Atlas inference analysis system for cell antigen treatment defects

Abstract

The invention discloses an atlas inference analysis system for cell antigen treatment defects, relates to the technical field of knowledge atlas inference, and is used for solving the problem of slow explaining and positioning of defect presentation. According to the method, a detection likelihood set is constructed by fusing HLA typing, somatic cell variation, transcriptional expression, immunopeptidomics and surface HLA detection, direct presentation and cross presentation sub-graphs are assembled by adopting a mode gating and loading complex integrity rule, a reachable path and observation consistency matrix is generated, structural contradictions and detection limitations are distinguished, and a detection result is obtained. According to constraint propagation, node cutting and link shrinkage are performed to form an individualized map, defect nodes and defect paths are searched and output based on a minimum interpretation path and a cause and effect graph cost, and a verification and intervention list and clinical decision support output are generated, so that unnecessary detection and intervention attempts are reduced, and the detection efficiency is improved. And the timeliness and consistency of case analysis and strategy making are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of knowledge graph reasoning, more particularly, the present application relates to a graph reasoning analysis system for cell antigen processing defects. BACKGROUND

[0002] In clinical immunoinformatics, researchers construct the antigen processing and presentation process into a knowledge graph, map the genome, transcription, immune peptide spectrum and surface human leukocyte antigen (HLA) detection to the graph according to a unified time and terminology, and perform path reasoning on the links of lysis, transport and modification to locate the presentation damage link and form an interpretable decision support.

[0003] However, the existing scheme has bottlenecks. First, the graph connection is usually static, which is difficult to update with interferon status and cell type, and it is easy to confuse reversible transcription inhibition with irreversible structural defects. Moreover, the evidence collection and processing period are inconsistent, and the cause and effect sequence on the graph is difficult to verify, leading to missed detection of immune peptide spectrum in limited sample size and modified peptide scenarios, which is often written as a neutral blank, causing the posterior defect positioning and verification sequence to shift, affecting clinical landing and reuse. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the following scheme is provided to solve the problem of slow interpretable positioning of defect presentation in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme: A graph reasoning analysis system for cell antigen processing defects, comprising a case data mapping module, an antigen assembly module, a presentation demand determination module, an individual strategy analysis module and a decision adjustment module, and the modules are connected through signals; The case data mapping module is used to obtain HLA locus typing, somatic variation and copy number, transcription expression, immune peptide omics and surface HLA detection results of the case, and complete standardized mapping; The antigen assembly module is used to assemble the antigen processing and presentation knowledge graph, including lysis, transport, modification, loading, display, and distinguishing between direct presentation and cross-presentation; The presentation demand determination module is used to calculate the link connectivity and observation consistency on the assembled antigen processing and presentation knowledge graph, and determine whether the presentation demand is met according to the preset presentation demand condition; The individual strategy analysis module is used to determine the unusable channel and presentation mode association according to the HLA locus copy loss, inactivation mutation and domain damage when it is determined that the presentation demand is not met, and generate a graph editing sequence using constraint propagation, trim the nodes and shrink the links to obtain an individualized graph; The decision adjustment module is used to perform evidence propagation and minimum interpretation path reasoning on the individualized graph by combining the detection likelihood set, outputting defect nodes, defect paths and evidence chains, and constructing a verification and intervention causal graph. It searches for the minimum cost path from the current presentation state to the target state that meets the presentation requirements, and generates clinical decisions for output.

[0006] Furthermore, the case data mapping module includes: Based on a unified time reference and glossary, the field standardization and aliasing of HLA locus typing, somatic cell variation and copy number, transcript expression, immunopeptidomics and surface HLA detection results were standardized and compared with aliases. Allele nomenclature standardization and heterozygosity consistency verification were performed on HLA locus typing. Genome coordinate alignment and functional annotation were performed on somatic cell variations and copy numbers, and markers of copy deletion, inactivation mutations and domain disruption were extracted. Gene expression status tags are generated from transcriptional expression. Immunopeptidomics was used to perform spectroscopic identification and confidence screening, and to map peptides to presentation sites. Surface HLA detection was numerically normalized and surface expression tags were generated. A detection likelihood set is established based on the sample input amount, peptide type, and experimental procedure parameters, and then output along with the label.

[0007] Furthermore, the requirement determination module includes: On the assembled antigen processing and presentation knowledge graph, the set of reachable paths from input to output is calculated in the order of cleavage, transport, modification, loading, and display. The labels output by the case data mapping module are compared with the corresponding nodes to form an observation consistency matrix; Based on the preset presentation requirements, the set of reachable paths and the observation consistency matrix are jointly determined, and the satisfaction result is output.

