Pathogen detection system and method based on metagenomic high-throughput sequencing

Abstract

The present invention discloses a pathogen detection system and method based on metagenomic high-throughput sequencing; the detection method comprises at least one of the following: cleaning of a public database, calculation and filtering of S_confidence and L_score of kmer classification information, and contamination identification of a basic background model and a DIC background model; the method can control the detection of false-positive species, and provide convenience for downstream report interpretation and assisting clinicians in determining the true pathogen.

Description

Technical Field

[0001] The present invention belongs to the technical field of pathogen detection, and in particular relates to a pathogen detection system and method based on metagenomic high-throughput sequencing. Background Art

[0002] Infectious diseases are a general term for diseases caused by pathogenic microorganisms (bacteria, viruses, fungi, parasites, etc.), and are major diseases that seriously threaten human health. For a long time, the diagnosis and efficacy monitoring of infectious diseases have relied on methods such as morphology, immunology, molecular biology, and pathogen isolation and culture. These methods have their own advantages and disadvantages and play an important role in the auxiliary diagnosis of infectious diseases.

[0003] In recent years, the emerging pathogen metagenomic high-throughput sequencing (mNGS) technology refers to a detection method that uses high-throughput sequencing technology to sequence all nucleic acids in a specific clinical sample and uses bioinformatics analysis to determine whether the sample contains pathogens. Compared with traditional pathogen detection technology based on isolation and culture, this technology can theoretically detect various types of microorganisms (such as viruses, bacteria, fungi, parasites, etc.) without bias, including pathogens that are difficult to culture and emerging pathogens. mNGS technology is an open analysis and diagnostic system. There is no clear regulation on the number of pathogens detected by mNGS technology. According to incomplete statistics, institutions that provide related testing services have included nearly 10,000 pathogens, including bacteria, viruses, fungi and parasites, which provides effective technical means for the diagnosis of difficult and critical diseases and rare pathogen infections. It has clinical significance in specific clinical application scenarios.

[0004] Pathogen species identification based on mNGS can be divided into several categories:

[0005] 1. Alignment-based methods. Align the reads of mNGS to different species assemblies, and determine the presence of specific species based on the sequence alignment position and reliability pair. mNGS has no assumptions about the scope of species identification, and the species assembly sequences in public databases increase massively every year, making it difficult for general application equipment to meet the challenges brought by its spatial complexity and time complexity. If you only focus on the sequences of common known pathogenic species, although you can alleviate the bottlenecks of hardware and time, you will limit the detection range of species and reduce the sensitivity of clinical pathogen detection, especially for rare or new pathogens;

[0006] 2. Species identification based on marker genes. This method first calculates species-level specific single-copy genes (marker genes) and aligns the mNGS data to these marker genes. The advantage of this method is that it requires little computation, but the disadvantage is that it requires a high amount of sequencing data (considering the need to fully cover marker genes), which limits the sensitivity of the method. At the same time, the nucleic acid similarity of some species is extremely high, which poses a challenge to the calculation of marker genes. The general operation is that multiple closely related species share marker genes, thereby reducing the resolution of closely related species;

[0007] 3. Kmer (continuous subsequence of nucleic acid sequence) based method. The species classification problem is converted into an mNGS sequence classification problem, and then the sequence classification problem is converted into kmer matching. While reducing the amount of calculation, it improves the overall analysis timeliness and has ultra-high sensitivity. These characteristics make the kmer-based species detection method popular in the field of pathogen detection. The ultra-high sensitivity of this method will have certain false positives due to the combined effects of poor database quality, biological variation, sequencing noise, and potential contamination in different experimental steps, which brings confusion to report interpretation and auxiliary doctor diagnosis, and has been difficult to solve.

[0008] Therefore, it is necessary to provide a pathogen detection system and method based on metagenomic high-throughput sequencing to suppress the false positives that occur in existing metagenomic high-throughput sequencing pathogen detection and improve the accuracy of clinical diagnosis. Summary of the invention

[0009] The purpose of the present invention is to propose a pathogen detection system and method based on metagenomic high-throughput sequencing, which is used to effectively reduce or inhibit false positives in pathogen detection and improve the accuracy of clinical diagnosis.

