Zero-code biological information analysis method and device based on multi-agent collaboration

Abstract

The invention provides a zero-code biological information analysis method and device based on multi-agent collaboration, and the method comprises the steps: setting a large language model as a plurality of agents through cue words, carrying out the scoring and rearrangement of tools in a tool library through the plurality of agents, and finding an optimal matching tool through discussion; exploring type assembly is carried out by utilizing intelligent agents of a plurality of explorers according to input and output parameters of different tools, fusion is carried out after multiple rounds of tests are successful, a workflow is established, the workflow is written into codes, and the codes are executed to obtain an analysis result; checking and screening the analysis result to obtain screening information; presenting the screening information in a graphic mode to obtain graphic information; and obtaining candidate answers of the to-be-analyzed task according to the screening information and the graphic information.

Description

Technical Field

[0001] This invention relates to the field of biomedical technology, and in particular to a multi-agent collaborative zero-code bioinformatics analysis method and apparatus based on a large language model. Background Technology

[0002] In current life science and medical research, researchers are facing an unprecedented surge in data volume. The rapid development of high-throughput sequencing technology and the application of emerging biotechnologies have enabled researchers to acquire vast amounts of biological data more quickly and economically. At the same time, the significant increase in sample size has also led to a dramatic increase in the demand for data management and analysis. This data is not only massive in quantity but also diverse in type, including genomic data, transcriptomic data, proteomic data, and metabolomic data, which are often referred to as multi-omics or multimodal data.

[0003] The accumulation of multi-omics and multimodal biological data has provided researchers with unprecedented opportunities, such as in disease diagnosis, new drug development, and biological evolution. However, it also brings significant challenges, particularly in data storage, management, and complex analysis. Effectively utilizing this data requires the ability to perform complex bioinformatics analyses, which typically involve a series of steps, including but not limited to data preprocessing, quality control, statistical analysis, machine learning modeling, and the visualization and interpretation of results.

[0004] The complexity of steps in bioinformatics analysis workflows makes planning, optimizing, and reusing analysis workflows more difficult. To reduce the complexity and technical barriers of bioinformatics analysis, the open-source community and commercial companies have developed online analysis platforms such as Galaxy and SevenBridge. Galaxy is a web application platform that provides researchers without a computer programming background with a visual graphical user interface (GUI). Through this interface, users can use mouse clicks and drag-and-drop operations to select, assemble, and run various analysis tools, thereby creating complex analysis workflows. A core feature of the Galaxy platform is that it provides a rich set of tools and predefined workflow templates. Users can choose analysis workflows suitable for their analytical needs and initiate analyses through simple graphical operations, which greatly simplifies the construction and execution of bioinformatics analysis workflows.

[0005] Galaxy and other predefined workflow bioinformatics analysis platforms reduce the complexity for users when building and executing complex analytical workflows through templated processes. This is particularly beneficial for non-expert users in the bioinformatics field, who do not need in-depth understanding of command-line operations or scripting to complete analyses. Furthermore, Galaxy supports sharing and reusing workflows, making bioinformatics analysis more standardized and reproducible.

[0006] Multi-omics and multimodal data have created a demand for complex bioinformatics analyses, involving the sequential execution of multiple bioinformatics steps, including but not limited to data preprocessing, quality control, statistical analysis, machine learning modeling, and result visualization and interpretation. Platforms with predefined workflows, such as Galaxy, serve as important tools for processing and analyzing large volumes of omics data. These platforms simplify the creation and execution of analytical workflows by providing graphical user interfaces (GUIs), particularly beneficial for researchers without a computer programming background. However, despite significantly improving user experience and efficiency, these platforms still have some technical limitations:

[0007] 1. Limitations of predefined processes

[0008] While templated predefined processes can greatly simplify workflow creation, they are often inflexible and struggle to adapt to highly personalized analytical needs. When faced with specific, non-standardized analytical tasks, researchers may find that existing templates are insufficient to support their needs, limiting the possibility of innovative and personalized analysis and hindering the ability to quickly adapt to the ever-changing and evolving demands of scientific research.

[0009] 2. Limitations in handling interface complexity

[0010] While the platform offers some visual aids to help users build workflows, effectively connecting multiple different analysis steps, especially when complex data format conversions are required between different tools, remains a challenge. When the analysis process involves numerous conditional statements, iterative processing, or dynamic parameter adjustments, operating through a GUI becomes less convenient and efficient.

[0011] 3. Lack of integration between downstream explanations and literature review.

[0012] A key aspect of bioinformatics data analysis is integrating analytical results with existing knowledge bases and literature resources to facilitate understanding, interpretation, and summarization. Current predefined workflow platforms typically focus on the analytical process itself without providing tools to link to external knowledge bases or automatically aggregate analytical results with relevant literature to generate summary results.

[0013] While platforms like Galaxy have made significant contributions to lowering technical barriers and improving efficiency, complex bioinformatics analysis needs to focus more on increasing the flexibility of workflow assembly, simplifying workflow assembly operations, and better integrating with downstream analytical interpretation tools. This will enable researchers to better analyze and mine the rapidly growing body of biological data, driving progress in life sciences and medicine. Summary of the Invention

[0014] The inventors addressed the shortcomings of existing bioinformatics analysis platforms that rely on workflow assembly and execution. These platforms suffer from limitations in predefined workflows, complex user interfaces, and a lack of downstream interpretation and literature integration, resulting in inefficiencies in handling complex and personalized analysis needs. This invention proposes a multi-agent collaborative bioinformatics analysis tool selection and workflow adaptive assembly method. Multiple agents collaboratively score and rearrange tools in a toolkit, identifying the optimal matching tool through discussion. Multiple explorer agents then perform exploratory assembly based on the input and output parameters of different tools. After multiple successful tests, these agents are merged to form the components of a complex workflow.