[0008] Furthermore, the individual strategy analysis module includes the following when it determines that the presentation requirements are not met: HLA site copy deletion, inactivation mutation, and domain disruption markers were extracted from the case data mapping results. The markers are projected onto the nodes and links associated with alleles to identify the set of unavailable channels and to build a channel association mapping table. Based on the channel association mapping table, constraint propagation is used to generate a graph editing sequence, which performs pruning on associated nodes and shrinking on dependent links to obtain an individualized graph.

[0009] Furthermore, the constraint propagation method for generating graph editing sequences includes: A constraint-satisfying network is constructed using node states and edge channel states as variables. Inject node predecessor availability, loading complex integrity, and allele existence as constraints into the network. Perform constraint propagation to reduce the feasible domain of variables and identify the minimum cut set that leads to inconsistency; The editing actions are sorted according to topological order and inconsistency resolution priority rules, and the graph editing sequence is output.

[0010] Furthermore, the decision adjustment module includes: Evidence propagation is performed on the individualized graph by combining the detection likelihood set and calculating the interpretation score of each node and each path; A verification and intervention causal graph is constructed based on the intervention points and the observed outputs. Nodes represent intervention points and observed outputs, and edges represent regulation or functional dependencies. A path cost function is constructed based on intervention costs, expected incremental increases, and penalties for inconsistencies in evidence. Heuristic search is used to search for the minimum cost path from the current presentation state to the target state that satisfies the presentation requirements in the causal graph; Output the minimum cost path and its corresponding evidence chain, and generate clinical decision support output.

[0011] Furthermore, the detection likelihood set established by the case data mapping module includes: Generate sample-level detection parameters based on the sample input volume, preprocessing method, and batch size. Peptide-level detection parameters are generated based on peptide charge-to-mass ratio range, retention time deviation, fragment ion coverage, identification confidence level, and modification type. Surface expression detection parameters are generated based on antibody cloning, labeling channel overflow correction, and instrument gain for surface HLA detection. Sample-level detection parameters, peptide-level detection parameters, and surface expression detection parameters are mapped to observation confidence labels and bound to corresponding nodes or observation records.

[0012] Furthermore, when distinguishing between direct presentation and cross-presentation, the antigen assembly module performs the following actions: Based on cell type labels and presentation pattern markers, evidence is propagated only within subplots that match them; Using the presentation layer nodes as the merge nodes, the consistency of the results from the direct presentation subgraph and the cross presentation subgraph is checked. When the verification is inconsistent, the conflict path is marked and a presentation mode conflict prompt is output for graph editing in the individual strategy analysis module.

[0013] Furthermore, the edit log includes: The unique identifier of the edited node and link, the corresponding allele, and the presentation pattern marker; The source of evidence that triggered the edit, the timestamp, and the version identifier; Editing action type and execution order, node state and link state before and after editing; Rollback mark and traceable check code for recovery.

[0014] Further, the structured message includes: Defect node list, defect path set and corresponding evidence chain; Reversibility label, verification item list and intervention item list; Data version, algorithm version and atlas version identification, evidence source identification and timestamp; Message header and field mapping relationship for medical information system docking, and provides backfilling and auditing interface to record receiving and calling state.

[0015] The technical effects and advantages of the antigen processing defect-oriented atlas reasoning analysis system of the present application are: The present application realizes the standardized integration, authenticity discrimination and individualized correction of defect evidence in the antigen processing and presentation pathway by constructing a closed-loop reasoning system of case data mapping, antigen assembly, presentation demand determination, individual adjustment and decision adjustment. The system introduces mode gating and loading complex integrity rules in the assembly stage to perform hierarchical assembly and conflict preposition resolution on direct presentation and cross-presentation. In the judgment stage, a reachable path set and an observation consistency matrix are generated in parallel, and the undetected observation is labeled as pending by detecting the likelihood set, which distinguishes structural contradictions and detection limitations, reduces the risk of misreading non-random missing as pathway disruption from the source, and thus can stably identify HLA-related structural defects and regulatory inhibition differences, improve the explainability and consistency of defect positioning, and reduce the probability of invalid verification and repeated experiments.

