A rapid sequencing and source tracing analysis system for imported infectious diseases

Abstract

This invention discloses a rapid sequencing and source tracing analysis system for imported infectious diseases, relating to the field of customs quarantine. The system includes a sample preprocessing unit, a high-throughput sequencing unit, a data processing and quality control unit, a rapid alignment and annotation unit, and a source tracing analysis unit. The rapid sequencing and source tracing analysis system for imported infectious diseases proposed in this application forms a complete chain from sample preprocessing, library construction, and data cleaning to pathogen identification and transmission map construction, avoiding module fragmentation and manual intervention, and facilitating automated task completion in large-scale, emergency scenarios.

Description

Technical Field

[0001] This specification pertains to the field of customs quarantine; more specifically, this application relates to a rapid sequencing and source tracing analysis system for imported infectious diseases. Background Technology

[0002] In recent years, with frequent international exchanges and the occurrence of global pandemics, imported infectious diseases have become a major public health security risk. Especially in areas with high population density and international connections, such as ports, border crossings, and harbors, rapid identification, accurate tracing, and risk warning of inbound personnel or goods carrying unknown pathogens have become important means to prevent the spread of imported epidemics.

[0003] Existing procedures for detecting and analyzing imported pathogens mainly rely on the following technological approaches: 1. Traditional PCR or antigen detection methods: Although they are fast and low-cost, they can only target known pathogens and cannot cover emerging or mutated pathogens, resulting in limited sensitivity and broad spectrum.

[0004] 2. Conventional high-throughput sequencing and manual analysis workflow: Currently, some research institutions have carried out whole-genome detection based on high-throughput sequencing, but this workflow usually has the following problems: poor sample information structure and lack of semantic coding, making it impossible to express the sample source background; the sequencing result analysis links are scattered and lack acceleration mechanisms, resulting in low overall processing efficiency; pathogen identification and transmission chain construction are fragmented, making it difficult to support rapid response and epidemic trend judgment; poor systematicity, making it impossible to output decision-making elements such as maps, risks, and node indicators in an integrated manner.

[0005] Furthermore, in real-world scenarios, such as ports where the daily sample volume is large and testing timeliness is critical, traditional methods struggle to simultaneously ensure timeliness, accuracy, and traceability throughout the entire process, necessitating a more systematic, intelligent, and structured technical solution. Summary of the Invention

[0006] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. This summary section is not intended to limit the key and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.

[0007] Firstly, this application proposes a rapid sequencing and source tracing analysis system for imported infectious diseases, comprising: The sample preprocessing unit is used to identify and encode the preprocessed sample to form preprocessed sample information. The sample processing includes nucleic acid extraction, purification and quality detection. The high-throughput sequencing unit is used to construct sequencing libraries from the preprocessed sample information and obtain raw sequencing data. The data processing and quality control unit is used to perform quality control, adapter removal, and host sequence removal on the above raw sequencing data to obtain effective pathogen sequencing data. The rapid alignment and annotation unit is used to align the above-mentioned effective pathogen sequencing data with a preset reference database to obtain alignment and annotation results; The source tracing analysis unit is used to construct an evolutionary relationship map between pathogens based on the above comparison and annotation results.

[0008] In one feasible implementation, the specific steps of identifying and encoding the preprocessed sample to form preprocessed sample information include: Based on the personnel information corresponding to the sample to be tested after the above sample processing, generate identity information coding segment and time coding segment; Environmental data is embedded in the coding segment based on port passenger flow level information and historical disease level; An extended coding segment is generated based on the vaccination information and travel itinerary information of the personnel corresponding to the above-mentioned test samples; The aforementioned identity information encoding segment, time encoding segment, environmental data embedding encoding segment, and extended encoding segment are bound to the processed sample to form the aforementioned preprocessed sample information.

[0009] In one feasible implementation, the specific steps for constructing a sequencing library from the preprocessed sample information and obtaining the raw sequencing data include: The target library type is determined based on the nucleic acid type information, concentration range information, and target pathogen type information of the preprocessed sample information described above. After the target library type is determined, the target library corresponding to the preprocessed sample is established according to the library construction rules corresponding to the target library type. Concentration and fragment distribution information of the target library were obtained through quantitative fluorescence and fragment analysis. The library quality score is calculated based on the above concentration information and fragment distribution information. The target library whose library quality score meets the preset quality is loaded into the sequencing platform to obtain the raw sequencing data.

[0010] In one feasible implementation, the specific steps for performing quality control, adapter removal, and host sequence removal on the above-mentioned raw sequencing data to obtain effective pathogen sequencing data include: The raw sequencing data were subjected to quality assessment, and the base quality distribution, N base ratio, length distribution and GC content deviation of the sequencing reads were analyzed. Low-quality sequencing reads were filtered and tails were pruned based on a set threshold to obtain quality-filtered sequencing data. Load the adapter sequence database of the corresponding library construction platform, use the adapter identification algorithm to match and locate the adapter sequences in the quality-filtered sequencing data, and perform adapter sequence trimming to obtain trimmed sequencing data; The host reference genome database is loaded, and the above-mentioned trimmed sequencing data is compared with the host genome. Sequencing reads originating from the host are identified and removed to obtain the above-mentioned effective pathogen sequencing data.

[0011] In one feasible implementation, the specific steps for comparing the aforementioned effective pathogen sequencing data with a preset reference database to obtain comparison and annotation results include: A pre-defined reference database containing multiple known pathogen reference genome sequences is constructed, and the pre-defined reference database is indexed and preprocessed to form a target index structure, wherein the target index structure is a two-layer index structure composed of a Bloom filter and a hash table. K-mer fragments are extracted from each sequencing read in the above effective pathogen sequencing data, and parallel hash matching is performed on the GPU. The above K-mer fragments are compared with the above target index structure of the above preset reference database, and a hit score for each pathogen is generated based on the number of hits to obtain the initial screening results. Based on the above preliminary screening results and the dynamic limitation of the high-precision alignment range, the reference sequences of the Top-N high-scoring candidate pathogens were selected as targets, and the alignment algorithm was used to perform a mapping operation on the above sequencing reads to obtain the alignment results. The sequencing reads from the successfully aligned results are annotated using an annotation hash table mechanism by calling the annotation module to quickly map virulence factors, resistance genes, and key functional regions, in order to obtain the alignment and annotation results.

[0012] In one feasible implementation, the specific steps for constructing the evolutionary relationship map between pathogens based on the above comparison and annotation results include: Feature extraction is performed on the above comparison and annotation results to obtain feature vectors, which include a list of SNP sites, InDel or amino acid substitution positions, and toxicity or resistance site tags. Construct enhanced variation distance information between any two samples in the above comparison and annotation results, wherein the enhanced variation distance information includes genotype differences, differences in countries of entry, and information on ports of entry; The enhanced mutation distance information is input into the DBSCAN algorithm to perform unsupervised clustering to obtain the clustering results. Construct a minimum spanning tree propagation graph based on the enhanced mutation distance information described above; The above clustering results are mapped to the above minimum spanning tree propagation graph to construct the evolutionary relationship map among the above pathogens.