[0015] like Figure 3 As shown, this invention proposes a zero-code bioinformatics analysis method based on multi-agent cooperation, including:

[0016] Step 1 of setting up multiple agents involves setting up the large language model as multiple agents using prompt words. These agents include a requirements analyst, selector, and planner for developing workflow plans; a programmer, scheduler, and executor for executing workflows; a data analyst, plotter, and biological expert for interpreting the execution results biologically; and an evidence-based data analyst, medical plotter, and medical expert for providing medical insights based on the biological interpretation results.

[0017] Workflow planning step 2 is used to acquire bioinformatics and the tasks to be analyzed. The requirements analyst determines the output format of the tasks to be analyzed. The selector selects the bioinformatics analysis tool that is closest to the requirements of the tasks to be analyzed from the tool library according to the description of each tool. The planner develops the workflow based on the selected bioinformatics analysis tool.

[0018] In step 3 of the process execution, the programmer writes the workflow into work code. After the scheduler arranges and optimizes the execution order of the work code, the executor executes the optimized work code and obtains the analysis results.

[0019] In biological interpretation step 4, the data analyst checks and filters the analysis results to obtain filtering information; the plotter presents the filtering information graphically to obtain graphical information; and the biologist obtains candidate answers for the task to be analyzed based on the filtering information and the graphical information.

[0020] In step 5 of the medical insight process, the evidence-based data analyst finds corresponding medical evidence and clinical data for the candidate answer based on clinical trials, research papers, and drug approval information. The medical illustrator then displays the candidate answer in chart form, resulting in a graphical answer. The candidate answer and the graphical answer are then input into the medical expert, who assigns a priority to the candidate answer based on the medical evidence and clinical data, and outputs the recommended answer and its priority as the analysis result for the task to be analyzed.

[0021] The zero-code bioinformatics analysis method based on multi-agent collaboration is described in which the bioinformatics and the task to be analyzed are respectively gene detection data and analysis of which genetic disease the gene detection data has; or tumor detection data and analysis of which targeted drugs the tumor detection data is suitable for.

[0022] The aforementioned zero-code bioinformatics analysis method based on multi-agent cooperation further includes:

[0023] Debugger, used to correct errors in the workflow;

[0024] The monitor is used to track the intermediate execution results of each stage of the working code to ensure the correct execution of the working code;

[0025] The inquirer raises questions and hypotheses about the analysis results based on a specified knowledge base, focusing on pathways, cells, and genes. The data analyst then verifies and filters the analysis results based on these questions and hypotheses.

[0026] The aforementioned zero-code bioinformatics analysis method based on multi-agent collaboration includes the following workflow planning steps: constructing a tool library from the background knowledge of each bioinformatics analysis tool; performing a corresponding retrieval for the task to be analyzed; obtaining bioinformatics analysis tools and their descriptions that meet the requirements of the task to be analyzed; and providing them as prompt words to the selector based on a large language model to enhance knowledge.

[0027] like Figure 4 As shown, this invention also proposes a zero-code bioinformatics analysis device B based on multi-agent collaboration, which includes:

[0028] The multi-agent setting module 100 sets the large language model as multiple agents through prompt words. These multiple agents include a requirements analyst, selector, and planner for developing workflow plans; a programmer, scheduler, and executor for executing workflows; a data analyst, plotter, and biological expert for biologically interpreting the execution results; and an evidence-based data analyst, medical plotter, and medical expert for providing medical insights from the biological interpretation results.

[0029] The workflow planning module 200 is used to acquire bioinformatics and the tasks to be analyzed. The requirements analyst determines the output format of the tasks to be analyzed. The selector selects the bioinformatics analysis tool that is closest to the requirements of the tasks to be analyzed from the tool library according to the description of each tool. The planner formulates the workflow based on the selected bioinformatics analysis tool.

[0030] In the process execution module 300, the programmer writes the workflow as work code, the scheduler arranges and optimizes the execution order of the work code, and then the executor executes the optimized work code to obtain the analysis results;

[0031] In the biological interpretation module 400, the data analyst checks and filters the analysis results to obtain filtering information; the plotter presents the filtering information graphically to obtain graphical information; and the biologist obtains candidate answers for the task to be analyzed based on the filtering information and the graphical information.

[0032] In the Medical Insights module 500, the evidence-based data analyst finds corresponding medical evidence and clinical data for the candidate answer based on clinical trials, research papers, and drug approval information; the medical illustrator displays the candidate answer in chart form to obtain a graphical answer; the candidate answer and the graphical answer are input into the medical expert, who assigns a priority to the candidate answer based on the medical evidence and clinical data, and outputs the recommended answer and its priority as the analysis result for the task to be analyzed.

[0033] The aforementioned zero-code bioinformatics analysis device based on multi-agent collaboration, wherein the bioinformatics and the task to be analyzed are respectively gene detection data and analysis of which genetic disease the gene detection data has; or tumor detection data and analysis of which targeted drugs the tumor detection data is suitable for.

[0034] The aforementioned zero-code bioinformatics analysis device based on multi-agent collaboration further includes, among other agents:

[0035] Debugger, used to correct errors in the workflow;

[0036] The monitor is used to track the intermediate execution results of each stage of the working code to ensure the correct execution of the working code;

[0037] The inquirer raises questions and hypotheses about the analysis results based on a specified knowledge base, focusing on pathways, cells, and genes. The data analyst then verifies and filters the analysis results based on these questions and hypotheses.