[0016] On this basis, the defect node, defect path and evidence chain are output by the minimum explanation path reasoning, and a verification and intervention causal graph is constructed accordingly. The reverse measurement of intervention cost and expected presentation improvement range, observation contradiction penalty and channel risk measurement are searched to find the minimum cost path from the current presentation state to the target state that meets the presentation demand, generate a verification item list and an intervention item list, and form a clinical decision support output. The execution result is backfilled with a structured message and carries data version, algorithm version, atlas version and evidence source identification. The supporting editing log and evidence chain realize process traceability and auditing, improve the accuracy and stability of defect path pointing, and also consider the interoperability and reuse in clinical processes, reduce unnecessary detection and intervention attempts, and improve the timeliness and consistency of case analysis and strategy formulation. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The structural diagram of the antigen processing defect-oriented atlas reasoning analysis system of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be apparently and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work belong to the protection scope of the present application.

[0019] To achieve the above-mentioned purpose, Figure 1 The structure diagram of the antigen processing and presentation knowledge graph is shown in the embodiment of the present application, and specifically comprises a case data mapping module, an antigen assembly module, a presentation demand determination module, an individual strategy analysis module, and a decision adjustment module, and the modules are connected through signals; The case data mapping module is used for obtaining HLA locus typing, somatic variation and copy number, transcription expression, immune peptidomics, and surface HLA detection results of a case, and completing standardized mapping; The antigen assembly module is used for assembling the antigen processing and presentation knowledge graph, including cleavage, transport, modification, loading, display, and distinguishing direct presentation and cross-presentation; The presentation demand determination module is used for calculating link connectivity and observation consistency on the assembled antigen processing and presentation knowledge graph, and judging whether the presentation demand is met according to a preset presentation demand condition; The individual strategy analysis module is used for determining the unusable channel and presentation mode association according to the HLA locus copy loss, inactivation mutation, and domain damage when it is judged that the presentation demand is not met, and generating a graph editing sequence by using constraint propagation, trimming nodes and shrinking links to obtain an individualized graph; The decision adjustment module is used for performing evidence propagation and minimum explanation path reasoning on the individualized graph combined with a detection likelihood set, outputting defect nodes, defect paths, and evidence chains, and constructing a verification and intervention causal graph, searching for a minimum cost path from a current presentation state to a target state meeting the presentation demand, and generating a clinical decision for output.

[0020] The case data mapping module is used for obtaining HLA locus typing, somatic variation and copy number, transcription expression, immune peptidomics, and surface HLA detection results of a case, and completing standardized mapping, and specifically implemented as: Access HLA typing, somatic mutation and copy number, transcription expression, immune peptidomics and surface HLA detection results from data sources one by one, and perform field specification, type checking and missing value labeling; perform allele naming standardization and heterozygosity consistency checking on HLA typing, if there are typing results from different platforms for the same case, then perform conflict mediation according to typing resolution priority and sequencing batch quality label, if the mediation condition is not met, generate a source audit field and keep double-track records for subsequent determination; Perform genomic coordinate alignment and functional annotation on somatic mutation and copy number data, after aligning to the same reference coordinate, identify and label copy loss markers, inactivation mutation markers and domain disruption markers; normalize the transcription expression matrix based on sequencing depth, gene length and batch effect, generate gene expression state labels combined with expression distribution thresholds, and bind the labels with specific samples, tissue sites and timestamps to form traceable mapping records.

[0021] Perform spectrum identification confidence screening on immune peptidomics data, use ion matching coverage, retention time deviation and identification confidence threshold to perform quality control on the results, after removing low-confidence spectrum records, map the remaining peptide segments to presentation sites, the mapping process is constrained by the HLA typing of the case, first select candidate presentation sites according to alleles, then match according to peptide segment length and binding site characteristics, peptide segments that cannot meet the matching conditions but have high experimental reliability are labeled as pending and the original evidence is retained.

[0022] Normalize the surface HLA detection results according to antibody clone, instrument gain and marker channel overflow correction parameters, the normalized results are converted into surface expression labels, and are bound with the corresponding cell population and detection time. To ensure the availability of multi-source evidence in subsequent graph reasoning, sample input quantity, pretreatment method and on-machine batch are summarized as sample-level detection parameters, peptide segment mass-to-charge ratio range, retention time deviation, fragment ion coverage, identification confidence and modification type are summarized as peptide-level detection parameters, and antibody clone, marker channel overflow correction and instrument gain of surface HLA detection are summarized as surface expression detection parameters. The three types of parameters together constitute the detection likelihood set.

[0023] The detection likelihood set is used to describe the likelihood of each observation being truly detected, when the sample-level detection parameters or peptide-level detection parameters indicate that the detection probability is low, the module does not regard the non-detection as missing, but labels the node or edge corresponding to the observation as pending and adds a detection likelihood weight.