[0010] In view of this, the scheme of the present invention is as follows:

[0011] The first aspect of the present invention provides a pathogen detection system based on metagenomic high-throughput sequencing, comprising:

[0012] Database module, used to build pathogenic species database;

[0013] The detection module is used to compare the metagenomic sequencing results with the database to obtain the pathogen species identification results;

[0014] A filtering module is used to filter the results of pathogen species identification; the filtering method is: constructing a species kmer database based on the pathogen species database, judging the species reliability based on the distribution of the metagenomic sequence kmer on the corresponding species classification tree in the species kmer database, and filtering the identified pathogen species; the reliability judgment process is: respectively calculating the proportion S_confidence of the number of kmers classified into the current species in the total number of kmers, and the proportion L_score of the number of kmers of the species sequence on the current species node and its direct nodes in the total number of kmers, taking sequences with S_confidence and L_score both close to 1 as reliable sequences, and taking species with more than two reliable sequences as filtered pathogen species.

[0015] Furthermore, the reliability judgment process is as follows: taking species with S_confidence quartile values ​​of all sequences in the identified species greater than 80% as the filtered pathogenic species;

[0016] And / or, the species with the maximum L_score of the sequence among the identified species that satisfies the following formula is taken as the pathogenic species after filtering: , count k ( all ) is the number of kmers under the maximum sequence length assigned to the current species.

[0017] Furthermore, the process of constructing a pathogenic species database by the pathogenic species database construction module includes: collecting and screening high-quality pathogenic species assembly sequences; taking high-quality assembly sequences in the species as the core, and retaining assembly sequences with high intraspecies similarity to the core assembly sequences.

[0018] Preferably, the intraspecies similarity is compared based on average nucleic acid identity and alignment coverage.

[0019] Preferably, the screening process of the pathogenic species assembly sequence is performed according to the assembly completion and contamination degree.

[0020] Preferably, the construction process of the pathogenic species database further includes filtering of abnormal assembly indicators, where the assembly indicators include total assembly length, number of contigs, and GC content;

[0021] and / or, merging highly related species in the assembled sequences;

[0022] and / or, eliminating moving elements from the assembly sequence;

[0023] and / or, masking low complexity sequences.

[0024] Furthermore, the filtering module also includes a step of removing internal contamination and / or external contamination from the pathogen species identification result.

[0025] Preferably, the internal contamination removal process is to calculate the significance of the rPM of the identified species relative to the rPM of the negative control sample, and species with less significance are regarded as contaminant species and removed;

[0026] Preferably, the external contamination removal process is to standardize the number of reads of the identified species according to the number of artificially inserted reference sequences, calculate the significance of the normalized number of species reads relative to the negative control sample, and regard those with low significance as contaminant species and remove them.

[0027] The second aspect of the present invention provides a pathogen detection method based on metagenomic high-throughput sequencing, including the steps of constructing a pathogen species database, and comparing the metagenomic sequencing results with the database to obtain pathogen species identification results;

[0028] The detection method also includes constructing a species kmer database based on a pathogenic species database, judging species reliability based on the distribution of metagenomic sequence kmer on the corresponding species classification tree in the species kmer database, and filtering the identified pathogenic species; the reliability judgment process is: respectively calculating the proportion S_confidence of the number of kmers of the species sequence classified to the current species to the total number of kmers, and the proportion L_score of the number of kmers of the species sequence on the current species node and its direct nodes to the total number of kmers, taking sequences with S_confidence and L_score both close to 1 as reliable sequences, and taking species with more than two reliable sequences as filtered pathogenic species.

[0029] As mentioned above, the percentage close to 1 refers to the percentage that is closer to 1 among the percentages. Thresholds such as more than 50%, more than 60%, more than 70%, and more than 80% can be set to select reliable sequences and then filter out reliable species.

[0030] According to a third aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, and when a processor executes the computer program, the pathogen detection method as described in the second aspect is implemented.

[0031] Compared with the prior art, the beneficial effects of the present invention include but are not limited to:

[0032] The pathogen detection system of the present invention determines the species reliability based on the distribution of the metagenomic sequence kmer on the corresponding species classification tree in the species kmer database, filters the identified pathogenic species, and can control the detection of false positive species, thus facilitating the interpretation of downstream reports and assisting clinicians in determining the true pathogen.

[0033] The pathogen detection system of the present invention constructs a pathogen species database by screening intraspecies similarity, removes assembly sequences with low credibility, and further controls the detection of false positive species by merging highly related species and removing mobile elements.