[0038] The aforementioned zero-code bioinformatics analysis device based on multi-agent collaboration includes a workflow planning module comprising: constructing a tool library from the background knowledge of various bioinformatics analysis tools; performing a corresponding retrieval for the task to be analyzed; obtaining bioinformatics analysis tools and their descriptions that meet the requirements of the task to be analyzed; and providing them as prompt words to the selector based on a large language model to enhance knowledge.

[0039] The present invention also proposes an electronic device, including the aforementioned zero-code bioinformatics analysis device based on multi-agent collaboration.

[0040] The electronic device is connected to an information display device, which displays the evaluation result using user-defined display parameters, attributes, or through an artificial intelligence model.

[0041] The present invention also proposes a storage medium for storing a computer program that executes the aforementioned zero-code bioinformatics analysis method based on multi-agent cooperation.

[0042] As can be seen from the above solutions, the advantages of the present invention are:

[0043] 1) Increased flexibility in workflow assembly and the upper limit of free exploration, covering more analytical needs: By having multiple agents collaboratively score and rearrange tools in the tool library, and perform exploratory assembly based on the input and output parameters of different tools, compared to traditional manual selection and assembly methods, this automated and intelligent approach can quickly find the most suitable combination for a specific task from a large number of tools. It can dynamically assemble the most suitable tool flow for each specific task, which not only greatly enhances the flexibility of the analytical workflow but also meets highly personalized analytical needs, covering more analytical requirements. The self-assembly mode of artificial intelligence has unlimited possibilities, raising the upper limit of workflow combination;

[0044] 2) This invention significantly simplifies the complexity of bioinformatics analysis and lowers the operational threshold. By introducing intelligent agents to automatically evaluate, select, and assemble analytical tools, it provides users with a simpler and more flexible way to create analytical workflows. No deep programming knowledge or bioinformatics background is required; task requirements are expressed through natural language commands. The collaborative work of the intelligent agents not only optimizes the tool selection process and improves analytical efficiency but also ensures the accuracy and adaptability of the final workflow through exploratory assembly and multiple rounds of testing. In short, this invention, through intelligent and automated analytical workflow design, provides researchers and clinicians with an easy-to-use and powerful bioinformatics analysis tool, greatly promoting the development of life science and medical research.

[0045] 3) Enhanced medical interpretability of analysis results: By combining the analysis results of multiple agents with downstream interpretation and literature tools, this invention can provide richer and more in-depth interpretations of analysis results. This approach not only helps researchers better understand the analysis results but also provides clinicians with more valuable information, contributing to advancements in medical research and clinical diagnosis.

[0046] In summary, this invention provides a novel and intelligent method for selecting bioinformatics analysis tools and adaptively assembling workflows. Compared with existing technologies, it not only improves analytical flexibility and covers a wider range of analytical needs while lowering the barrier to entry, but also enhances the medical interpretability of analytical results, making it a powerful tool for promoting the development of life science and medical research. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of a preferred embodiment of the present invention.

[0048] Figure 2 This is a flowchart of the overall method of the present invention;

[0049] Figure 3 This is a flowchart of the method of the present invention;

[0050] Figure 4 This is a block diagram of the device of the present invention;

[0051] Figure 5 This is a schematic diagram of the structure of the first electronic device of the present invention;

[0052] Figure 6 This is a schematic diagram of the application environment structure of the first electronic device of the present invention;

[0053] Figure 7 This is a schematic diagram of the structure of the second electronic device of the present invention;

[0054] Figure 8 This is a diagram illustrating the prompts for demand analysts.

[0055] Figure 9 This is a diagram illustrating the prompts for the selector;

[0056] Figure 10 A diagram illustrating the planner's prompts;

[0057] Figure 11 A diagram illustrating the prompts for programmers;

[0058] Figure 12 A diagram illustrating the prompts given to the executor;

[0059] Figure 13 This is a diagram illustrating the prompts for data entry personnel.

[0060] Figure 14A diagram illustrating the prompts used by the draftsman;

[0061] Figure 15 A diagram illustrating the prompts used by biology experts;

[0062] Figure 16 A diagram illustrating the prompts for evidence-based data analysts;

[0063] Figure 17 A diagram illustrating prompts for medical illustrators;

[0064] Figure 18 This is an illustration of prompts provided by medical experts.

[0065] Figure label:

[0066] 1-Multi-agent setup steps;

[0067] 2-Workflow planning steps;

[0068] 3. Process execution steps;

[0069] 4- Biological explanation steps;

[0070] 5-Steps to Medical Insight;

[0071] 6-Planner;

[0072] 7-Selector;

[0073] 8-Demand Analyst;

[0074] 9-Executor;

[0075] 10-Scheduler;

[0076] 11-Programmer;

[0077] 12-Biology expert;

[0078] 13-Draftsman;

[0079] 14-Data Clerk;

[0080] 15-Medical experts;

[0081] 16-Medical Draftsman;

[0082] 17-Medical Insights;

[0083] 100-Multi-agent setting module;

[0084] 200 - Workflow Planning Module;

[0085] 300 - Process Execution Module;

[0086] 400 - Biological Explanation Module;

[0087] 500-Medical Insights Module;

[0088] A - First electronic device;

[0089] B- A zero-code bioinformatics analysis device based on multi-agent collaboration;

[0090] C-Data acquisition equipment;

[0091] D-Information display device;

[0092] 1000 - Second electronic device;

[0093] Ⅰ-Computational Unit;

[0094] II-ROM;

[0095] III-RAM;

[0096] N-bus;

[0097] V-Interface;

[0098] VI - Input Unit;

[0099] VII - Output Unit;

[0100] VIII - Storage medium;

[0101] IX - Communication Unit. Detailed Implementation

[0102] It should be noted that the processor described in this invention is the control center of an electronic device. It can be a single processor or a collective term for multiple processing elements. For example, it can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of this invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0103] Alternatively, the processor can perform various functions of the electronic device by running or executing software programs stored in memory, and by calling data stored in memory.