[0024] The final output includes two parts: the first part is a label set, including copy loss markers, inactivation mutation markers, domain disruption markers, gene expression state markers, peptide segment to presentation site mapping, and surface expression markers, all of which have source identification, version identification, and time stamp, and are in one-to-one correspondence with specific nodes or edges; the second part is a detection likelihood set, including sample-level detection parameters, peptide segment-level detection parameters, and surface expression detection parameters, which are mapped to observation confidence labels and bound to corresponding nodes or observation records.

[0025] An antigen assembly module for assembling an antigen processing and presentation knowledge graph, including cleavage, transport, modification, loading, display, and distinguishing between direct presentation and cross-presentation, is implemented as follows: Read the label set and the detection likelihood set, where the label set includes copy loss markers, inactivation mutation markers, domain disruption markers, gene expression state markers, peptide segment to presentation site mapping, and surface expression markers, and the detection likelihood set includes sample-level detection parameters, peptide segment-level detection parameters, and surface expression detection parameters.

[0026] The assembly process takes layered mapping as the core, first establishing nodes and directed edges of the cleavage layer, transport layer, modification layer, loading layer, and display layer in the direct presentation subgraph in turn. The cleavage layer includes proteasome activity nodes and substrate nodes connected to them, the transport layer includes TAP (antigen processing-related transporter) transporter nodes, the modification layer includes ERAP (endoplasmic reticulum aminopeptidase) No. 1 and ERAP No. 2 nodes, the loading layer includes loading complex nodes and case-specific HLA class I heavy chain nodes and β2 microglobulin nodes, and the display layer includes surface HLA class I complex nodes. Then establish the exogenous uptake layer, endosome and lysosome degradation layer, transmembrane translocation or vacuole processing layer, loading layer, and display layer in the cross-presentation subgraph, where the loading layer includes HLA class II molecules and paired invariant segments, as well as exchange regulation nodes of HLA-DM and HLA-DO.

[0027] The gene expression state markers are attached to the corresponding coding gene nodes, the copy loss markers, inactivation mutation markers, and domain disruption markers are attached to the corresponding structure nodes, the peptide segment to presentation site mapping is attached to the loading layer and display layer, and the surface expression markers are attached to the display layer nodes. Each node generates an initial state field and an evidence reference field, with the initial state taking values of available, pending, and unavailable, and the evidence reference field recording the label source, time stamp, and version identification. The detection likelihood set is used to generate observation confidence labels and write into the evidence weight field of the edges and nodes. When the sample-level detection parameters or peptide segment-level detection parameters show low detection probability, the weight field of the corresponding nodes and edges is set to pending weight, which is used for subsequent observation consistency calculation by the presentation demand determination module.

[0028] The fine assembly and conflict pre-elimination of the twin sub-graphs are achieved by mode gating and loading complex integrity rules. The evidence is only propagated in the sub-graphs consistent with the cell type label and presentation mode marker. The direct presentation sub-graph and the cross-presentation sub-graph are set to merge nodes in the display layer for result checking. If two sub-graphs give contradictory conclusions for the same display layer node, the module generates a presentation mode conflict prompt and marks the conflict path during the assembly phase, which is handed over to the subsequent individual strategy analysis module for processing.

[0029] When any of the TAPBP (tapasin), calreticulin, ERP57 or beta2 microglobulin in the key components of the loading layer is determined to be unavailable, the loading edge and the display edge connected to the component are immediately marked as unavailable. The above processing is performed in the assembly phase to directly solidify structural defects as graph structure constraints, rather than waiting for the subsequent reasoning phase to make a unified judgment. The allele existence rule indicates that once the HLA site copy deletion marker or inactivation mutation marker of the case is confirmed, the corresponding HLA class I heavy chain node and its link to the display layer are marked as unavailable, while the feasible link associated with another allele is not affected. The module finally outputs the assembled antigen processing and presentation knowledge graph object, including the node set and edge set of the direct presentation sub-graph and the cross-presentation sub-graph, the node initial state field, the evidence weight field, the mode marker and the assembly log.

[0030] The presentation demand determination module is used to calculate link connectivity and observation consistency on the assembled antigen processing and presentation knowledge graph, and determine whether the presentation demand is met according to the preset presentation demand conditions. The specific implementation is as follows: The antigen processing and presentation knowledge graph output by the antigen assembly module is read, including the direct presentation sub-graph and the cross-presentation sub-graph. The label set includes the copy deletion marker, the inactivation mutation marker, the domain damage marker, the gene expression state label, the mapping of the peptide segment to the presentation site and the surface expression label. The detection likelihood set includes the sample level detection parameter, the peptide segment level detection parameter and the surface expression detection parameter.