[0034] The pathogen detection system of the present invention can further reduce false positive detection by setting a basic background model and a reference sequence background model to identify potential contaminant species using the statistical indicator z-score and remove them. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0036] Figure 1 This is a flow chart of the pathogen detection method based on metagenomic high-throughput sequencing of the present invention;

[0037] Figure 2 This is a flow chart of the species database processing described in the present invention;

[0038] Figure 3 is a schematic diagram of the intraspecific similarity screening process of the present invention;

[0039] Figure 4 A schematic diagram of the mobile component removal process of the present invention;

[0040] Figure 5 This is a schematic diagram of the species filtering according to the distribution of sequence kmer on the classification tree according to the present invention. DETAILED DESCRIPTION

[0041] The following provides definitions of some terms used in this specification. Unless otherwise specified, all terms used herein are generally understood by those skilled in the art to which this solution belongs.

[0042] Terminology explanation:

[0043] mNGS: Metagenomic Next-Generation Sequencing, uses high-throughput sequencing methods to sequence all biological sequences (DNA or / and RNA) in a sample. It is used to analyze the species, abundance, and function of organisms in a sample. It should be stated that the sample source for the metagenomic sequencing data can be clinical samples, or samples from animals and living environments.

[0044] kmer: refers to k consecutive base pairs in a DNA or RNA sequence. kmer is a characteristic representation of the sequence.

[0045] z-score: In statistics, the Z-score is also called the standard score, which represents the distance of a data point from the mean of the data set in units of standard deviation. In other words, it measures how far a data point deviates from the mean of the data set.

[0046] ANI: Average Nucleotide Identity, is the average consistency of nucleotides between two genome sequences. It is obtained by comparing the entire genome sequences of two genomes and calculating the proportion of identical base pairs between them.

[0047] AF: Alignment fraction. In bioinformatics, Alignment fraction is translated into Chinese as "alignment score" or "alignment ratio". It refers to the ratio of the number of bases that can be successfully aligned to the total number of bases after two sequences are aligned.

[0048] NTC: Negative control samples usually refer to No Template Control negative control samples. In experiments, negative control samples are an important control group, which can help exclude the influence of other factors on experimental results, thus ensuring the reliability of experimental results.

[0049] In one embodiment, a method for pathogen detection based on metagenomic high-throughput sequencing is provided, including the steps of comparing the metagenomic sequencing results with the pathogen species database to obtain the pathogen species identification results, and filtering the pathogen species identification results. In order to reduce or suppress false positives in the species classification process, improvements are made in terms of the assembly of the pathogen database and the filtering of the identification results. The specific process is as follows: Figure 1 The specific improvements include the following aspects:

[0050] 1. Assembly and cleaning of species in public databases

[0051] The inventors found that the reliability of the kmer method for pathogen identification depends first on the quality of the database. Based on this, the inventors collected assembly sequences of viruses, bacteria, archaea, parasites, etc. in public databases (including but not limited to Refseq, GeneBank). Bacteria and archaea account for the vast majority of the database. Therefore, the database quality of bacteria and archaea is processed, and the operation process is as follows: Figure 2 As shown, the details are as follows:

[0052] 1. Assembly quality screening

[0053] Assemblies that meet the following conditions are retained as high-quality candidate assemblies for further processing.

[0054] (1) Assembly completeness < 60%;

[0055] (2) Assembly contamination level > 5%;

[0056] (3) (assembly completion degree*5) × assembly contamination degree <50;

[0057] This assembly evaluation metric is calculated by checkM2.

[0058] 2. Intraspecific similarity screening

[0059] There is some erroneous species information in public databases, which leads to deviations in species classification and even affects the sensitivity of species.

[0060] To this end, we took the highest quality or recognized classic strains and representative assemblies in the species as the core, and calculated the average nucleic acid similarity and alignment coverage of other assemblies of the same species to the core assembly, such as Figure 3 shown. Figure 3 In the figure, the central red dot represents the core assembly of a specific species, and the other dots (black and orange) represent other assemblies of a specific species. The average nucleic acid identity (ANI) is used to measure the similarity of two assemblies at the nucleic acid level (between 0 and 1, the closer to 1, the closer they are). Due to its asymmetry, the ANI of the two assemblies is calculated twice. We define the ANI of other assemblies to the core assembly as a negative value, and the ANI of the core assembly to other assemblies as a positive value. The alignment score (AF) indicates the alignment coverage of the two assemblies. Similarly, for the sake of uniformity, we define it as:

[0061] in, sgn () is a common positive and negative sign function. v_ANI (Equation 1) falling in the interval [-5, 5] is considered to be similar to the core assembly of the species from the perspective of nucleic acid similarity. v_AF(Equation 2) falling in the interval [-10, 10] is considered to be close to and complete with the core assembly in terms of coverage. Figure 3 As shown, the assemblies indicated by orange dots are trusted assemblies within the species and are retained; the assemblies indicated by black dots are suspicious assemblies within the species and are kicked out.