[0104] In a specific implementation, as one example, the processor may include one or more CPUs. Each of these processors may be a single-core processor or a multi-core processor. Here, "processor" can refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions). Electronic devices may include servers, desktop computers, laptops, smartphones, tablets, embedded computers, etc., where the embedded computer includes vehicles and robots, etc.

[0105] The memory is used to store the software program that executes the solution of the present invention, and the execution is controlled by the processor. For specific implementation methods, please refer to the above method embodiments, which will not be repeated here.

[0106] It should be noted that the structure of the electronic device shown in the accompanying drawings of this invention does not constitute a limitation thereof. The actual knowledge structure recognition device may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0107] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0108] It should also be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0109] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0110] It should also be understood that, in various embodiments of the present invention, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0111] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0112] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0113] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0114] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0115] While conducting bioinformatics analysis research on multi-omics and multimodal data, the inventors discovered that existing bioinformatics analysis platforms, which rely on workflow assembly, are not efficient enough in handling complex and personalized analytical needs. This is due to limitations in predefined workflows, the complexity of the user interface, and the lack of downstream interpretation and literature integration. These limitations highlight the technical challenge proposed in this invention: how to provide a bioinformatics analysis platform that simultaneously satisfies high flexibility, ease of operation, and integration with downstream interpretation tools.

[0116] The inventors discovered that this deficiency can be addressed through a multi-agent collaborative approach. In this method, multiple agents work together, each scoring and ranking tools in a toolkit, collaboratively discussing and finding the best-matching tool. Next, explorer agents perform exploratory assembly based on the tool's input and output parameters. Through multiple rounds of testing and feedback adjustments, successful workflows are identified and merged, ultimately forming a complex yet efficient workflow. Multiple agents are then recruited to assemble and run this workflow according to their dependencies.

[0117] To make the above-mentioned features and effects of the present invention clearer and easier to understand, specific embodiments are described below in conjunction with the accompanying drawings. This specification discloses one or more embodiments incorporating the features of the present invention. The disclosed embodiments are merely illustrative. The scope of protection of the present invention is not limited to the disclosed embodiments, but is defined by the appended claims.

[0118] The overall system architecture of this invention is shown in the appendix. Figure 2As shown, the architecture of the entire system embodies a complex multidisciplinary collaborative system. Multiple agents are generated through a large language model and prompts. For example, the large language model could be GPT-4. Prompts could be phrased like, "If you were a tool selector, and your task is to analyze tumor detection data, and a tool is available to analyze the protein impact of tumor mutations, would you consider including this tool in your selection?" This sets the large language model as the tool selector, enabling it to perform this function. Thus, various roles (agents) collaborate throughout the entire process, from posing a question to deriving medical insights. Based on the large language model, it utilizes retrieval enhancement and Few-shot technology to generate different agents, each responsible for different functions. This breaks down a complex bioinformatics analysis into four main parts, involving the participation of different roles: effectively analyzing user input data and questions, and ultimately generating conclusions and reports, including various tables and visualizations.

[0119] Retrieval enhancement involves building a local knowledge base from the background knowledge of bioinformatics analysis tools. This allows for targeted retrieval based on the user's task, resulting in better access to tools and background knowledge relevant to the user's needs. This information is then provided as prompts to the large language model, thus enhancing its knowledge base. Few-shots provide examples to the large language model using prompts, essentially offering multiple examples as prompts.

[0120] 1) Workflow plan:

[0121] Requirement analysts define requirements and ensure that tools and processes are aligned with those requirements.

[0122] The character's definition code (hint) is as follows: Figure 8 As shown, for example, "You are playing the role of a requirements analyst. You have been assigned a {task}. Propose the steps to complete the task. In addition, provide guidelines for solving key problems and specify the development environment and programming language."

[0123] The tool manager selects the appropriate tool from a wide range of tools. The definition code for this role is as follows: Figure 9 For example, "You are a selector, and your task is to choose the most suitable tool for a specific task {task}. Based on guidelines provided by an analyst, you should provide a list of the tools you have selected. You should also describe your selection process, including the factors you considered and the criteria you used for decision-making. For example, you might consider functionality, performance, cost, usability, etc."

[0124] The Workflow Planner is responsible for the overall workflow design. The code defining this role is as follows: Figure 10 For example, "You are a workflow planner responsible for designing the overall workflow. Your task involves considering user task requests, requirements from requirements analysts, and the toolset selected by the toolset manager. Please describe your approach to designing the workflow, taking these inputs into account, and ultimately provide a combination of tools from the toolset."

[0125] 2) Execution:

[0126] Programmers write code for the workflow, ensuring the generation of code for the process. The code defining this role is as follows: Figure 11 For example, as shown below: "You are a programmer responsible for coding workflows to ensure efficient code generation. The workflow planner has designed the workflow as {workflow}. Your responsibility is to translate the workflow design into executable code. You should also follow the requirements analyst's {requirements} guidelines. Describe your approach to programming the workflow, considering the steps listed by the workflow planner. Explain how you structure the code, implement the logic, handle exceptions, and ensure scalability and maintainability. Additionally, detail any testing or debugging processes used to validate the functionality of the coded workflow."