[0013] In one feasible implementation, the construction of the minimum spanning tree propagation graph based on the enhanced mutation distance information includes: Based on the above enhanced distance information, a graph structure G(V, E) is constructed, where node V represents a sample, edge E represents the potential propagation relationship between samples, and the edge weight is the above enhanced mutation distance information; The minimum spanning tree algorithm is applied to the above graph structure to construct a minimum spanning tree propagation graph between samples.

[0014] In one feasible implementation, mapping the clustering results to the minimum spanning tree propagation graph to construct the evolutionary relationship map among the pathogens includes: Write the above clustering labels into the attributes of each node in the above minimum spanning tree propagation graph, so that each node in the graph contains information about its clustering category; Traverse the edge structure of the minimum spanning tree propagation graph and determine whether the clustering labels of the nodes at both ends of the edge are consistent; If they are inconsistent, mark the edge as a cross-cluster propagation path; The edge weights of the above cross-cluster propagation paths are weighted and adjusted to enhance the ability to represent the risk of propagation jumps in the graph; The output includes an enhanced propagation graph structure containing the clustering labels and cross-cluster connection markers mentioned above, in order to construct an evolutionary relationship map among the pathogens.

[0015] In one feasible implementation, the system further includes a disease transmission path analysis unit, which includes a construction module, a traversal module, a calculation module, and a display module. The aforementioned construction module is used to call the edge weight information, timestamp information, and transmission path connection information in the above-mentioned evolutionary relationship graph between pathogens to construct a transmission time sequence graph between samples; The above traversal module is used to perform graph traversal operations based on the above propagation time sequence graph to determine the propagation subgraph starting from the specified source sample. The above graph traversal operations include breadth-first search or depth-first search. The above calculation module is used to perform propagation characteristic analysis on the nodes of the above propagation subgraph and calculate key propagation node indicators, including node degree, betweenness centrality, and propagation path length. The aforementioned display module is used to display the aforementioned propagation timeline diagram and the aforementioned node indicators in a visualization graph.

[0016] In one feasible implementation, the system further includes a propagation risk assessment unit, which includes a statistics module, an identification module, and a risk calculation module. The aforementioned statistics module is used to count the number of cross-cluster connections and the average edge weight in the evolutionary relationship graph of the pathogens. The aforementioned identification module is used to identify high-frequency regions and high-frequency propagation channel nodes with high-frequency cross-cluster connections in the evolutionary relationship map of the aforementioned pathogens; The aforementioned risk calculation module is used to generate a propagation risk score based on the number of edges, the average edge weight, the high-frequency region, and the high-frequency propagation channel nodes.

[0017] In summary, the rapid sequencing and source tracing analysis system for imported infectious diseases proposed in this application constitutes a complete chain from sample preprocessing, library construction, and data cleaning to pathogen identification and transmission map construction, avoiding module fragmentation and manual intervention, and facilitating automated task completion in large-scale, sudden scenarios. The sample preprocessing stage introduces a dynamically scalable coding strategy, transforming high-order semantics such as population movement, port access, and travel history into structured labels, effectively supporting subsequent sample priority scheduling and semantic enhancement of the transmission map. The rapid alignment unit combines a Bloom filter and a hash table dual-layer structure to achieve a second-level response for initial screening of millions of reads, and improves alignment efficiency through a Top-N candidate constraint strategy, significantly outperforming traditional full alignment algorithms. The source tracing unit not only constructs genotypic difference maps between samples but also integrates information such as entry countries and port access to generate an enhanced transmission relationship graph, supporting more realistic epidemic transmission modeling and abnormal jump detection. The system can automatically calculate indicators such as node degree, betweenness centrality, and path depth, and display key nodes and chains through an interactive transmission map. This system can also be integrated with downstream risk warning platforms and port dispatch systems to optimize the prevention and control mechanism through data-driven mechanisms, such as alarms for high-risk transmission chains and priority ranking of entry inspections.

[0018] Other advantages, objectives and features of this application will be partly apparent from the description below, and partly understood by those skilled in the art through study and practice of this application. Attached Figure Description

[0019] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit this specification. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1This is a structural schematic diagram of a rapid sequencing and source tracing analysis system for imported infectious diseases provided in an embodiment of this application. Detailed Implementation

[0020] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus. The technical solutions of the embodiments of this application will now be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0021] Please see Figure 1 The diagram below illustrates the structure of a rapid sequencing and source tracing system for imported infectious diseases, as provided in this application embodiment. Specifically, it may include: The sample preprocessing unit 101 is used to identify and encode the sample to be tested after sample processing to form preprocessed sample information, wherein the sample processing includes nucleic acid extraction, purification and quality detection; The high-throughput sequencing unit 102 is used to construct a sequencing library from the preprocessed sample information and obtain raw sequencing data. The data processing and quality control unit 103 is used to perform quality control, adapter removal and host sequence removal on the above raw sequencing data to obtain effective pathogen sequencing data. The rapid alignment and annotation unit 104 is used to align the above-mentioned effective pathogen sequencing data with a preset reference database to obtain alignment and annotation results; The source analysis unit 105 is used to construct an evolutionary relationship map between pathogens based on the above comparison and annotation results.

[0022] For example, the sample preprocessing unit 101 is used to perform standardized preprocessing operations on the collected samples to be tested. Its functions include extracting and purifying nucleic acids from the samples, and performing quality testing to ensure the accuracy and stability of subsequent sequencing. Based on the completion of biochemical processing, the system also generates a unique identifier code for each sample. This code can integrate information such as sample source information, epidemiological background of the personnel (such as travel history, port of entry, population flow level, etc.), and timestamps to form "preprocessed sample information".

[0023] Based on the nucleic acid type, concentration range, and target pathogen type in the preprocessed sample information, the high-throughput sequencing unit 102 automatically selects a matching library preparation strategy and sequencing platform to construct a high-quality sequencing library. Samples that pass library preparation are loaded onto the platform (such as Illumina or Nanopore), and the system acquires the corresponding raw sequencing data, providing basic sequence information for subsequent analysis.

[0024] The data processing and quality control unit 103 performs a series of cleaning processes on the raw sequence data output from the sequencing platform. This process includes: sequencing quality control, adapter sequence identification and removal, and alignment with the host reference genome to remove host-source reads. The final output is "effective pathogen sequencing data," which consists of target information sequences used for source tracing and identification analysis.

[0025] The fast alignment and annotation unit 104 is then invoked to efficiently align the aforementioned valid pathogen sequencing data with a pre-constructed pathogen reference database. This module typically combines GPU parallel acceleration technology with a two-layer index structure (such as Bloom filters and hash tables) to first perform initial screening and matching of the data, and then perform high-precision alignment of the Top-N candidate pathogens. Simultaneously, this unit has a built-in annotation module that can functionally label the aligned locations, identify virulence factors, drug resistance genes, key mutation sites, etc., and generate structured "alignment and annotation results".

