A multi-omics detection system for lung cancer

Abstract

A multi-omics detection system for lung cancer, which relates to the field of lung cancer research, includes a clinical data collection module, a clinical sample and information acquisition module, and a test data standardization and analysis module; the clinical sample and information acquisition module includes a rule setting unit, a peripheral blood collection and processing unit, an epigenome test unit, and a proteome and metabolome test unit; the test data standardization and analysis module includes a lung nodule AI automated analysis unit, a multi-omics data analysis, and a lung nodule benign and malignant differentiation model construction unit. The present invention is based on medical imaging, and combines peripheral blood epigenomics, proteomics, metabolomics and other related technical means to detect and evaluate lung nodule patients, study the biomarkers of benign and malignant differences in lung nodules, the protein expression and metabolic differences caused by the epigenetic modification mechanism and its effects, so as to establish a high-sensitivity and high-specificity clinical lung cancer detection scheme for early warning of lung cancer and differentiation of benign and malignant lung nodules.

Description

Technical Field

[0001] The present invention relates to the field of lung cancer research, and in particular to a multi-omics detection system for lung cancer. Background Art

[0002] The incidence and mortality of lung cancer in China rank first in the world. Early detection and diagnosis of lung cancer are the key to lung cancer prevention and treatment and improving survival rate. Accurate assessment of the invasiveness of lung adenocarcinoma directly determines the choice of treatment plan and prognosis assessment. The accuracy and robustness of evaluating the invasiveness of lung adenocarcinoma by traditional imaging, computer-aided diagnosis and radiomics are limited to a certain extent. Deep learning methods (Deep Learning) provide a new opportunity for realizing automatic deep analysis of medical images, high-precision intelligent auxiliary diagnosis and integrated multi-omics liquid biopsy, and for clinical differential diagnosis of benign and malignant lung nodules, and for developing highly sensitive and specific lung cancer markers. At present, the improvement of imaging diagnostic efficiency faces two bottleneck problems: one is that the scanning scheme and image resolution standards have not yet been established; the other is that the resolution of traditional CT images cannot support in-depth research on the correlation between tumors and surrounding tissues. Summary of the invention

[0003] (I) Purpose of the invention

[0004] In order to solve the technical problems existing in the background technology, the present invention proposes a multi-omics detection system for lung cancer. The present invention is based on artificial intelligence medical imaging, and relies on related technical means such as epigenomics, proteomics and metabolomics to test the peripheral blood of lung nodules, study the biomarkers of benign and malignant lung nodules, and the potential epigenetic modification mechanism and the protein expression and metabolic differences caused by its action, so as to establish a high-sensitivity and high-specificity clinical lung cancer detection scheme for early warning of lung cancer and differentiation of benign and malignant lung nodules.

[0005] (II) Technical solution

[0006] To solve the above problems, the present invention provides a multi-omics detection system for lung cancer, which relates to the field of lung cancer research, and includes a clinical data collection module, a clinical sample and information acquisition module, and an experimental data standardization and analysis module; the clinical data collection module collects the original DICOM data of patient CT; the clinical sample and information acquisition module includes a rule setting unit, a peripheral blood collection and processing unit, an epigenome test unit, and a proteome and metabolome test unit; the experimental data standardization and analysis module includes a lung nodule AI automated analysis unit, a multi-omics data analysis and a lung nodule benign and malignant differentiation model construction unit.

[0007] Preferably, the above-mentioned lung cancer multi-omics detection system has the following working steps:

[0008] S1. Patients who are scheduled for surgical pathological examination should undergo low-dose CT examination, with the scanning range from the apex of the lung to the lower edge of the posterior costophrenic angle, including the chest wall and axilla on both sides;

[0009] S2. Set the detection object rules and establish the test group according to the rules;

[0010] S3, collecting and processing peripheral blood to obtain test samples;

[0011] S4. Perform epigenomic testing on the test samples;

[0012] S5. Perform protein metabolomics testing on the test samples;

[0013] S6. Collect the above test data, conduct AI automated analysis of lung nodules, multi-omics data analysis and construction of a model for distinguishing benign and malignant lung nodules; multi-omics data analysis and identification model construction include image AI processing and analysis, epigenetic data preprocessing and analysis, protein and metabolomics analysis, and multi-omics integrated analysis.

[0014] Preferably, in S2, the rules are as follows: Collect clinical information related to the enrolled patients, including but not limited to: Collect clinical information related to the patients, including but not limited to: age, gender, smoking history, clinical evaluation results at enrollment, and original DICOM information about lung nodules through CT imaging examination.