[0127] The scheduler is responsible for task allocation and time management. Python code is used to return a Task ID after a task is submitted, which is used to track progress. The task pool can be dynamically adjusted.

[0128]

[0129]

[0130] The Executor runs the analysis process. The code defining this role is as follows: Figure 12 As shown below, for example: "You will play the role of an executor, responsible for running the analysis process based on the {code} provided by the programmer. Your responsibility is to execute the code accurately and efficiently to perform the expected analysis. Describe your method for running the analysis process, taking into account the steps listed in the programmer's code. Explain how you interpret and execute the code, handle any errors or exceptions, and ensure the validity and reliability of the analysis results."

[0131] 3) Biological interpretation:

[0132] The Curator integrates and filters information. The code defining this role is as follows: Figure 13As shown below, for example, "You will play the role of an administrator, responsible for integrating and filtering information based on the {tables and files} generated from the workflow. Your tasks include organizing and refining the data to ensure its accuracy, relevance, and usability. Describe your methods for managing information, taking into account the content and structure of the intermediate documents. Explain how you assess the quality of the data, identify any inconsistencies or errors, and apply filters or transformations to improve its consistency and usefulness. Furthermore, detail any strategies you employ throughout the process to maintain the integrity and comprehensiveness of the planned information."

[0133] The Visualizer presents data graphically to aid understanding. The role is defined as shown in code example 14: "You are playing the role of a visualization tool, responsible for converting data into graphical representations to aid understanding. Your responsibility is to create visualizations based on the {tables and files} generated from the workflow. Describe your methods for visualizing the data, taking into account the content and format of the generated tables and files. Explain how you chose appropriate visualization techniques, designed aesthetically pleasing graphics, and effectively communicated insights."

[0134] Biologists draw conclusions based on the above information. The definition code for this role is shown in Figure 15, for example: "As a biologist, your responsibility is to extract biological meaning and draw conclusions based on the {results} of the workflow, the {information} synthesized and curated by the curator, and the {visualizations} created by visualization tools. Describe your methods for analyzing results, considering the curated information and visual representations, and synthesizing them to obtain meaningful biological insights. Explain how you identify patterns, correlations, and trends in the data, and how you interpret these findings within the context of your biological expertise."

[0135] 4) Medical insights:

[0136] Evidence-Based Data Clerks (EBM Curators) collect and organize relevant medical evidence. The definition code for this role is as follows: Figure 16 For example, it might say, "You will serve as an evidence-based data officer, responsible for collecting and organizing relevant medical evidence. Your duties include searching for conclusions from literature and public databases to support the evidence-based medicine practices outlined in {Table}. Describe your methods for collecting and synthesizing evidence, considering various sources such as academic journals, clinical trials, and databases. Explain how you assess the quality and reliability of the collected evidence to ensure it meets the standards of evidence-based medicine."

[0137] The MedVisualizer role presents data in medically relevant graphical formats to aid understanding. The definition code for this role is as follows: Figure 17For example, "You are playing the role of a medical illustrator, responsible for converting data into graphical representations to aid understanding. Your task is to create visualizations based on {tables and files} generated from the workflow. Describe your methods for visualizing the data, taking into account the content and format of the generated tables and files. Explain how you chose appropriate visualization techniques, designed aesthetically pleasing graphics, and effectively communicated insights."

[0138] Physicians make clinical decisions. The definition code for this role is as follows: Figure 18 As shown, for example, "Your role is that of a medical expert responsible for making clinical decisions based on submitted {task} inputs, {medical evidence} collected and organized by evidence-based data analysts, and {visualizations} created by visualization tools. Your responsibility is to analyze the collected evidence and visual representations to inform your clinical decision-making process. Describe how you review medical evidence and graphical representations, considering their relevance, accuracy, and applicability to the patient's condition. Explain how you integrate the information provided by evidence-based data analysts and visualization tools into your decision-making process to ensure that your decisions are evidence-based and patient-centered."

[0139] Preferred, such as Figure 1 As shown, in addition to the basic intelligent agent described above, the present invention can also add intelligent agents to generate higher quality reports.

[0140] 1) Workflow plan:

[0141] The planner is responsible for designing the overall workflow.

[0142] Explorers work together to explore different tool assembly methods and strategies.

[0143] The selector selects the appropriate tool from a large number of tools.

[0144] Requirement analysts define requirements and ensure that tools and processes are aligned with those requirements.

[0145] 2) Execution:

[0146] The executor runs the analysis process.

[0147] The scheduler is responsible for task allocation and time management.

[0148] Programmers perform programming work to ensure the implementation of processes.

[0149] The debugger fixes errors in the program.

[0150] The monitor tracks and analyzes the results to ensure the correct execution of the process.

[0151] Testers A and B are responsible for the testing and ensuring the accuracy of the results.

[0152] 3) Biological interpretation:

[0153] Inquirers, drawing upon specialized knowledge bases in the life sciences, raise questions and hypotheses about pathways, cells, and genes, thereby driving in-depth research.

[0154] Curators integrate and filter information.

[0155] Visualizers present data graphically to aid understanding.

[0156] Senior biologists draw conclusions based on the above information.

[0157] 4) Medical insights:

[0158] Evidence curators collect and organize relevant medical evidence.

[0159] Evidence analysts (EBM analysts) conduct analysis based on evidence.

[0160] Senior Physicians make clinical decisions.

[0161] Visualizers also play a role in this section, visualizing information to assist in medical decision-making.