[0026] Finally, based on the above comparison and annotation results, the source tracing analysis unit 105 extracts variation feature information, constructs an enhanced propagation distance matrix between samples, and combines clustering and minimum spanning tree algorithms to generate an evolutionary propagation relationship map between samples. This map visually displays the propagation paths, variation levels, and possible source tracing directions between different samples, and can be used to support tasks such as entry prevention and control risk assessment, abnormal transmission behavior identification, and epidemic evolution trend analysis.

[0027] In summary, the rapid sequencing and source tracing analysis system for imported infectious diseases proposed in this application constitutes a complete chain from sample preprocessing, library construction, and data cleaning to pathogen identification and transmission map construction, avoiding module fragmentation and manual intervention, and facilitating automated task completion in large-scale, sudden scenarios. The sample preprocessing stage introduces a dynamically scalable coding strategy, transforming high-order semantics such as population movement, port access, and travel history into structured labels, effectively supporting subsequent sample priority scheduling and semantic enhancement of the transmission map. The rapid alignment unit combines a Bloom filter and a hash table dual-layer structure to achieve a second-level response for initial screening of millions of reads, and improves alignment efficiency through a Top-N candidate constraint strategy, significantly outperforming traditional full alignment algorithms. The source tracing unit not only constructs genotypic difference maps between samples but also integrates information such as entry countries and port access to generate an enhanced transmission relationship graph, supporting more realistic epidemic transmission modeling and abnormal jump detection. The system can automatically calculate indicators such as node degree, betweenness centrality, and path depth, and display key nodes and chains through an interactive transmission map. This system can also be integrated with downstream risk warning platforms and port dispatch systems to optimize the prevention and control mechanism through data-driven mechanisms, such as alarms for high-risk transmission chains and priority ranking of entry inspections.

[0028] In one feasible implementation, the specific steps of identifying and encoding the preprocessed sample to form preprocessed sample information include: Based on the personnel information corresponding to the sample to be tested after the above sample processing, generate identity information coding segment and time coding segment; Environmental data is embedded in the coding segment based on port passenger flow level information and historical disease level; An extended coding segment is generated based on the vaccination information and travel itinerary information of the personnel corresponding to the above-mentioned test samples; The aforementioned identity information encoding segment, time encoding segment, environmental data embedding encoding segment, and extended encoding segment are bound to the processed sample to form the aforementioned preprocessed sample information.

[0029] For example, the system obtains the personnel information corresponding to the sample to be tested, including their ID card number or passenger identification number. Based on this information, a hash function is used to generate a unique identity information encoding segment, for example, using SHA-256 truncated to the first 8 bits as an individual identity identifier.

[0030] Simultaneously, the sample collection time (accurate to the hour or minute) is obtained and formatted as a timestamp (e.g., YYYYMMDDHHmm) to generate a time-coded segment. The combination of these two methods can be used to uniquely identify a sample at a specific time and from a specific source.

[0031] For example: Identity information: Passenger ID = ZL2345XH89, hash is 4E91A5C2, sampling time: April 5, 2025, 13:26 → time encoding segment is 02504051326, generated encoding prefix: 4E91A5C2-202504051326.

[0032] The system calls the port's real-time database to obtain the passenger flow level information of the port where the sample entered the country. For example, port number T3 corresponds to "high passenger flow level" and is coded as H. At the same time, it queries the historical epidemic database. If there are confirmed cases at the port in the past 7 days, the epidemic level is "medium" and is coded as M. The above two types of information are combined to generate environmental data and embedded in the coding segment, for example: HM (High flow + Medium epidemic level).

[0033] The following two types of information are extracted from the inbound traveler health declaration system: 1. Vaccination information: If all three doses of vaccination have been completed, it is coded as V3; if not, it is coded as V0. 2. Travel itinerary information: If the traveler has stayed in a country with a high incidence of COVID-19 within the past 14 days, such as Italy (IT), the country code is recorded; if the travel itinerary is normal, it is coded as NA. These are then combined to form an extended coding segment: for example, V3-IT. The four segments of information are concatenated sequentially to form the final sample's identifier string: <Identity Information Encoding Segment> - <Time Encoding Segment> - <Environmental Data Segment> - <Extended Segment> → 4E91A5C2-202504051326-HM-V3-IT. The system writes this encoding into the database and binds it to the sample's barcode or electronic tag, using it as a unique sample identification ID in subsequent processes and for functions such as task scheduling priority ranking, tracking, and source mapping.

[0034] Through this implementation method, the system can achieve multi-dimensional semantic identification of samples, significantly improve the sample priority screening capability and intelligent scheduling efficiency in high-concurrency scenarios, and at the same time ensure that the source and characteristics of samples are traceable and analyzable in the subsequent tracing process.

[0035] In one feasible implementation, the specific steps for constructing a sequencing library from the preprocessed sample information and obtaining the raw sequencing data include: The target library type is determined based on the nucleic acid type information, concentration range information, and target pathogen type information of the preprocessed sample information described above. After the target library type is determined, the target library corresponding to the preprocessed sample is established according to the library construction rules corresponding to the target library type. Concentration and fragment distribution information of the target library were obtained through quantitative fluorescence and fragment analysis. The library quality score is calculated based on the above concentration information and fragment distribution information. The target library whose library quality score meets the preset quality is loaded into the sequencing platform to obtain the raw sequencing data.

[0036] For example, based on the sample metadata passed from the sample preprocessing module, the system automatically completes the entire process of library type determination, library construction, quality control, and sequencing data acquisition. Sample metadata includes: nucleic acid type information (e.g., DNA or RNA), concentration range information (e.g., 8–50 ng / μL), and target pathogen type information (e.g., virus, bacteria, or fungus). The system has a built-in library construction rule library that can automatically match a suitable sequencing platform and corresponding library construction scheme based on the above information.

[0037] For example, when the sample is RNA, the target pathogen is a virus, and high sequencing accuracy is required, the system will match the strand-specific transcription library preparation protocol of the Illumina platform; while when the sample is DNA, and the target is whole genome coverage or the response speed of the tracing task is high, the full-length fragment library preparation protocol of the Nanopore platform will be preferentially matched. For example, if a sample is RNA with a concentration of 12.4 ng / μL and the target pathogen is a coronavirus, the system will automatically select the TruSeq RNALibrary protocol and allocate it to the Illumina sequencing platform.

[0038] Once the target library type is determined, the system calls the corresponding library construction module based on the matching results to complete the library construction operation. If the Illumina platform library construction scheme is selected, the process includes nucleic acid fragmentation (by enzyme digestion or sonication), end repair, A-tailing, sequencing adapter ligation, PCR amplification of the library, and purification and recovery using magnetic beads. If the Nanopore platform library construction scheme is selected, the process is simplified to skip fragmentation, directly repairing the ends of long DNA and ligating sequencing adapters with barcode tags, while omitting the PCR amplification step to preserve the original fragment length information.