[0015] Preferably, in S3, peripheral blood collection and processing operations are as follows:

[0016] (1) For each patient, fasting arm venous blood was collected using one tube of EDTA anticoagulant tube (purple head) and one tube of procoagulant tube (yellow head);

[0017] (2) The collected blood was allowed to stand at room temperature for 30 minutes, and all blood samples were separated within 4 hours; blood samples collected from the procoagulant tube were centrifuged at 2000 g / min in a horizontal centrifuge for 15 minutes; blood samples collected from the EDTA anticoagulant tube were centrifuged at 2000 g for 15 minutes at 4°C;

[0018] (3) Use a pipette to draw up the blood supernatant in turn and dispense it into pre-labeled nuclease-free cryopreservation tubes, 1000 μl per tube;

[0019] (4) After centrifugation, the cryopreserved tubes of plasma and serum samples should be stored in a -80°C refrigerator. They should not be frozen or thawed before extraction to avoid repeated freezing and thawing.

[0020] (5) The samples are placed in special freezing boxes and transported via cold chain to the laboratory for testing.

[0021] Preferably, in S4, the epigenome detection step is as follows:

[0022] (1) Extracting and quantifying free DNA in sample plasma;

[0023] (2) Internal quality control preparation: Use dNTP mixture to amplify λ-DNA, prepare 3 sets of dNTP mix according to the proportion and use the corresponding amplification primers for amplification;

[0024] (3) 5hmC library construction, labeling, capture and sequencing: Take 1-10 ng of free DNA and add internal quality control amplification product at 0.01 pg per 10 ng of free DNA; then prepare the library; take 25 μl of labeling solution to label the prepared library with 5hmC and incubate at 37°C for 2 hours; then take 2.5 μl of DBCO-PEG4-labeled biotin, mix well, and continue incubating at 37°C for 2 hours to complete the labeling; add 10 μg of salmon sperm DNA fragments to the labeled library, mix well, and purify using a MicroBio-Spin 30 separation column;

[0025] (4) The purified free DNA was incubated with 0.5 μl of pre-treated M270 streptavidin magnetic beads at room temperature for 30 minutes. The M270 streptavidin magnetic beads were pre-blocked with salmon sperm DNA fragments in buffer A. After the incubation, the magnetic beads were gently washed four times at room temperature, namely: washing with buffer A for 5 minutes, washing with buffer B for 5 minutes, and washing with buffer C and buffer D. The washed magnetic beads were suspended in 20 μl of nuclease-free water.

[0026] (5) The captured free DNA was amplified and enriched, and the amplification cycle was set to 14 weeks; the amplified PCR product was purified using AMPure XP beads, and the purified library was sequenced according to the paired 75 base pairs.

[0027] Preferably, in S5, the protein metabolomics test steps are as follows:

[0028] (1) Take 0.5 μl of sample serum and place it on the polished steel target plate MTP384. After air drying, add 1 μl of mass spectrometry matrix of three-dimensional nanomaterials for ionization;

[0029] (2) Use matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF) to perform mass spectrometry detection of samples;

[0030] (3) Each sample was randomly measured 25 times at 20 different grating points to obtain 500 satisfactory measurement results;

[0031] (4) Use data analysis software and machine learning algorithms to study the peak spectra of serum protein metabolomics in patients with pulmonary nodules, determine the characteristic fingerprint of pulmonary nodule protein metabolomics and the characteristic peaks with significant differences between benign and malignant nodules, and further analyze and discover the metabolic pathways in which the differential protein metabolites are mainly involved and the closely related enzymes.

[0032] Preferably, in S6, the data in S1 is used for AI automated analysis of lung nodules, and automatic analysis is performed based on pre-developed artificial intelligence and deep learning lung nodule three-dimensional segmentation technology and auxiliary diagnosis technology.

[0033] Preferably, in S6, the epigenetic data analysis preprocessing steps are as follows:

[0034] (1) After the epigenetic sequencing data is obtained, the data is quality controlled by Illumina's own tools to obtain the fastq original file;

[0035] (2) The obtained raw files were evaluated for quality by FastQC, and Picard was used to remove adapter sequences and duplicate sequences;

[0036] (3) Bowtie2 was used to align HCSC / hg19 as the reference sequence. After alignment, samtools was used to remove duplicates and only unique sequences were retained for subsequent analysis.