[0162] 2. Application Examples

[0163] Using the application of this system as an example, the following question illustrates the analysis process of the system:

[0164] User input question: You are analyzing the genetic testing data of a cancer patient. The patient's somatic mutation data is stored in the following VCF file. Based on the mutation results, please recommend possible targeted drugs for the patient. If there are multiple drugs, please sort them by priority.

[0165] The input file VCF format is as follows:

[0166]

[0167]

[0168] In this example of tumor analysis, we can utilize the aforementioned large-model-based multi-agent collaborative system to accomplish the task. This task involves analyzing somatic mutation data of a tumor patient stored in a VCF file to ultimately recommend potential targeted drugs. We can break this process down into the following steps based on the overall system architecture:

[0169] 1) Workflow Plan:

[0170] The requirements analyst defines the analysis task in detail and determines the specific output format, such as the structure of a drug recommendation report. The tool selector chooses the most suitable tool from the tool pool based on the proposed task requirements. Each tool has a tool description and input / output; the most suitable tool is selected. The process compares the functionality of the tool description with the functionality required by the user's task. This invention first determines which tools to select, and then the planner connects them into a workflow. The explorer evaluates possible analysis tools and algorithms, selecting the tool most suitable for tumor mutation analysis. The planner defines the entire analysis workflow, ensuring that each step from extracting data from the VCF file to the final drug recommendation is planned. For example, if the explorer believes that A->B and B->C can be connected (ABC here refers to different tools), then the planner can connect ABC using the A->B->C workflow, i.e., the defined analysis workflow. Downstream from this workflow, other processes will continue to generate and run code.

[0171] 2) Process execution:

[0172] The executor runs the selected analysis tool to perform a preliminary analysis of the VCF file. The scheduler arranges and optimizes the task execution order to ensure efficient data processing. Programmers write or adjust code as needed to suit specific analysis requirements. Debuggers resolve any technical issues encountered during the analysis process. Monitors track the execution process to ensure the quality of the data analysis. Testers verify the accuracy of the analysis results.

[0173] The system planning process is as follows:

[0174]

[0175]

[0176]

[0177] 3) Biological explanation:

[0178] Based on the analysis results, the inquirer asks questions about the specific impact of the variants on the patient's cells or pathology. The data analyst collects biological information and known associations for each variant. The plotter presents the variant results and their associations with known tumor markers graphically to aid understanding for non-specialists. Senior biologists review the data and propose potential candidates based on the biological significance of the variants.

[0179] The results of the biological analysis are as follows:

[0180]

[0181] 4) Medical Insights:

[0182] Evidence data specialists compile relevant clinical trial and drug approval information. Evidence analysts assess the level of evidence for each recommended drug. Visualization experts present drug recommendations in an intuitive way, such as priority charts.

[0183] Senior medical experts determine the most appropriate drug selection and its priority based on the level of evidence and clinical experience.

[0184] The result of this process is a report that generates targeted drug recommendations and their priorities, based on patient-specific somatic variant data. The entire analysis considers drug efficacy, patient variant characteristics, drug side effects and interactions, and existing clinical evidence. The final drug recommendations will help clinicians make more informed treatment decisions for patients.

[0185] The medical analysis results are as follows:

[0186]

[0187]

[0188] Description of results: Priority 1 drugs are preferred, including Afatinib, Dacomitinib, Erlotinib, Erlotinib+Ramucirumab, Gefitinib, and Osimertinib. These drugs target specific parts of the patient's genetic mutations, so they may have a positive impact on treatment outcomes. The priority 2 drug PatritumabDeruxtecan is a secondarychoice and may serve as an alternative when the priority 1 drug is ineffective orunusable. Please note that these recommendations require further evaluation and confirmation from doctors."

[0189] The following are system embodiments corresponding to the above method embodiments. This embodiment can be implemented in conjunction with the above embodiments. The relevant technical details mentioned in the above embodiments are still valid in this embodiment, and will not be repeated here to reduce repetition. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the above embodiments.

[0190] like Figure 4 As shown, this invention also proposes a zero-code bioinformatics analysis device B based on multi-agent collaboration, which includes:

[0191] The multi-agent setting module 100 sets the large language model as multiple agents through prompt words. These multiple agents include a requirements analyst, selector, and planner for developing workflow plans; a programmer, scheduler, and executor for executing workflows; a data analyst, plotter, and biological expert for biologically interpreting the execution results; and an evidence-based data analyst, medical plotter, and medical expert for providing medical insights from the biological interpretation results.

[0192] The workflow planning module 200 is used to acquire bioinformatics and the tasks to be analyzed. The requirements analyst determines the output format of the tasks to be analyzed. The selector selects the bioinformatics analysis tool that is closest to the requirements of the tasks to be analyzed from the tool library according to the description of each tool. The planner formulates the workflow based on the selected bioinformatics analysis tool.

[0193] In the process execution module 300, the programmer writes the workflow as work code, the scheduler arranges and optimizes the execution order of the work code, and then the executor executes the optimized work code to obtain the analysis results;

[0194] In the biological interpretation module 400, the data analyst checks and filters the analysis results to obtain filtering information; the plotter presents the filtering information graphically to obtain graphical information; and the biologist obtains candidate answers for the task to be analyzed based on the filtering information and the graphical information.

[0195] In the Medical Insights module 500, the evidence-based data analyst finds corresponding medical evidence and clinical data for the candidate answer based on clinical trials, research papers, and drug approval information; the medical illustrator displays the candidate answer in chart form to obtain a graphical answer; the candidate answer and the graphical answer are input into the medical expert, who assigns a priority to the candidate answer based on the medical evidence and clinical data, and outputs the recommended answer and its priority as the analysis result for the task to be analyzed.

