-attention recognition based on the targeted gene fragment and judgment based on the gene interaction network are performed to determine the variation trend of the targeted fragment, and the drug resistance prediction data under the time series is determined according to the variation trend.

2. The method for predicting lung cancer drug resistance based on user feature analysis according to claim 1, characterized in that: Mining gene interaction networks through statistical analysis of gene variation based on targeted drugs, including: Based on the biological gene library, extracting a first sample genome, wherein the first sample genome includes gene sequences at multiple period time nodes under a medication cycle; Traversing the first sample genome for comparison, identifying variant gene segments, and determining a first targeted gene segment group; Complete the extraction and determination of the Nth sample genome and determine the Nth targeted gene fragment group; Statistical analysis is performed on the first targeted gene fragment group to the Nth targeted gene fragment group to construct the gene interaction network, wherein the gene interaction network identifies a linear relationship between gene variability and drug resistance.

3. The method for predicting lung cancer drug resistance based on user feature analysis according to claim 2, characterized in that: Performing statistical analysis on the first targeted gene segment group to the Nth targeted gene segment group comprises: Traversing the first targeted gene segment group to the Nth targeted gene segment group, performing cluster analysis, and determining M gene segment clusters, where M is the number of distinct targeted gene segments; Traversing the M gene fragment clusters, screening and determining X targeted gene fragments according to the preset number of clusters, wherein X is a positive integer less than or equal to M, and the X targeted gene fragments are gene fragments with strong correlation with drug resistance; The gene interaction network is constructed for the X targeted gene fragments.

4. The method for predicting lung cancer drug resistance based on user feature analysis according to claim 3, characterized in that: For the X targeted gene fragments, constructing the gene interaction network includes: Traversing the X targeted gene fragments, for a single targeted gene fragment, traversing the sample genome to mine the variation evolution trend under periodic time sequence, and determining a first interaction network; For the combined targeted gene fragments, the sample genome is traversed to mine the variation evolution trend under periodic time sequence and determine the second interaction network; The first interaction network and the second interaction network are integrated to determine the gene interaction network.

5. The method for predicting lung cancer drug resistance based on user feature analysis according to claim 1, characterized in that: By controlling the sequencer to perform gene sequencing detection, the user's gene sequence is determined, including: Controlling the sequencer to perform parallel testing on the data submitted by the user to determine multiple gene testing sequences; Map and proofread the multiple gene detection sequences, perform sequence splicing and reconstruction, and determine the user gene sequence; Among them, the largest proportion of mapped gene fragments is used for screening and splicing, and the mapped gene fragments are the same gene fragments in each gene detection sequence.

6. The method for predicting lung cancer drug resistance based on user feature analysis according to claim 3, characterized in that: Supervised training of drug resistance detector based on self-attention mechanism, including: configuring the self-attention mechanism based on the identification of the X target gene fragments; Constructing a recognition unit based on the self-attention mechanism, constructing a detection unit based on the gene interaction network, and post-connecting the detection unit to the recognition unit to generate a drug resistance detector; The sample data is called to perform sample-driven training on the drug resistance detector until convergence.

7. The method for predicting lung cancer drug resistance based on user feature analysis according to claim 4, characterized in that: Determine the variation trend of the targeted fragment, including: Transmitting the user gene sequence to the identification unit to identify the targeted gene fragment set; Transmitting the targeted gene fragment set to the detection unit, performing gene interaction variation analysis based on targeted drugs, and determining the variation trend; If the number of the targeted gene fragments is 1, the first interaction network is used as the analysis standard; if the number of the targeted gene fragments is greater than 1, the second interaction network is used as the analysis standard.

8. The method for predicting lung cancer drug resistance based on user feature analysis according to claim 2, characterized in that: Determine the drug resistance prediction data in the time series according to the variation trend, including: Identify the linear relationship between genetic variability and drug resistance; Based on the linear relationship, the variation trend is measured for drug resistance trend to determine the drug resistance prediction data.

9. A lung cancer drug resistance prediction system based on user feature analysis, characterized in that: Executing a method for predicting lung cancer drug resistance based on user feature analysis according to any one of claims 1 to 8, the system comprising: A mining module, the mining module is used to call the biological gene library to mine the gene interaction network based on the statistical analysis of gene variation of targeted drugs, wherein the gene interaction network is constructed by the interaction between targeted drugs and gene fragments; A construction module, wherein the construction module is used to supervise the training of a drug resistance detector based on a self-attention mechanism using the gene interaction network as a basic database, wherein the drug resistance detector includes a recognition unit and a detection unit; A sequencing module, which is used to receive data submitted by users, and to control a sequencer to perform gene sequencing to determine the user's gene sequence, including multiple parallel tests; A prediction module is used to transmit the user gene sequence to the drug resistance detector, perform self-attention recognition based on the targeted gene fragment and judgment based on the gene interaction network, determine the variation trend of the targeted fragment, and determine the drug resistance prediction data under the time series according to the variation trend.

Images