[0031] First, two traversals are performed on the knowledge graph from top to bottom and from bottom to top, and a set of reachable paths is generated in the order of cleavage, transport, modification, loading and display. During traversal, mode gating is enabled, and the extension is only performed within the sub-graph consistent with the cell type label and the presentation mode marker. When a node is marked as unavailable by the label set or all upstream edges of the node are determined to be unavailable, the extension is terminated and the breakpoint reason is recorded. Each path that completes from the cleavage layer to the display layer is registered as a reachable path, and the source identification, timestamp and version identification of each node and edge in the path are attached for subsequent comparison and audit. Paths that cannot pass through the display layer are registered as candidate paths and marked with the interruption link.

[0032] Subsequently, an observation consistency matrix is constructed for comparing the label set with the set of reachable paths one by one. The comparison rules include four categories: The first category is structural consistency. Whether the copy loss label, inactivation mutation label and domain disruption label are inconsistent with the retention of the corresponding node in the path is compared. If inconsistent, the node and dependent edge are marked as observation contradiction; The second category is functional consistency. The gene expression state label and the functional requirement of the path node are compared. When the essential component of the loading complex is marked as expression loss, the loading edge directly connected thereto is marked as observation contradiction; The third category is peptide segment consistency. The mapping of the peptide segment to the presentation site and the matching relationship of the loading layer and the display layer are compared. When the peptide segment has mapping evidence and the peptide segment level detection parameter meets the detection confidence requirement, the corresponding position of the path is marked as observation support. When no detection is made and the sample level detection parameter or the peptide segment level detection parameter indicates that the detection probability is low, it is recorded as pending rather than contradiction, so as to avoid misjudging non-random loss as chain breakage; The fourth category is surface consistency. The surface expression label and the display layer node are compared. When the surface expression label is positive and the surface expression detection parameter meets the detection confidence requirement, it is determined as observation support. When the surface expression label is negative and there is sufficient observation support upstream, it is marked as observation contradiction. The unit value in the matrix is limited to observation support, observation contradiction and pending, and the corresponding source identifier, timestamp and version identifier are retained. For the inconsistent situation of the conclusions of the direct presentation subgraph and the cross-presentation subgraph on the same display layer node, a presentation mode conflict item is generated in the matrix and recorded.

[0033] Finally, the satisfaction result is output according to the preset presentation requirement condition. The presentation requirement condition includes three judgment rules. The path penetration rule requires that at least one reachable path from the cleavage layer to the display layer without observation contradiction exists on each existing HLA allele channel. The observation coverage rule requires that there is at least one item marked as observation support by peptide segment consistency or surface consistency in the display layer for each case. One of the two can satisfy the requirement to recognize that the display end has evidence support. The conflict prohibition rule requires that there is no unexplained presentation mode conflict or structural consistency contradiction.

[0034] The module inputs the reachable path set and the observation consistency matrix into the above rules to obtain a satisfaction result and a non-satisfaction list, the non-satisfaction list including an unavailable channel set, a pending channel set, a presentation mode conflict item and a breakpoint reason, serving as an input of an individual strategy analysis module for constraint propagation and generation of a graph editing sequence; the satisfaction result and the support item serving as an input of a decision adjustment module for establishing a start point and a cost function in a causal graph, the innovation of the module being that the reachable path set and the observation consistency matrix are generated in parallel and pending labeling is performed by using a detection likelihood set, so that observation contradiction and detection limitation are distinguished from the source, thereby reducing false injury to a real feasible path, and a direct presentation and a cross presentation are exposed in advance in a judgment stage by using a mode gating and a conflict item mechanism.