[0037] (4) Use the bedGraphToBigWig application in bedtools to convert the unique sequence tags into bedgraph format, and normalize the converted sequences by dividing the number of paired sequences by the total number of paired sequences; use the UCSC Genome Browser to view the converted data;

[0038] (5) Capture efficiency evaluation;

[0039] (6) Through sequence alignment, the source region of cfDNA on the genome is located to infer the degree of 5hmC modification in the DNA region and perform stacking peak mapping; the corresponding 5hmC levels of genes and chromosome fragments are calculated based on genes or chromosome fragments, and the genes or chromosome fragments with different 5hmC modification levels between patients with benign and malignant pulmonary nodules are further compared; combined with the construction of the 5hmC model and the input of clinical parameters, potential biological markers are found.

[0040] Preferably, in S6, the protein and metabolomics analysis steps are as follows:

[0041] (1) Collecting 100-5000Da mass spectrometry results, collecting the original mass spectrometry data in the range of 100-5000Da obtained by MALDI-TOF-MS instrument detection, and importing the generated MzML file using MALDIquantForeign to generate a mass spectrometry CSV result that is easy to analyze;

[0042] (2) Perform inter-group and intra-group quality checks on the obtained CSV results, and perform noise reduction, image smoothing, baseline adjustment, intensity calibration, peak detection and alignment on the qualified data to obtain tens of thousands of metabolite characteristic peaks;

[0043] (3) Create a feature matrix based on the sample grouping, use machine learning or neural network methods to analyze and build models, and find substances with characteristic differences;

[0044] (4) Compare the characteristic differential substances with the proteome and metabolome databases to explore their metabolic pathways in lung cancer samples.

[0045] Preferably, in S6, multi-omics integrated analysis targets the acquired epigenomics, protein metabolomics and LDCT image data, uses neural networks, deep machine learning methods and artificial intelligence algorithms to extract features, cluster and analyze the lung nodule patient data, and establishes a corresponding multi-omics lung nodule benign and malignant differential diagnosis model.

[0046] The above technical solution of the present invention has the following beneficial technical effects:

[0047] The system in the present invention adopts the research approach of "omics to mechanism to application", uses medical imaging omics as pre-screening data for the research subjects, and performs epigenomics on the peripheral blood samples obtained, combining proteomics and metabolomics, as well as multi-omics research ideas of machine learning, to explore the molecular mechanism of the difference between benign and malignant lung nodules, screen for differentially characteristic small molecules, and thus establish a highly sensitive and specific clinical lung cancer research program; the present invention combines imaging omics, self-developed small molecule enrichment nanomaterials, mass spectrometry technology, epigenome detection technology, and artificial intelligence technology, with a data-driven methodological approach to distinguish benign and malignant lung nodules, providing a detailed and reliable methodological theoretical basis for establishing the molecular mechanism of the occurrence and development of the disease. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a technical roadmap for the multi-omics detection system for lung cancer proposed in the present invention.

[0049] Figure 2 This is a flow chart of multi-omics integrated analysis in a multi-omics detection system for lung cancer proposed by the present invention.

[0050] Figure 3This is a flow chart of the epigenetic group test in the multi-omics detection system for lung cancer proposed by the present invention.

[0051] Figure 4 This is a flow chart of the protein and metabolomics test in a lung cancer multi-omics detection system proposed by the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0053] Example 1

[0054] A multi-omics detection system for lung cancer proposed in the present invention comprises a clinical data collection module, a clinical sample and information acquisition module and a test data standardization and analysis module; the clinical data collection module collects the original DICOM data of patient CT; the clinical sample and information acquisition module comprises a rule setting unit, a peripheral blood collection and processing unit, an epigenome test unit and a protein metabolome test unit; the test data standardization and analysis module comprises a lung nodule automated analysis unit, a multi-omics data analysis and a lung nodule benign and malignant differentiation model construction unit.

[0055] Example 2

[0056] like Figure 1 As shown, the above-mentioned lung cancer multi-omics detection system has the following working steps:

[0057] S1. Low-dose CT scans are performed on patients with pulmonary nodules who are scheduled for surgical pathological examination. The scanning range is from the apex of the lung to the lower edge of the posterior costophrenic angle, including the chest wall and axilla on both sides. The hardware conditions use 16-row CT or above. The parameters are set as tube voltage 100KV, tube current 30-50mAs, and received dose 0.51-0.8mSV (<1.5mSV).