[0062] 3. Assembly index abnormal filtering

[0063] Intraspecific assemblies should be similar in total assembly length and GC content.

[0064] For assemblies with significantly abnormal assembly length, GC content, N50, L50, and contig number compared to other assemblies, they are kicked out. This step can further reduce the risk of contamination or inaccuracy of assembly biological samples.

[0065] 4. Merger of highly related species

[0066] Since there are differences between traditional taxonomy and genomic species taxonomy, and mNGS essentially infers taxonomy from the nucleic acid level. Therefore, we will v_ANI Species with a similarity ratio <1 are merged to increase the sensitivity of species detection (if the nucleic acids of two species are highly similar, the single sequence of mNGS will be assigned to the common minimum ancestor node of the two species, thereby reducing the detection sensitivity at the species level. The classic example is Escherichia coli and Shigella).

[0067] 5. Removal of moving components

[0068] During the species assembly process, some submitters will retain sequences such as plasmids, bacterial viruses (phages), and other sequences that can be horizontally transmitted between different closely related species or sequences of organisms in the species assembly, which will cause confusion in the downstream sequence species classification. We remove related horizontally mobile elements based on relevant plasmid databases, phage databases, and sequence name information to ensure the cleanliness of the assembly;

[0069] It mainly includes two steps: Figure 4 As shown: The first step is to remove the complete mobile element sequences assembled by species such as bacteria based on the sequence ID, and the second step is to remove the sequence fragments of mobile elements formed by incorrect assembly or sequence integration in the species assembly based on sequence similarity.

[0070] 6. Low complexity sequence masking

[0071] In the process of biological sequence comparison, low-complexity sequences contain low information and may cause erroneous results in sequence comparison. We use software such as dust (not limited to dust) to shield low-complexity sequences and avoid false positive results in the process of biological sequence comparison.

[0072] 2. Filtering of sequence kmer classification information

[0073] After obtaining the clean database, use kmer-based classification software (such as kraken2 but not limited to kraken2) to build a species kmer database. And perform classification analysis on the sequences of mNGS samples. Due to biological mutations, sequencing noise and limitations of different kmer classification methods, there is a certain probability of error in the classification information of a single kmer. In order to reduce the false positives caused by this probability error, the distribution of the sequence kmer on the species classification tree was further analyzed, and two scores were used to judge the classification reliability of an independent sequence, namely S_confidence (species-levelconfidence) and L_score (Lineage score). The calculation method is as follows Figure 5 As shown, the specific process is as follows:

[0074] S_confidence is defined as: the ratio of the number of kmers classified into the current species (and its subnodes) in all kmers of a specific sequence (read or assembled contig) to the total number of kmers in the sequence, with a value range of 0-1. The closer it is to 1, the more unique kmers in the sequence kmers belong to the current species, indicating a greater possibility that it comes from the current species (as shown in Formula 3).

[0075]

[0076] in, count k Represents the number of kmers under the maximum sequence length assigned to the current species, the same below.

[0077] L_score definition: The proportion of all kmers of a specific sequence assigned to the current species node and its direct nodes (including direct ancestor nodes and direct descendant nodes) to the total number of kmers. Figure 5 The red circle represents "species 1", the blue shadow represents its direct nodes, and the others are indirect nodes. If the kmers of a specific sequence are concentrated on the direct nodes, the surface classification process and data construction are reliable. The value range is between 0 and 1, and the closer to 1, the more reliable it is. If the kmer is assigned to other indirect branches of the classification tree (such as Figure 4 If there are too many non-negative nodes in the dataset, it means that the classification of the current sequence is unreliable (as shown in Formula 4).

[0078]

[0079] According to the above definition, sequences with both S_confidence and L_score close to 1 were taken as reliable sequences, and species with more than 2 reliable sequences were taken as filtered pathogenic species.