[0035] The individual strategy analysis module is configured to determine an unavailable channel and a presentation mode association according to HLA site copy loss, inactivation mutation and domain damage when the judgment is not satisfied, and generate a graph editing sequence by using constraint propagation, and obtain an individualized graph by node pruning and link contraction. The satisfaction result and the non-satisfaction list are taken as a start condition, and copy loss markers, inactivation mutation markers, domain damage markers, gene expression state tags, mappings of peptide segments to presentation sites, surface expression tags and a detection likelihood set output by a case data mapping module are read, the detection likelihood set being composed of sample-level detection parameters, peptide segment-level detection parameters and surface expression detection parameters, a channel association mapping table is first generated, a channel being defined as a directed link set from a cleavage layer to a display layer with a specific HLA allele as an anchor point and a presentation mode being limited, an allele existence rule marking nodes of the loading layer and the display layer as unavailable when the allele has a copy loss marker, an inactivation mutation marker or a domain damage marker, and integrating upstream and downstream links thereof into an unavailable channel set; A loading complex integrity rule requires that necessary components of a loading complex satisfy availability at the same time, otherwise a loading edge and a display edge connected thereto enter the unavailable channel set; a mode gating rule separates evidences of a direct presentation subgraph and a cross presentation subgraph, and records two groups of paths corresponding to a presentation mode conflict item as candidate editing objects; An observation preservation rule marks nodes and edges having a mapping of a peptide segment to a presentation site and satisfying a peptide segment-level detection parameter credibility threshold as observation priority preservation objects, and for a situation that a sample is not detected and a detection probability is low as shown by sample-level detection parameters or peptide segment-level detection parameters, the module does not integrate the link into the unavailable channel set, but integrates the link into a pending channel set and records a pending weight, and a channel association mapping table generates the unavailable channel set, the pending channel set, a presentation mode marker and evidence source identification in accordance with this, thereby providing a boundary for subsequent graph editing.

[0036] After obtaining the channel association mapping table, the module generates a graph editing sequence using constraint propagation. The variables are node states and edge channel states, taking values of available, pending, and unavailable. The constraints include node predecessor availability constraints, loading complex integrity constraints, allele existence constraints, presentation mode consistency constraints, and observation preservation constraints.

[0037] The unavailable channel set is assigned as unavailable, the pending channel set is assigned as pending, and constraint propagation is performed on the topological order of the knowledge graph to shrink the variable feasible region. When all upstream edges of a node are unavailable or conflict with the loading complex integrity constraint, the node is marked as unavailable and propagated downstream. When a presentation mode conflict entry occurs, the two mode subgraphs are solved separately, the subgraph that is consistent with the observation preservation constraint and has higher observation support is retained, and the other subgraph is set as unavailable or pending.

[0038] After propagation stabilizes, the smallest cut set that causes inconsistency is identified, and the ordering of the smallest cut set follows the inconsistency resolution priority rule, which prioritizes deleting loading layer and display layer nodes directly triggered by allele existence constraints, followed by necessary components of the loading complex, followed by mode-specific transport or modification links, and finally, objects marked as pending due to insufficient detection likelihood are down-weighted rather than deleted. The graph editing sequence consists of two types of atomic actions: node pruning, which removes nodes confirmed as unavailable and not subject to observation preservation constraints, and link contraction, which removes breakpoint edges formed by mode separation or allele deletion from the channel and establishes redirection between alternative edges at the same level to maintain connectivity of the remaining channels. After executing the complete graph editing sequence, the individualized graph is output, and an editing log is generated to record the unique identifiers of edited nodes and links, corresponding allele and presentation mode labels, sources and timestamps of triggering evidence, editing action types and execution order, and pre- and post-editing states, which serve as the starting point for causal search in the audit and decision adjustment module.

[0039] In summary, using channels as the basic editing unit and introducing pending down-weighting and observation preservation protection, structural defects are rigidly removed and detection limitations are flexibly handled, avoiding misjudgment of non-random deletions as structural breaks, and ultimately obtaining an individualized graph consistent with the molecular evidence of the case and usable for intervention design.

[0040] The decision adjustment module is used to perform evidence propagation and minimum explanation path reasoning on the individualized graph in combination with the detection likelihood set, outputting defect nodes, defect paths, and evidence chains, and constructing verification and intervention causal graphs, searching for the minimum cost path from the current presentation state to the target state that meets the presentation requirements, and generating clinical decisions for output. The specific implementation is as follows: The personalized atlas and the detection likelihood set are received as inputs, and evidence propagation is performed on the personalized atlas: the mapping of the peptide segment to the presentation site and the surface expression label are taken as observations, and the observation confidence label corresponding to the detection likelihood is written in the evidence weight field of the node and edge; when a certain observation is derived from the sample-level detection parameter or the peptide segment-level detection parameter, the module regards the observation as pending and participates in the propagation in a reduced weight manner.

[0041] The propagation is performed in the order of cleavage, transport, modification, loading, and display, and the interpretation score of the node is composed of three parts: the weighted support amount of the upstream observation evidence, the downstream ability to form a through link consistent with the observation, and the structural penalty amount applied by the copy loss marker, inactivation mutation marker, and domain damage marker related to the node; The interpretation score is used as a criterion, and the minimum interpretation path reasoning is used to select a path set that meets the observation consistency in all reachable paths, and the path that can explain the same observation with fewer nodes and fewer conflicts is preferred, to obtain defect nodes, defect paths, and evidence chains. The defect node refers to the node whose interpretation score is significantly biased towards abnormality due to structural penalty and downstream non-through; the defect path refers to the path containing one or more defect nodes and having contradictory items in the observation consistency matrix; and the evidence chain is an ordered record of the observation source, source identification, timestamp, and version identification supporting each conclusion, which is used for auditing and review.