[0039] The entire cytology preparation process is completed through an automated pipetting system (such as a liquid workstation), which simultaneously records the batch number, parameters used, consumable numbers, and metadata of the operation process to ensure traceability and standardization.

[0040] After library construction is complete, the system automatically performs quality checks. This process includes measuring the overall concentration of the library using quantitative fluorescence instruments (such as Qubit) and obtaining fragment length distribution, main peak fragment size, and peak shape uniformity indices using fragment analyzers (such as Bioanalyzer or FragmentAnalyzer). The system inputs these test results into a built-in quality scoring model to calculate the Library Quality Score (LQS) and determines whether the library meets the requirements for use on the instrument.

[0041] Document quality rating It can be calculated based on the following formula: ; ; ; in, For penalty items (such as pollution, failure, etc.). To prevent small constants with a denominator of zero.

[0042] The concentration standard score (range 0-1) represents the degree of match between the measured library concentration and the platform's recommended optimal concentration. The calculation formula is as follows: ; in, To measure the concentration, This is the recommended concentration. When Falling When the value is within the specified range, it is considered optimal, and the score can be directly assigned as 1.

[0043] The main peak accuracy score (range 0-1) indicates the degree of similarity between the length of the main peak fragment in the library and the target length. The calculation formula is as follows: ; in, The actual length of the main peak. For the target length, This is a moderating factor that reflects the sensitivity of the score to the deviation in segment length.

[0044] The peak width variation score, used to characterize the concentration of the segment distribution, is defined as the coefficient of variation of the main peak, i.e.: ; in, The main peak width (standard deviation) The CV value indicates the length of the main peak. A smaller CV value indicates a more concentrated fragment and higher library quality.

[0045] The regularization term is used to avoid the case where the denominator is zero in division operations. This is a non-linear enhancement index used to control the degree of non-linear enhancement of the overall score. It is suitable for boosting, compressing, or differentiating and amplifying the score results. It can be set to a value between 1.2 and 2.0 and can be optimized based on data experience.

[0046] Scoring based on concentration standards, For the quality consistency angle, if the concentration is high, the main peak is concentrated and the peak width is small, and the three factors are consistent, then... , This indicates consistent quality; if a certain item deviates significantly, the angle increases, and the score is suppressed.

[0047] Generally, if the LQS score is higher than 85 (out of 100), it is considered a qualified library and can proceed to the sequencing process; if the score is lower than the threshold, the system will automatically mark the sample as "requiring library reconstruction" and return to the library construction process to be re-executed.

[0048] For libraries that pass the scoring, the system loads them into the corresponding sequencing platform and configures the necessary parameters to start the sequencing task. For example, in Illumina platforms (such as MiSeq or NextSeq), the system will set sample index tags, read modes (such as PE150), etc.; while in Nanopore platforms (such as GridION), the libraries are directly loaded into the sequencing machine, and the system will monitor their sequencing status in real time.

[0049] During sequencing, the system continuously records key performance indicators, including the number of valid reads, error rate, and Q30 ratio. After sequencing is complete, the system exports raw sequence data files in either FASTQ (containing sequencing reads and quality values) or BAM (aligned format). Finally, the sequencing data is bound to the corresponding sample identifier and uploaded to a central database, providing data support for subsequent data cleaning, pathogen identification, and source tracing analysis modules.

[0050] This implementation method enables sample-specific automatic library construction strategy selection, automated execution of high-throughput library construction processes, fine-grained scoring control of library quality, and seamless integration of multi-platform sequencing tasks, significantly improving the overall system's processing efficiency, sequencing success rate, and data reliability in rapid response to imported infectious diseases.

[0051] In one feasible implementation, the specific steps for performing quality control, adapter removal, and host sequence removal on the above-mentioned raw sequencing data to obtain effective pathogen sequencing data include: The raw sequencing data were subjected to quality assessment, and the base quality distribution, N base ratio, length distribution and GC content deviation of the sequencing reads were analyzed. Low-quality sequencing reads were filtered and tails were pruned based on a set threshold to obtain quality-filtered sequencing data. Load the adapter sequence database of the corresponding library construction platform, use the adapter identification algorithm to match and locate the adapter sequences in the quality-filtered sequencing data, and perform adapter sequence trimming to obtain trimmed sequencing data.

[0052] The host reference genome database is loaded, and the above-mentioned trimmed sequencing data is compared with the host genome. Sequencing reads originating from the host are identified and removed to obtain the above-mentioned effective pathogen sequencing data.

[0053] For example, the system receives raw sequencing data from the sequencing platform, typically provided in FASTQ format. This format contains each sequencing read and its corresponding base quality value (i.e., Phred quality score). The system first performs a comprehensive quality assessment and cleaning of this raw data to ensure the accuracy and validity of the data for downstream analysis.

[0054] During the quality assessment phase, the system primarily analyzes the following indicators: 1. The quality distribution of each base, with particular attention to the ratio of Q20 (quality value ≥ 20) and Q30 (quality value ≥ 30), which reflects the overall sequencing accuracy; 2. The proportion of N bases: N bases represent uncertain base sites, and an excessively high proportion will affect the reliability. 3. Analyze the length distribution of sequencing fragments to determine if there are any abnormally short fragments; 4. GC content deviation: Compare with the reference GC composition of the target species to identify possible contamination or library construction bias.

[0055] Based on the analysis, the system implements the following quality filtering strategy: 1. Use a sliding window algorithm to perform local quality trimming on reads. For example, set trimming to be triggered when the average quality value of four consecutive bases is lower than 20. 2. Delete short reads whose length is less than a set threshold (e.g., 50 bp); 3. Remove low-quality reads with an N base content exceeding 10%; 4. Optional poly-G tail trimming, specifically designed for pseudo-tail problems common in platforms such as Illumina NovaSeq.

[0056] After the above processing, the retained high-quality reads are output as the `clean_qc.fastq` file. At the same time, the system generates and saves the quality control log for subsequent traceability and analysis (this process can be completed through FastQC or the self-developed quality control module).

[0057] After quality control is completed, the system proceeds to the adapter identification and removal step. First, based on the library construction platform information recorded during the sample preprocessing stage, a matching adapter sequence database is automatically selected. For example: 1. If the sample uses the Illumina TruSeq library construction scheme, then call the corresponding TruSeq connector sequence; 2. If using the Nanopore Native Barcoding solution, load its dedicated connector sequence; 3. The system also supports specific connector sequence configurations for other platforms (such as MGI, Ion Torrent, etc.).

[0058] During the adapter identification phase, the system invokes existing sequence processing tools (such as Cutadapt and Trimmomatic) to perform adapter sequence alignment at the 5' and 3' ends of each read. The matching method supports both exact matches and fuzzy matching strategies with a maximum of 1-2 base mismatches to enhance the sensitivity of the identification. Successfully matched adapter sequences are pruned, retaining only the valid target sequence region in the middle.