[0080] Specifically, a reliable and efficient filtering method is as follows: for the sequences classified by the kmer method, we group them according to species, set a threshold of >80% for the upper quartile of S_confidence of all sequences of identified species, and the maximum value of L_score of the identified species satisfies Formula 5 and is considered a reliable species.

[0081]

[0082] 3. Establish a background model to eliminate internal and external pollution species.

[0083] The species obtained by high-throughput sequencing may come from the clinical samples themselves, environmental pollution, reagents and consumables, and background bacteria in library construction and sequencing. The purpose of pathogen detection is to detect microorganisms from clinical samples, while microorganisms from other sources are considered false positives. In order to identify microorganisms from the environment, consumables, instruments, etc., we set up multiple negative control samples (NTC) and add a specified amount of reference sequences (DIC) to all samples. Based on these samples and data, we establish a background model at the bioinformatics level, and use the statistical indicator z-score to identify potential contaminant species to reduce false positive detection.

[0084] There are two background models: basic background model and reference sequence background model. The specific description is as follows:

[0085] 1. Basic background model:

[0086] The basic background model does not consider the reference fragments and is used for general process experiments. First, the rPM (reads per million, sequence number per million) of the specific species (sp) of the biological sample (biosample) is calculated, as shown in formula (6); the purpose of this calculation is to offset the reads difference caused by the difference in sequencing data volume. The rPM of the three negative control samples (NTC) of the experiment on the same day or the negative control samples (NTC) of the past three days is calculated. Assuming that the DNA amount of the experimental contaminant species is normally distributed in multiple NTC samples, the standard deviation (SD) is used to calculate whether the rPM (s) of the current biological sample is significant under the NTC model. The Z-score (such as formula 7) threshold is set according to the characteristics of the species to define the significance (such as viruses and bacteria set z-score ≥ 1.96).

[0087]

[0088]

[0089] in, mean ( ) represents the mean value function.

[0090] 2. Reference sequence (DIC) background model:

[0091] The reference sequence is added for quality normalization to further accurately identify exogenous contaminant species.

[0092] First, the number of reads of a specific species sp in the sample was standardized according to the number of DICs to obtain the standardized number of sp reads, which is recorded as reads_norm ( sp ), as shown in formula (8). The z-score of the background sample (NC) model is calculated with the standardized reads, as shown in formula (9), with the same threshold as above.

[0093]

[0094]

[0095] When there is a significant difference between the species reads of a biological sample and any of the two background models, it is considered a reliable species, otherwise it is considered a suspected contaminant species.

[0096] Example

[0097] (1) mNGS data preprocessing (high-throughput data quality control, human data removal, low-complexity sequence removal) obtains quality control, human reads, low-complexity reads and microbial reads. The statistical table is as follows:

[0098] Sample Sample1 NTC1 NTC2 NTC3 Sample_type Bio NTC NTC NTC Raw_reads 20000000 1578463 1941510 1499540 Raw_bases 1000000000 78923150 97075500 74977000 Clean_reads 19999782 1565835 1925977 1487543 Clean_bases 979989318 77508832 95335861 73633378 Q20(%) 98.53 98.55 98.28 98.56 Q30(%) 91.48 91.28 90.66 91.14 Reads_median_length 50 50 50 50 GC(%) 42.7 43.08 42.88 42.9 Host_reads 19803696 1061636 1161364 1014504 Low_complexity_reads 2459 192 236 182 Micro_reads 193627 504007 764377 472857 Micro_rate(%) 0.01 0.322 0.397 0.318

[0099] (2) Kmer-based sequence species classification (taking kraken2 as an example)

[0100] The number of species-unique reads was obtained using kmer-based sequence classification software (Kraken2).

[0101] Sample name UniqReads Sample1 Mycobacterium tuberculosis 24 Sample1 Leptotrichia wadei 3 Sample1 Escherichia coli 11 Sample1 Oribacterium sinus 1 Sample1 Klebsiella pneumoniae 21

[0102] (3) S_confidence and L_score calculation and filtering

[0103] Unreliable reads in kmer distribution were filtered out based on the upper quartile threshold of S_confidence and the maximum threshold of L_score.