[0042] After obtaining the defect nodes and defect paths, the module constructs a verification and intervention causal graph and searches for the minimum cost path from the current presentation state to the target state that meets the presentation requirements. The nodes of the causal graph are composed of intervention points and observation outputs, and the edges represent the regulatory or functional dependence relationship confirmed by the personalized atlas. The module sets the cost composition for each intervention action, including intervention cost, inverse measure of expected presentation improvement amplitude, punishment related to observation inconsistency items, and risk measure of potential adverse effects on other channels. All four measures are derived from the comprehensive quantification of the personalized atlas, reachable path set, and observation consistency matrix.

[0043] Heuristic strategies are used in the search process: The priority expansion can simultaneously reduce the number of observation contradictions and increase the intervention nodes of the observation support items in the display layer, and the branches that only rely on pending observations are given a lower priority. When there is a presentation mode conflict, the module respectively evaluates the path cost on the two mode subgraphs and retains the branch with a smaller total cost, and finally outputs the minimum cost path, the corresponding evidence chain, the defect node list and defect path set, the reversibility label, the verification item list and the intervention item list. At the same time, a structured message for clinical decision support output is generated, which includes data version, algorithm version and atlas version identification, evidence source identification and timestamp, and message header and field mapping relationship for medical information system docking.

[0044] In summary, the module internalizes the detection likelihood set as propagation weights and conflict penalties, so that non-random missing is flexibly processed in the decision-making stage. The minimum explanation path reasoning is used to first complete abnormal attribution, and then the path search of executable intervention is completed on the causal graph by using the cost function, so that the explanation and intervention are integrated and form a closed loop with the individualized atlas and the presentation demand determination module.

[0045] The above embodiments can be realized wholly or partially by software, hardware, firmware or any other combination. When realized by software, the above embodiments can be realized wholly or partially in the form of a computer program product.

[0046] Those skilled in the art can realize that the modules and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0047] In addition, the functional modules in each embodiment of the present application can be integrated in one processing module, or each module can exist physically alone, or two or more modules can be integrated in one module.

[0048] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0049] Finally, the above is only the preferred embodiment of the present application and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A map-based reasoning and analysis system for addressing defects in cell antigen processing, characterized in that: It includes a case data mapping module, an antigen assembly module, a presentation requirement determination module, an individual strategy analysis module, and a decision adjustment module, and the modules are connected by signals. The case data mapping module is used to obtain the HLA locus typing, somatic cell variation and copy number, transcriptional expression, immunopeptidomics and surface HLA detection results of cases, and to complete the standardized mapping. The antigen assembly module is used to assemble a knowledge graph of antigen processing and presentation, including lysis, transport, modification, loading, and display, and distinguishes between direct presentation and cross-presentation. The presentation requirement determination module is used to calculate link connectivity and observation consistency on the assembled antigen processing and presentation knowledge graph, and determine whether the presentation requirement is met based on the preset presentation requirement conditions. The individual strategy analysis module is used to determine the association between unavailable channels and presentation patterns based on HLA site copy deletion, inactivation mutation and domain destruction when the conditions are not met. It then uses constraint propagation to generate a map editing sequence, prunes nodes and shrinks links to obtain an individualized map. The decision adjustment module is used to perform evidence propagation and minimum interpretation path reasoning on the individualized graph by combining the detection likelihood set, outputting defect nodes, defect paths and evidence chains, and constructing a verification and intervention causal graph. It searches for the minimum cost path from the current presentation state to the target state that meets the presentation requirements, and generates clinical decisions for output.

2. The atlas reasoning and analysis system for addressing defects in cell antigen processing according to claim 1, characterized in that: The case data mapping module includes: Based on a unified time reference and glossary, the field standardization and aliasing of HLA locus typing, somatic cell variation and copy number, transcript expression, immunopeptidomics and surface HLA detection results were standardized and compared with aliases. Allele nomenclature standardization and heterozygosity consistency verification were performed on HLA locus typing. Genome coordinate alignment and functional annotation were performed on somatic cell variations and copy numbers, and markers of copy deletion, inactivation mutations and domain disruption were extracted. Gene expression status tags are generated from transcriptional expression. Immunopeptidomics was used to perform spectroscopic identification and confidence screening, and to map peptides to presentation sites. Surface HLA detection was numerically normalized and surface expression tags were generated. A detection likelihood set is established based on the sample input amount, peptide type, and experimental procedure parameters, and then output along with the label.