[0059] After processing, the system outputs the reads file `clean_adapter.fastq` after connector removal, and records metadata such as connector location and connector type for each read, for subsequent traceability and quality control statistics.

[0060] Subsequently, the system performs host sequence removal on the adapter-processed data. The system first loads the genome sequence of the target host species from its built-in reference database. It then uses efficient alignment algorithms, such as Bowtie2, BWA, or minimap2, to align the reads in `clean_adapter.fastq` with the reference host genome. Strict parameter constraints are applied during the alignment process, such as setting an alignment homology greater than 90% and a sequence coverage greater than 50% as matching thresholds. Reads that completely or partially match the host sequence are marked as "host source" and removed from the dataset.

[0061] Ultimately, reads that were not aligned to the host genome were retained as valid pathogen sequencing data. The system outputs these as a `clean_hostfree.fastq` file and simultaneously generates a host removal statistics report, which includes the percentage of host reads, the number of retained reads, and the removal rate, for use in quality traceability and automated evaluation mechanisms.

[0062] Through the above three-stage processing flow, the system can effectively clean the raw sequencing data, retain the true, reliable and analytically valuable pathogen sequencing information, and provide a reliable data foundation for subsequent pathogen identification, variant annotation and source tracing analysis.

[0063] This implementation method significantly improves the effectiveness and relevance of sequencing data through a multi-step quality control and decontamination process, ensuring that the data entering the subsequent alignment and annotation modules has high fidelity and low background contamination characteristics, laying a data foundation for accurate pathogen identification and source tracing analysis.

[0064] In one feasible implementation, the specific steps for comparing the aforementioned effective pathogen sequencing data with a preset reference database to obtain comparison and annotation results include: A pre-defined reference database containing multiple known pathogen reference genome sequences is constructed, and the pre-defined reference database is indexed and preprocessed to form a target index structure, wherein the target index structure is a two-layer index structure composed of a Bloom filter and a hash table. K-mer fragments are extracted from each sequencing read in the above effective pathogen sequencing data, and parallel hash matching is performed on the GPU. The above K-mer fragments are compared with the above target index structure of the above preset reference database, and a hit score for each pathogen is generated based on the number of hits to obtain the initial screening results. Based on the above preliminary screening results and the dynamic limitation of the high-precision alignment range, the reference sequences of the Top-N high-scoring candidate pathogens were selected as targets, and the alignment algorithm was used to perform a mapping operation on the above sequencing reads to obtain the alignment results. The sequencing reads from the successfully aligned results are annotated using an annotation hash table mechanism by calling the annotation module to quickly map virulence factors, resistance genes, and key functional regions, in order to obtain the alignment and annotation results.

[0065] For example, the system integrates reference genome sequences of known pathogens from multiple authoritative public databases (such as NCBI RefSeq, GISAID, VFDB, etc.), covering multiple species and different strains of viruses, bacteria, and fungi. Each reference sequence is accompanied by relevant metadata, including species name, strain number, country of origin, etc. All reference sequences are standardized to FASTA format and uniformly constructed into a pathogen reference database.

[0066] To improve comparison efficiency and reduce false recognition rate, the system performs index preprocessing on the reference database and constructs a two-level index structure: The first layer is a Bloom filter, which generates a fixed-length set of k-mers (e.g., k=31) for each reference species to construct a spatially compressed Bloom filter, which is used to quickly determine whether a certain k-mer may come from a certain species in the initial screening stage; The second layer is a hash table, which establishes a precise hash mapping for each k-mer, recording its corresponding species ID and location information in the reference sequence. This structure supports parallel access on the GPU, facilitating high-speed queries of large-scale data.

[0067] This two-level index structure effectively combines space efficiency and query accuracy, significantly reducing redundant comparison calculations while ensuring initial screening speed.

[0068] During the alignment phase, the system first performs k-mer extraction on each read in the valid pathogen sequencing data. It uses a sliding window method to extract all possible fixed-length k-mers (e.g., 31 bp) with bases as the step size, forming the k-mer vector sequence for each read.

[0069] Next, the system transmits these k-mer vectors in batches to the GPU, initiating the GPU parallel hash matching process. The comparison process consists of two stages: first, a Bloom filter is used to quickly eliminate reference species that are unlikely to be matched; then, the remaining k-mers are precisely matched in the hash table, the matching position and its corresponding species are recorded, and the number of times each reference pathogen is matched is counted.

[0070] Based on the number of hits, the system calculates a pathogen score vector for each read, representing its probability of matching across different candidate species. The higher the score, the more likely the sample is to contain the corresponding pathogen sequence.

[0071] Based on this, the system selects the Top-N candidate pathogens (e.g., Top-3) according to their scores, and proceeds to the next step of high-precision comparison. This strategy significantly reduces the computational resources required for subsequent comparisons and improves the overall system throughput efficiency.

[0072] During the high-precision alignment stage, the system calls efficient alignment algorithms, such as BWA-MEM (Burrows-WheelerAligner - Maximal Exact Matches), to map the reads to be tested onto the Top-N reference sequences and generate standard alignment output files (such as SAM or BAM format), recording the alignment position, quality score, mutation information, and other information for each read.

[0073] After accurate alignment, the system enters the functional annotation stage. In this stage, the system calls the built-in annotation hash table module, which pre-establishes a mapping relationship between specific sites in the reference sequence and functional regions, including virulence factors, drug resistance genes, and characteristic variant regions. The system can quickly locate each successfully aligned read to its corresponding functional region without needing to access a large database again.

[0074] For each aligned read, the system can quickly determine whether it falls into a specific functional region, such as whether it hits a drug resistance gene (e.g., blaKPC, mecA), whether it is located in a typical virulence factor region (e.g., the Type III secretion system), or whether it contains a specific SNP mutation (e.g., D614G).

[0075] The annotation results are output in a structured format, such as JSON, for use by downstream modules of the system, such as source tracing analysis, cluster construction, and propagation risk assessment.

[0076] The method proposed in this embodiment achieves fast matching and low false recognition rate in the initial screening process through a two-layer index structure of Bloom filter and hash table. Based on GPU parallel execution, it significantly improves the initial screening speed for millions of reads. It adopts a dynamic Top-N screening mechanism to effectively avoid invalid comparisons and save computing resources. It also utilizes an annotation hash mapping mechanism to achieve real-time and fast annotation of key functional areas.