[0196] The aforementioned zero-code bioinformatics analysis device based on multi-agent collaboration, wherein the bioinformatics and the task to be analyzed are respectively gene detection data and analysis of which genetic disease the gene detection data has; or tumor detection data and analysis of which targeted drugs the tumor detection data is suitable for.

[0197] The aforementioned zero-code bioinformatics analysis device based on multi-agent collaboration further includes, among other agents:

[0198] Debugger, used to correct errors in the workflow;

[0199] The monitor is used to track the intermediate execution results of each stage of the working code to ensure the correct execution of the working code;

[0200] The inquirer raises questions and hypotheses about the analysis results based on a specified knowledge base, focusing on pathways, cells, and genes. The data analyst then verifies and filters the analysis results based on these questions and hypotheses.

[0201] The aforementioned zero-code bioinformatics analysis device based on multi-agent collaboration includes a workflow planning module comprising: constructing a tool library from the background knowledge of various bioinformatics analysis tools; performing a corresponding retrieval for the task to be analyzed; obtaining bioinformatics analysis tools and their descriptions that meet the requirements of the task to be analyzed; and providing them as prompt words to the selector based on a large language model to enhance knowledge.

[0202] The present invention also proposes an electronic device, including the aforementioned zero-code bioinformatics analysis device based on multi-agent collaboration.

[0203] The electronic device is connected to an information display device, which displays the evaluation result using user-defined display parameters, attributes, or through an artificial intelligence model.

[0204] like Figure 5 As shown, in another embodiment of the present invention, a first electronic device A is also proposed, which includes the aforementioned zero-code bioinformatics analysis device based on multi-agent collaboration.

[0205] like Figure 6 As shown, the first electronic device A can also be connected to the data acquisition device C and the information display device D through a wired or wireless information transmission scheme. The data acquisition device C is used to collect and acquire biological information and the tasks to be analyzed, and the information display device D is used to display the analysis results obtained by the present invention.

[0206] Information display device D can process and organize the data output by the first electronic device A based on an information display mechanism to improve the readability of the data. This information display mechanism can be manually preset, for example, visualizing the data output by the first electronic device A. It can present the user with specified key information, such as news updates or system information, based on user-defined display parameters and / or attributes. Display parameters could be, for example, the data range to be displayed, and display attributes could be, for example, the font, color, or whether scrolling is enabled. This allows the user to access this information more quickly without having to navigate to secondary pages or scroll through pages, saving user effort. Alternatively, this information display mechanism can be an artificial intelligence (AI) display model, which can learn the user's key information interests based on previous usage habits, such as viewing time, number of clicks, and number of edits, and then automatically present the user with rich and necessary key information.

[0207] In another embodiment, the present invention also proposes a storage medium VIII for storing a computer program that executes the aforementioned zero-code bioinformatics analysis method based on multi-agent cooperation. It should be understood that the storage medium in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which serves as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0208] Figure 7 A schematic block diagram of a second electronic device 1000 that can be used to implement embodiments of the present invention is shown. The second electronic device 1000 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The second electronic device 1000 can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components described herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein. The second electronic device 1000 may be the same as or different from the first electronic device A.

[0209] The second electronic device 1000 includes a computing unit I, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory II (ROM) or a computer program loaded from storage medium VIII into random access memory (RAM) III. The RAM III may also store various programs and data required for the operation of the device 1000. The computing unit I, ROM II, and RAM III are interconnected via bus IV. An input / output (I / O) interface V is also connected to bus IV.

[0210] Multiple components in the second electronic device 1000 are connected to I / O interface V, including: input unit VI, such as a keyboard, mouse, etc.; output unit VII, such as various types of displays, speakers, etc.; storage medium VIII, such as a disk, optical disk, etc.; and communication unit IX, such as a network card, modem, wireless transceiver, etc. Communication unit IX allows the second electronic device 1000 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0211] Computing unit I can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of computing unit I include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Computing unit I performs the various methods and processes described above, such as method steps 1-5. For example, in some embodiments, the methods can be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage medium VIII. In some embodiments, part or all of the computer program can be loaded and / or installed on device 1000 via ROM II and / or communication unit IX. When the computer program is loaded into RAM III and executed by computing unit I, one or more steps of the methods described above can be performed. Alternatively, in other embodiments, computing unit I can be configured to perform methods by any other suitable means (e.g., by means of firmware).

[0212] In summary, this invention employs a multi-agent collaborative approach to bioinformatics analysis tool selection. Multiple agents, based on tool scoring and reordering, engage in scoring discussions to collaboratively select relevant tools. Through adaptive assembly of bioinformatics analysis tools via multi-agent collaborative workflow, exploratory assembly is performed based on different tool parameters, and complex workflows are built after multiple successful tests. Furthermore, by combining downstream interpretation and literature tools, along with knowledge bases, literature retrieval interfaces, and large-scale medical model tools, interpretability tools in biology and medicine are integrated to achieve interpretable analysis of bioinformatics results.