3. The atlas reasoning and analysis system for addressing defects in cell antigen processing according to claim 2, characterized in that: The module for determining presentation requirements includes: On the assembled antigen processing and presentation knowledge graph, the set of reachable paths from input to output is calculated in the order of cleavage, transport, modification, loading, and display. The labels output by the case data mapping module are compared with the corresponding nodes to form an observation consistency matrix; Based on the preset presentation requirements, the set of reachable paths and the observation consistency matrix are jointly determined, and the satisfaction result is output.

4. The atlas reasoning and analysis system for addressing defects in cell antigen processing according to claim 3, characterized in that: The individual strategy analysis module includes the following when it determines that the presentation requirements are not met: HLA site copy deletion, inactivation mutation, and domain disruption markers were extracted from the case data mapping results. The markers are projected onto the nodes and links associated with alleles to identify the set of unavailable channels and to build a channel association mapping table. Based on the channel association mapping table, constraint propagation is used to generate a graph editing sequence, which performs pruning on associated nodes and shrinking on dependent links to obtain an individualized graph.

5. The atlas reasoning and analysis system for addressing defects in cell antigen processing according to claim 4, characterized in that: The graph editing sequence generated using constraint propagation includes: A constraint-satisfying network is constructed using node states and edge channel states as variables. Inject node predecessor availability, loading complex integrity, and allele existence as constraints into the network. Perform constraint propagation to reduce the feasible domain of variables and identify the minimum cut set that leads to inconsistency; The editing actions are sorted according to topological order and inconsistency resolution priority rules, and the graph editing sequence is output.

6. The atlas reasoning and analysis system for addressing defects in cell antigen processing according to claim 4, characterized in that: The decision adjustment module includes: Evidence propagation is performed on the individualized graph by combining the detection likelihood set and calculating the interpretation score of each node and each path; A verification and intervention causal graph is constructed based on the intervention points and the observed outputs. Nodes represent intervention points and observed outputs, and edges represent regulation or functional dependencies. A path cost function is constructed based on intervention costs, expected incremental increases, and penalties for inconsistencies in evidence. Heuristic search is used to search for the minimum cost path from the current presentation state to the target state that satisfies the presentation requirements in the causal graph; Output the minimum cost path and its corresponding evidence chain, and generate clinical decision support output.

7. A map-based reasoning and analysis system for addressing defects in cell antigen processing according to claim 2, characterized in that: The detection likelihood set established by the case data mapping module includes: Generate sample-level detection parameters based on the sample input volume, preprocessing method, and batch size. Peptide-level detection parameters are generated based on peptide charge-to-mass ratio range, retention time deviation, fragment ion coverage, identification confidence level, and modification type. Surface expression detection parameters are generated based on antibody cloning, labeling channel overflow correction, and instrument gain for surface HLA detection. Sample-level detection parameters, peptide-level detection parameters, and surface expression detection parameters are mapped to observation confidence labels and bound to corresponding nodes or observation records.

8. The atlas reasoning and analysis system for addressing defects in cell antigen processing according to claim 1, characterized in that: When distinguishing between direct presentation and cross-presentation, the antigen assembly module performs the following actions: Based on cell type labels and presentation pattern markers, evidence is propagated only within subplots that match them; Using the presentation layer nodes as the merge nodes, the consistency of the results from the direct presentation subgraph and the cross presentation subgraph is checked. When the verification is inconsistent, the conflict path is marked and a presentation mode conflict prompt is output for graph editing in the individual strategy analysis module.

9. A map-based reasoning and analysis system for addressing defects in cell antigen processing according to claim 4, characterized in that: The edit log includes: The unique identifier of the edited node and link, the corresponding allele, and the presentation pattern marker; The source of evidence that triggered the edit, the timestamp, and the version identifier; Edit action types and execution order; edit the status of nodes and links before and after editing. Rollback markers and traceability check codes used for recovery.

10. A map-based reasoning and analysis system for addressing defects in cell antigen processing according to claim 6, characterized in that: Structured messages include: List of defective nodes, set of defective paths, and corresponding chains of evidence; Reversibility labeling, validation program list, and intervention program list; Data version, algorithm version, and graph version identifiers; evidence source identifiers and timestamps; This is used for mapping message headers and fields for integration with medical information systems, and provides backfilling and auditing interfaces to record the status of receiving and calling data.

Images