[0077] In one feasible implementation, the specific steps for constructing the evolutionary relationship map between pathogens based on the above comparison and annotation results include: Feature extraction is performed on the above comparison and annotation results to obtain feature vectors, which include a list of SNP sites, InDel or amino acid substitution positions, and toxicity or resistance site tags. Construct enhanced variation distance information between any two samples in the above comparison and annotation results, wherein the enhanced variation distance information includes genotype differences, differences in countries of entry, and information on ports of entry; The enhanced mutation distance information is input into the DBSCAN algorithm to perform unsupervised clustering to obtain the clustering results. Construct a minimum spanning tree propagation graph based on the enhanced mutation distance information described above; The above clustering results are mapped to the above minimum spanning tree propagation graph to construct the evolutionary relationship map among the above pathogens.

[0078] In one feasible implementation, the construction of the minimum spanning tree propagation graph based on the enhanced mutation distance information includes: Based on the above enhanced distance information, a graph structure G(V, E) is constructed, where node V represents a sample, edge E represents the potential propagation relationship between samples, and the edge weight is the above enhanced mutation distance information; The minimum spanning tree algorithm is applied to the above graph structure to construct a minimum spanning tree propagation graph between samples.

[0079] In one feasible implementation, mapping the clustering results to the minimum spanning tree propagation graph to construct the evolutionary relationship map among the pathogens includes: Write the above clustering labels into the attributes of each node in the above minimum spanning tree propagation graph, so that each node in the graph contains information about its clustering category; Traverse the edge structure of the minimum spanning tree propagation graph and determine whether the clustering labels of the nodes at both ends of the edge are consistent; If they are inconsistent, mark the edge as a cross-cluster propagation path; The edge weights of the above cross-cluster propagation paths are weighted and adjusted to enhance the ability to represent the risk of propagation jumps in the graph; The output includes an enhanced propagation graph structure containing the clustering labels and cross-cluster connection markers mentioned above, in order to construct an evolutionary relationship map among the pathogens.

[0080] For example, based on sequencing alignment and functional annotation results, the system constructs a pathogen evolutionary relationship map that integrates genetic variation characteristics and epidemiological information to support transmission chain reconstruction and source tracing analysis. This process includes five key steps: feature extraction, enhanced distance calculation, unsupervised clustering, transmission map construction, and map output.

[0081] First, the system performs structured analysis on the alignment and annotation results of each sample, extracts the following key variation information, and constructs a unified feature vector representation: 1. SNP site list: Records the single base mutation positions (e.g., G23403A) that occur in the sample in the reference sequence compared to the standard strain.

[0082] 2. InDel information: includes the starting position of the insertion or deletion, the length, and the mutation type.

[0083] 3. Amino acid substitution information: such as D614G, P681R, etc., indicating characteristic sites where substitutions have occurred at the protein level.

[0084] 4. Functional annotation tags: used to mark whether a sample hits a virulence factor, drug resistance gene or other key functional region, usually represented in the form of a sparse binary vector.

[0085] All features are uniformly encoded into structured feature vectors. , denoted as: ; Structured feature vectors Used for subsequent inter-sample difference calculation. After feature extraction is complete, the system performs analysis on any two samples. and Calculate the enhanced variant distance. This distance function comprehensively considers genotypic differences, differences in countries of entry, and port information, and is defined as follows: ; in, Hamming distance, based on SNP / InDel / functional annotation vectors, is used to measure the degree of genetic diversity. For the country difference item, 1 is assigned if the country of entry is different, and 0 is assigned if they are the same. This is the port difference item; 1 indicates a different port, and 0 indicates the same port. , , As a weighting factor, controlling the relative contribution of the three indicators, it can be set to... Other adjustments will balance the impact of regional factors. This is a non-linear enhancement index for genetic distance, used to amplify the effects of genetic differences, for example... , The distance normalization exponent controls the overall shape of the function. It can be used to compress extreme value differences. For joint structure penalty terms, it represents the consistency correction amount for propagation paths between samples.

[0086] ; in, This represents the propagation similarity direction angle between sample feature vectors; That is, standard cosine similarity. , These are the feature vectors of two samples; This is the structural penalty coefficient, controlling the incremental penalty when the propagation directions are inconsistent (e.g., 0.3~0.5); if the two samples have highly consistent mutation directions (i.e., the angle between the directions is small, and the cosine value is close to 1), then... If the value is close to 0, no penalty is imposed; if the two samples show a significant difference in the direction of variation (small cosine value), it indicates that the propagation logic is weak, and the overall distance is widened.

[0087] The system constructs a distance matrix from the enhanced variation distances between all samples, inputs it into the DBSCAN unsupervised clustering algorithm, and sets an appropriate distance threshold. The clustering results, including the minimum number of points parameter minPts, are output as propagation clusters composed of multiple similar samples (such as ClusterA, ClusterB, etc.), as well as some outliers, used to identify anomalous propagation samples. Each sample is assigned a cluster label, representing the structure of the propagation group to which it belongs.

[0088] Building upon this foundation, the system further constructs a transmission graph structure between pathogens. First, a weighted undirected graph G(V, E) is constructed, with samples as nodes V and the enhanced distance between each pair of samples as edge weights E. Then, the system applies Prim's algorithm to this graph to generate a Minimum Spanning Tree (MST) that connects all samples and has the minimum total edge weight. In this tree, each edge represents the most probable transmission path between two samples, with smaller edge weights indicating a higher probability of transmission.

[0089] To enhance the expressive power of the graph structure for propagation risk, the system maps DBSCAN clustering results back to a minimum spanning tree. The clustering label of each sample node is written into the node attribute field of the graph. Simultaneously, the system traverses all edges in the tree; if the clustering labels connecting two nodes are different, the edge is marked as a "cross-cluster propagation edge," indicating a potential propagation jump or multi-source crossover. To enhance risk expressiveness, the system can adjust the edge weights of these edges (e.g., by increasing the weight by 1.0) to highlight their jump characteristics in propagation path calculation and visualization. Finally, the system outputs the generated propagation graph structure in a standard graph format (such as GraphML or JSON). Nodes in the output graph represent samples, accompanied by clustering labels, entry country, port information, and mutation characteristics. Edges represent propagation paths, accompanied by attributes such as enhanced mutation distance and whether it is a cross-cluster edge. The graph can be used for tasks such as propagation chain visualization, origin tracing, jump path identification, and propagation risk assessment.

[0090] This implementation method achieves more realistic modeling of inter-sample transmission distance by integrating genetic variation and epidemiological background information between samples. By combining DBSCAN clustering with minimum spanning tree, the constructed transmission map has structural rationality, transmission interpretability, and jump recognition capabilities. It is particularly suitable for reconstructing cross-border epidemic transmission chains, assisting in the evaluation of entry prevention and control strategies, and prioritizing the handling of high-risk samples.

[0091] In one feasible implementation, the system further includes a disease transmission path analysis unit, which includes a construction module, a traversal module, a calculation module, and a display module. The aforementioned construction module is used to call the edge weight information, timestamp information, and transmission path connection information in the above-mentioned evolutionary relationship graph between pathogens to construct a transmission time sequence graph between samples; The above traversal module is used to perform graph traversal operations based on the above propagation time sequence graph to determine the propagation subgraph starting from the specified source sample. The above graph traversal operations include breadth-first search or depth-first search. The above calculation module is used to perform propagation characteristic analysis on the nodes of the above propagation subgraph and calculate key propagation node indicators, including node degree, betweenness centrality, and propagation path length. The aforementioned display module is used to display the aforementioned propagation timeline diagram and the aforementioned node indicators in a visualization graph.