[0104] Sample name S_confidence(UQ) Max(L_score) Pass Sample1 Mycobacterium tuberculosis 0.9375 1 True Sample1 Leptotrichia wadei 0.6875 0.8125 False Sample1 Escherichia coli 0.9375 1 True Sample1 Oribacterium sinus 0.5625 0.5625 False Sample1 Klebsiella pneumoniae 0.9375 1 True

[0105] (4) Background model filtering (basic background model, DIC background model)

[0106] Sample name Basic background model Z DIC Background Model Z Pass Sample1 Mycobacterium tuberculosis 100 100 True Sample1 Escherichia coli 1.03 1.25 False Sample1 Klebsiella pneumoniae 7.32 6.91 True

[0107] *Note: When no relevant species were detected in the NC sample, the z-score value was defined as 100.

[0108] (5) Final species information output

[0109] Reliable detected species information obtained after false positive filtering in steps (3) and (4).

[0110] Sample name UniqReads Sample1 Mycobacterium tuberculosis 24 Sample1 Klebsiella pneumoniae 21

[0111] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A pathogen detection system based on metagenomic high-throughput sequencing, characterized in that: include: Database module, used to build pathogenic species database; The detection module is used to compare the metagenomic sequencing results with the database to obtain the pathogen species identification results; A filtering module is used to filter the results of pathogen species identification; the filtering method is: constructing a species kmer database based on the pathogen species database, judging the species reliability based on the distribution of the metagenomic sequence kmer on the corresponding species classification tree in the species kmer database, and filtering the identified pathogen species; the reliability judgment process is: respectively calculating the proportion S_confidence of the number of kmers classified into the current species in the total number of kmers, and the proportion L_score of the number of kmers of the species sequence on the current species node and its direct nodes in the total number of kmers, taking the sequence with S_confidence and L_score both close to 1 as the reliable sequence, and taking the species with more than 2 reliable sequences as the filtered pathogen species; The reliability judgment process is as follows: The species with S_confidence quartile values ​​greater than 80% among all sequences in the identified species are selected as the pathogenic species after filtering; and the species with the maximum L_score of the sequence in the identified species that satisfies the following formula is selected as the pathogenic species after filtering: , count k ( all ) is the number of kmers under the maximum sequence length assigned to the current species.

2. The pathogen detection system according to claim 1, characterized in that: The pathogenic species database construction process includes: collecting and screening high-quality pathogenic species assembly sequences; taking high-quality assembly sequences in the species as the core, and retaining assembly sequences with high intraspecies similarity to the core assembly sequences.

3. The pathogen detection system according to claim 2, characterized in that: The intraspecific similarity is compared based on average nucleic acid identity and alignment scores.

4. The pathogen detection system according to claim 2, characterized in that: The screening process of the pathogenic species assembly sequence is performed according to the assembly completion and contamination degree.

5. The pathogen detection system according to claim 2, characterized in that: The construction process of the pathogenic species database also includes filtering of abnormal assembly indicators, where the assembly indicators include total assembly length, number of contigs, and GC content; and / or, merging highly related species in the assembled sequences; and / or, eliminating moving elements from the assembly sequence; and / or, masking low complexity sequences.

6. The pathogen detection system according to claim 1, characterized in that: The filtering module also includes a step of removing internal contamination and / or external contamination from the pathogen species identification result.

7. The pathogen detection system according to claim 6, characterized in that: The internal contamination removal process is to calculate the significance of the rPM of the identified species relative to the rPM of the negative control sample, and the species with less significance are regarded as contaminant species and removed; the rPM is the number of sequences per million; And / or, the external contamination removal process is to standardize the number of reads of the identified species according to the number of artificially inserted reference sequences, calculate the significance of the normalized number of species reads relative to the negative control sample, and regard those with low significance as contaminant species and remove them.

8. A pathogen detection method based on metagenomic high-throughput sequencing, using the pathogen detection system of claim 1; characterized in that: This includes the construction of a pathogenic species database and the steps of comparing the metagenomic sequencing results with the database to obtain pathogenic species identification results; The detection method also includes constructing a species kmer database based on a pathogenic species database, judging species reliability based on the distribution of metagenomic sequence kmer on the corresponding species classification tree in the species kmer database, and filtering the identified pathogenic species; the reliability judgment process is: respectively calculating the proportion S_confidence of the number of kmers of the species sequence classified to the current species to the total number of kmers, and the proportion L_score of the number of kmers of the species sequence on the current species node and its direct nodes to the total number of kmers, taking sequences with S_confidence and L_score both close to 1 as reliable sequences, and taking species with more than two reliable sequences as filtered pathogenic species.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the processor executes the computer program to implement the pathogen detection method as claimed in claim 8.

Images