[0213] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A zero-code bioinformatics analysis method based on multi-agent collaboration, characterized in that, include: The multi-agent setup step involves setting up a large language model as multiple agents using prompt words. These agents include a requirements analyst, a selector, and a planner for developing workflow plans. Programmers, schedulers, and executors who execute workflows; data scientists, plotters, and biological experts who interpret the execution results biologically. Evidence-based data analysts, medical illustrators, and medical experts who provide medical insights based on biological interpretations; The workflow planning steps are used to acquire bioinformatics and the tasks to be analyzed. The requirements analyst determines the output format of the tasks to be analyzed. The selector selects the bioinformatics analysis tool that is closest to the requirements of the tasks to be analyzed from the tool library based on the description of each tool. The planner develops the workflow based on the selected bioinformatics analysis tool. The process execution steps are as follows: the programmer writes the workflow as work code, the scheduler arranges and optimizes the execution order of the work code, and then the executor executes the optimized work code to obtain the analysis results; In the biological interpretation process, the data analyst checks and filters the analysis results to obtain filtering information; The drafter presents the filtered information graphically, thus obtaining graphical information. Based on the screening information and the graphic information, the biologist obtained candidate answers for the task to be analyzed. The medical insight step involves the evidence-based data analyst finding corresponding medical evidence and clinical data for the candidate answer based on clinical trials, research papers, and drug approval information. The medical illustrator presented the candidate answers in a chart format, resulting in a graphical answer. The candidate answer and the graphical answer are input into the medical expert. Based on the medical evidence and clinical data of the candidate answer, the medical expert assigns a priority to the candidate answer and outputs the recommended answer and its priority as the analysis result of the task to be analyzed.

2. The zero-code bioinformatics analysis method based on multi-agent collaboration as described in claim 1, characterized in that, The bioinformatics and the task to be analyzed are respectively gene detection data and analysis of which genetic disease the gene detection data indicates; or tumor detection data and analysis of which targeted drugs the tumor detection data is suitable for.

3. The zero-code bioinformatics analysis method based on multi-agent collaboration as described in claim 1, characterized in that, These multiple intelligent agents also include: Debugger, used to correct errors in the workflow; The monitor is used to track the intermediate execution results of each stage of the working code to ensure the correct execution of the working code; The inquirer raises questions and hypotheses about the analysis results based on a specified knowledge base, focusing on pathways, cells, and genes. The data analyst then verifies and filters the analysis results based on these questions and hypotheses.

4. The zero-code bioinformatics analysis method based on multi-agent collaboration as described in claim 1, characterized in that, The planned steps of this workflow include: constructing a tool library by accumulating background knowledge of various bioinformatics analysis tools; performing a corresponding search for the task to be analyzed; obtaining bioinformatics analysis tools and their descriptions that meet the needs of the task to be analyzed; and providing these as prompt words to the selector based on the large language model to enhance knowledge.

5. A zero-code bioinformatics analysis device based on multi-agent collaboration, characterized in that, include: The multi-agent setting module sets the large language model as multiple agents through prompt words. These multiple agents include a requirements analyst, a selector, and a planner for formulating workflow plans. Programmers, schedulers, and executors who execute workflows; data scientists, plotters, and biological experts who interpret the execution results biologically. Evidence-based data analysts, medical illustrators, and medical experts who provide medical insights based on biological interpretations; The workflow planning module is used to acquire bioinformatics and the tasks to be analyzed. The requirements analyst determines the output format of the tasks to be analyzed. The selector selects the bioinformatics analysis tool that is closest to the requirements of the tasks to be analyzed based on the description of each tool in the tool library. The planner develops the workflow based on the selected bioinformatics analysis tool. In the process execution module, the programmer writes the workflow as work code. After the scheduler arranges and optimizes the execution order of the work code, the executor executes the optimized work code and obtains the analysis results. In the biological interpretation module, the data analyst checks and filters the analysis results to obtain filtering information; The drafter presents the filtered information graphically, thus obtaining graphical information. Based on the screening information and the graphic information, the biologist obtained candidate answers for the task to be analyzed. In the Medical Insights module, the evidence-based data analyst searches for corresponding medical evidence and clinical data for the candidate answer based on clinical trials, research papers, and drug approval information. The medical illustrator presented the candidate answers in a chart format, resulting in a graphical answer. The candidate answer and the graphical answer are input into the medical expert. Based on the medical evidence and clinical data of the candidate answer, the medical expert assigns a priority to the candidate answer and outputs the recommended answer and its priority as the analysis result of the task to be analyzed.

6. The zero-code bioinformatics analysis device based on multi-agent collaboration as described in claim 5, characterized in that, The bioinformatics and the task to be analyzed are respectively gene detection data and analysis of which genetic disease the gene detection data indicates; or tumor detection data and analysis of which targeted drugs the tumor detection data is suitable for.

7. The zero-code bioinformatics analysis device based on multi-agent collaboration as described in claim 5, characterized in that, These multiple intelligent agents also include: Debugger, used to correct errors in the workflow; The monitor is used to track the intermediate execution results of each stage of the working code to ensure the correct execution of the working code; The inquirer raises questions and hypotheses about the analysis results based on a specified knowledge base, focusing on pathways, cells, and genes. The data analyst then verifies and filters the analysis results based on these questions and hypotheses.

8. The zero-code bioinformatics analysis device based on multi-agent collaboration as described in claim 5, characterized in that, The workflow planning module includes: constructing a tool library by accumulating background knowledge of various bioinformatics analysis tools; performing a corresponding search for the task to be analyzed; obtaining bioinformatics analysis tools and their descriptions that meet the needs of the task to be analyzed; and providing them as prompt words to the selector based on a large language model to enhance knowledge.

9. An electronic device, characterized in that, This includes a zero-code bioinformatics analysis device based on multi-agent collaboration as described in claims 5-8.

10. The electronic device as claimed in claim 9, characterized in that, The electronic device is connected to an information display device, which displays the evaluation results using user-defined display parameters, attributes, or through an artificial intelligence model.

11. A storage medium for storing a computer program that executes the zero-code bioinformatics analysis method based on multi-agent cooperation as described in claims 1-4.

Images