[0092] For example, the propagation path analysis unit in the system consists of a construction module, a traversal module, a calculation module, and a display module, which aims to construct a time-series propagation structure diagram based on the pathogen evolution relationship map and to track and analyze potential propagation chains.

[0093] First, the construction module constructs a propagation time sequence diagram based on the edge weight information in the pathogen evolution relationship graph, the sample collection timestamp information, and the propagation path connection relationship between nodes. Each node in the graph Corresponding to a sample obtained from actual sampling; edge This indicates the existence of a potential propagation relationship between samples; the direction indicates the order of propagation probability, and the weight reflects the propagation strength. (Node acquisition time) It is recorded as a timestamp and used to constrain the directionality of edges, making the graph structure conform to temporal causal logic.

[0094] The weights of the propagation edges are derived from the enhanced mutation distances defined in the system. Furthermore, to improve the temporal rationality of propagation modeling, the system introduces a time factor correction term based on the original mutation distance. For example, edge weights... It can be defined as: ; in, Indicates the sampling time difference between samples. Adjust the weighting coefficient for the time difference.

[0095] The traversal module performs graph traversal operations on the propagation sequence graph to trace the possible propagation paths of any given source sample node. This module supports classic graph traversal algorithms, such as Breadth-First Search (BFS) and Depth-First Search (DFS), which can start from the earliest or most powerful node and gradually expand its potential influence range. During the traversal, the system dynamically generates a propagation subgraph starting from the source node. This indicates the upstream and downstream transmission network of the sample in the epidemic chain.

[0096] The computation module performs propagation characteristic analysis on the nodes in the propagation subgraph, including but not limited to the following key graph metrics: 1. Node Degree: Represents the number of direct propagation connections between a node and other nodes. If a node... The outgoing range (spread to others) is In-degree (being spread by others) is Then the total degree of the node is Nodes with higher degrees are more likely to be points of convergence or diffusion in propagation.

[0097] 2. Betweenness Centrality: Measures a node's ability to act as a hub for information exchange. For all node pairs... The number of shortest paths is denoted as , which passes through nodes The number of paths is ,but: ; The higher the index, the more likely the node is in a "bridge" position in the propagation path, and may be a potential key propagation node or super-spreader.

[0098] 3. Path Length: The shortest path length from a specified source sample node to any other node, used to assess the propagation depth of a sample in the propagation chain or its distance from the source.

[0099] Finally, the system provides a graphical representation of the transmission map and its analytical indicators through a display module. This module supports interactive visualization; in the output transmission map, node colors identify transmission clusters or stages, node size can be dynamically adjusted based on their degree or betweenness centrality, and edge thickness and direction reflect transmission intensity and directionality. This graphical display also supports interactive operations such as clicking on specific nodes, filtering path chains, and expanding cluster structures. This visualization function helps users intuitively understand the pathogen transmission trajectory, identify key nodes and paths, and provide intuitive data support for epidemic tracing and prevention strategy formulation.

[0100] In one feasible implementation, the system further includes a propagation risk assessment unit, which includes a statistics module, an identification module, and a risk calculation module. The aforementioned statistics module is used to count the number of cross-cluster connections and the average edge weight in the evolutionary relationship graph of the pathogens. The aforementioned identification module is used to identify high-frequency regions and high-frequency propagation channel nodes with high-frequency cross-cluster connections in the evolutionary relationship map of the aforementioned pathogens; The aforementioned risk calculation module is used to generate a propagation risk score based on the number of edges, the average edge weight, the high-frequency region, and the high-frequency propagation channel nodes.

[0101] For example, the transmission risk assessment module in the system is used to identify and quantitatively assess potential cross-cluster transmission phenomena in the pathogen evolutionary relationship map, and generate a transmission risk score that can be used for early warning.

[0102] First, the system extracts all cross-cluster connections in the graph using a statistical module. Cross-cluster connections refer to edges connecting sample nodes with different cluster labels, reflecting potential propagation jumps or multi-source fusion propagation paths. These edges do not belong to the regular evolutionary paths within the same cluster, and therefore have high risk indication value in the propagation chain structure.

[0103] The statistics module performs quantitative analysis on these cross-cluster edges in the following two key dimensions: The number of cross-cluster connection edges N cross This indicates the number of edges connecting different clusters in the entire graph structure. The average edge weight across clusters is [value missing]. This represents the average cost of cross-cluster connections in terms of the combined distance of genetic variation and geographical path.

[0104] After completing the basic statistics, the identification module further extracts potential high-risk areas and key propagation channel nodes from the structural map. Specifically, this includes: High-frequency propagation area identification: The system uses a sliding window algorithm or graph clustering analysis to find local areas in the graph where multiple cross-cluster edges intersect densely. These areas often represent hotspots where propagation paths intersect, super-spreading events occur, or multiple cases converge, and have significant monitoring value.

[0105] Identification of hotspot nodes in the transmission channel: The system counts the number of cross-cluster edges connected to each sample node. If the number exceeds a set threshold (e.g., 3 edges), the node is marked as a "hotspot in the transmission channel". Such nodes usually represent important circulation hubs, key passenger distribution centers, or super-spreader samples that have been associated with multiple transmission paths.

[0106] Based on the above statistical and identification results, the system constructs a propagation risk scoring model R to comprehensively quantify the overall propagation complexity and potential risk level in the map.

[0107] ; in, This represents the number of cross-cluster connection edges; The average variation distance across cluster edges (including genetic / national / portal differences); The number of high-frequency propagation regions (locally highly clustered cross-cluster regions); The number of hotspot nodes in the propagation channel (e.g., samples of super-spreaders); The source diversity index is calculated using Shannon entropy and represents the diversity of samples from different countries / ports of entry. The larger the value, the more complex the transmission path. , , These are the risk model coefficients, used to adjust the weights of various risk factors, and can be obtained through training on historical epidemics.

[0108] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A rapid sequencing and source tracing analysis system for imported infectious diseases, characterized in that, include: The sample preprocessing unit is used to identify and encode the preprocessed samples to form preprocessed sample information. The sample processing includes nucleic acid extraction, purification and quality detection. The high-throughput sequencing unit is used to construct sequencing libraries from preprocessed sample information and obtain raw sequencing data. The data processing and quality control unit is used to perform quality control, adapter removal, and host sequence removal on the raw sequencing data to obtain effective pathogen sequencing data. The rapid alignment and annotation unit is used to align valid pathogen sequencing data with a preset reference database to obtain alignment and annotation results; The source tracing analysis unit is used to construct an evolutionary relationship map between pathogens based on the alignment and annotation results. Specific steps include: Feature vectors are extracted from the comparison and annotation results. The feature vectors include a list of SNP sites, InDel or amino acid substitution positions, and virulence or resistance site tags. Enhanced variation distance information between any two samples in the comparison and annotation results is constructed. The enhanced variation distance information includes genotype differences, differences in countries of entry, and port of entry information. All features are uniformly encoded into structured feature vectors. After feature extraction is complete, the system performs analysis on any two samples. and Calculate its enhanced variation distance : ； in, Hamming distance is based on SNP / InDel / functional annotation vectors. For the country difference item, 1 is assigned if the country of entry is different, and 0 is assigned if they are the same. This is the port difference item; 1 indicates a different port, and 0 indicates the same port. , , As a weighting factor, This is a nonlinear enhancement index of genetic distance. The distance normalization index, For joint structure penalty terms; ;in, This represents the propagation similarity direction angle between sample feature vectors; , and These are the feature vectors of two samples; The structural penalty coefficient controls the increment when the propagation directions are inconsistent. punish; The enhanced mutation distance information is input into the DBSCAN algorithm for unsupervised clustering to obtain the clustering results; a minimum spanning tree propagation graph is constructed based on the enhanced mutation distance information, including: A graph structure G(V, E) is constructed based on enhanced distance information, where node V represents a sample, edge E represents the potential propagation relationship between samples, and the edge weight is the enhanced mutation distance information mentioned above. Apply the minimum spanning tree algorithm to the graph structure to construct the minimum spanning tree propagation graph between samples; The clustering results are mapped to a minimum spanning tree propagation graph to construct an evolutionary relationship map among pathogens, including: Write clustering labels into the attributes of each node in the minimum spanning tree propagation graph, so that each node in the graph contains information about its clustering category. Traverse the edge structure of the minimum spanning tree propagation graph and determine whether the cluster labels of the nodes at both ends of the edge are consistent; If they are inconsistent, mark the edge as a cross-cluster propagation path; The edge weights of cross-cluster propagation paths are weighted and adjusted to enhance the ability to represent the risk of propagation jumps in the graph; The output is an enhanced propagation graph structure containing clustering labels and cross-cluster connection markers to construct an evolutionary relationship map among pathogens.

2. The rapid sequencing and source tracing analysis system for imported infectious diseases according to claim 1, characterized in that, The specific steps for identifying and encoding the pre-processed samples to form pre-processed sample information include: Based on the personnel information corresponding to the sample to be tested after sample processing, generate identity information coding segment and time coding segment; Environmental data is embedded in the coding segment based on port passenger flow level information and historical disease level; An extended coding segment is generated based on the vaccination information and travel itinerary information of the personnel corresponding to the test sample; The identity information coding segment, time coding segment, environmental data embedding coding segment, and extended coding segment are bound to the sample to be tested after sample processing to form preprocessed sample information.

3. The rapid sequencing and source tracing analysis system for imported infectious diseases according to claim 1, characterized in that, The specific steps for constructing sequencing libraries from preprocessed sample information and obtaining raw sequencing data include: The target library type is determined based on the nucleic acid type, concentration range, and target pathogen type information of the preprocessed sample information; After the target library type is determined, the target library corresponding to the preprocessed sample is built according to the library construction rules corresponding to the target library type. Concentration and fragment distribution information of the target library were obtained through quantitative fluorescence and fragment analysis. Calculate the library quality score based on concentration and fragment distribution information; Load the target library that meets the preset quality score into the sequencing platform to obtain raw sequencing data.

4. The rapid sequencing and source tracing analysis system for imported infectious diseases according to claim 1, characterized in that, The specific steps for quality control, adapter removal, and host sequence removal of raw sequencing data to obtain effective pathogen sequencing data include: Quality assessment was performed on the raw sequencing data, analyzing the base quality distribution, N base ratio, length distribution, and GC content deviation of the sequencing reads. Low-quality sequencing reads were filtered and tails were pruned based on a set threshold to obtain quality-filtered sequencing data. Load the adapter sequence database of the corresponding library construction platform, use the adapter identification algorithm to match and locate the adapter sequences in the quality-filtered sequencing data, and perform adapter sequence trimming to obtain trimmed sequencing data; The host reference genome database is loaded, and the cropped sequencing data is compared with the host genome to identify and remove sequencing reads from the host in order to obtain effective pathogen sequencing data.

5. The rapid sequencing and source tracing analysis system for imported infectious diseases according to claim 1, characterized in that, The specific steps for comparing valid pathogen sequencing data with a pre-defined reference database to obtain comparison and annotation results include: A pre-defined reference database containing multiple known pathogen reference genome sequences is constructed, and the pre-defined reference database is indexed and preprocessed to form a target index structure, wherein the target index structure is a two-layer index structure composed of a Bloom filter and a hash table. k-mer fragments are extracted from each sequencing read in the effective pathogen sequencing data, and parallel hash matching operations are performed on the GPU. The k-mer fragments are compared with the target index structure of the preset reference database, and a hit score for each pathogen is generated based on the number of hits to obtain the initial screening results. Based on the initial screening results and the dynamic limitation of the high-precision alignment range, the reference sequences of the Top-N high-scoring candidate pathogens are selected as targets, and the alignment algorithm is used to perform mapping operations on the sequencing reads to obtain the alignment results. The annotation module is called to quickly map virulence factors, resistance genes and key functional regions of the sequencing reads of the successfully aligned results for annotation using the annotation hash table mechanism, so as to obtain the alignment and annotation results.

6. The rapid sequencing and source tracing analysis system for imported infectious diseases according to claim 1, characterized in that, The system also includes a disease transmission path analysis unit, which further comprises a construction module, a traversal module, a calculation module, and a display module. The building module is used to call the edge weight information, timestamp information and transmission path connection information in the evolutionary relationship graph between pathogens to construct the transmission time sequence graph between samples; The traversal module is used to perform graph traversal operations based on the propagation time sequence graph to determine the propagation subgraph starting from a specified source sample. The graph traversal operations include breadth-first search or depth-first search. The calculation module is used to perform propagation characteristic analysis on the nodes of the propagation subgraph and calculate key propagation node indicators, including node degree, betweenness centrality, and propagation path length. The display module is used to show the propagation timeline and node indicators in the visualization graph.

7. The rapid sequencing and source tracing analysis system for imported infectious diseases according to claim 1, characterized in that, It also includes a transmission risk assessment unit, which comprises a statistical module, an identification module, and a risk calculation module. The statistics module is used to count the number of cross-cluster connections and the average edge weight in the evolutionary relationship graph of pathogens; The identification module is used to identify high-frequency regions and high-frequency propagation channel nodes with high-frequency cross-cluster connections in the evolutionary relationship map between pathogens; The risk calculation module is used to generate a propagation risk score based on the number of edges, average edge weight, high-frequency regions, and high-frequency propagation channel nodes.

Images