[0058] S2. Set the test object rules and establish the test group according to the rules; the rules are as follows: collect clinical information related to the enrolled patients, including but not limited to: age, gender, smoking history, clinical diagnosis results at enrollment, tumor marker results, past medical history, medication history in the past week, blood transfusion history in the past week, and CT sign information about nodules described in the imaging examination report;

[0059] Among them, the inclusion criteria of patients are:

[0060] (1) Aged 18-75, regardless of gender;

[0061] (2) Patients who plan to undergo complete resection or puncture biopsy of pulmonary nodules;

[0062] (3) Preoperative thin-section CT examination revealed a single solid nodule with a diameter of ≤3 cm in the lung, without ground glass nodules, pneumonia, or other lung manifestations;

[0063] (4) no history of malignant tumor in the past 5 years and no neoadjuvant therapy before surgery;

[0064] (5) No distant metastasis or pleural effusion or ascites were found in the preoperative evaluation;

[0065] (6) Voluntarily sign the informed consent form;

[0066] Exclusion criteria for patients:

[0067] (1) Imaging examination cannot be performed due to pregnancy, breastfeeding, etc.;

[0068] (2) no thin-section CT examination record at the clinical center receiving treatment at the time of enrollment evaluation;

[0069] (3) received blood transfusion therapy or had a history of alcoholism or drug abuse in the month before enrollment;

[0070] (4) patients who could not obtain sufficient blood specimens (≥12 ml);

[0071] (5) Any condition that is not suitable for blood drawing, such as anemia, fever, low immunity, coagulation dysfunction, etc.;

[0072] (6) Autoimmune diseases and blood system diseases.

[0073] S3. Collect and process peripheral blood to obtain test samples. The peripheral blood collection and processing operations are as follows:

[0074] (1) For each patient, fasting arm venous blood was collected using an EDTA anticoagulant tube (purple head) and a procoagulant tube (yellow head). After blood sample collection, it should be gently inverted and mixed immediately, and care should be taken to avoid hemolysis.

[0075] (2) The collected blood was allowed to stand at room temperature for 30 minutes, and all blood samples were separated within 4 hours; blood samples collected from the procoagulant tube were centrifuged at 2000 g / min in a horizontal centrifuge for 15 minutes; blood samples collected from the EDTA anticoagulant tube were centrifuged at 2000 g for 15 minutes at 4°C;

[0076] (3) Use a pipette to draw out the blood supernatant (the supernatant in the EDTA anticoagulant tube is plasma, and the supernatant in the procoagulant tube is serum) in sequence, and dispense them into pre-labeled nuclease-free cryopreservation tubes, 1000 μl per tube;

[0077] (4) After centrifugation, the cryopreserved tubes of plasma and serum samples should be stored in a -80°C refrigerator. They should not be frozen or thawed before extraction to avoid repeated freezing and thawing.

[0078] (5) Place the obtained samples (each sample contains: 2 tubes of cryopreserved plasma and 2 tubes of cryopreserved serum) in a special cryopreservation box and use cold chain transportation to arrive at the laboratory for testing to prevent dry ice from squeezing the sample tubes.

[0079] S4. Perform a methylation test on the test sample. The steps of the methylation test are as follows:

[0080] (1) Extract and quantify free DNA in the sample plasma; extraction is performed using the Circulating Nucleic Acid Kit (Qiagen); quantification is performed using the Qubit Fluorometer (Life Technologies);

[0081] (2) Internal quality control preparation: Use dNTP mixture to amplify λ-DNA. Prepare 3 sets of dNTP mix according to the ratio in Table 1 and use the corresponding amplification primers for amplification, and use Qubit Fluorometer (Life Technologies) for quantification:

[0082]

[0083] Table 1

[0084] (3) 5hmC library construction, labeling, capture and sequencing: 1-10 ng of free DNA was taken, and the internal quality control amplification product was added at 0.01 pg per 10 ng of free DNA; then the library was prepared according to the library construction operation manual of KAPA Hyper Prep Kit (Kapa Biosystems); 25 μl of labeling solution was taken to label the prepared library with 5hmC (25 μl of labeling solution contained: 50 mM HEPES buffer (pH 8), 25 mM MgCl2, 60 μM UDP-6-N3-Glc (Active Motif) and 12.5 U βGT (Thermo)) and incubated at 37°C for 2 hours; then 2.5 μl of DBCO-PEG4-labeled biotin (Click Chemistry Tools, 20 mM DMSO solution) was taken, mixed, and incubated at 37°C for another 2 hours to complete the labeling; 10 μg of salmon sperm DNA fragments (Life Technologies) were added to the labeled library. Technologies), mixed and purified using a Micro Bio-Spin 30 separation column (Bio-Rad);

[0085] (4) The purified free DNA was incubated with 0.5 μl pre-treated M270 streptavidin magnetic beads (Life Technologies) at room temperature for 30 minutes. The M270 streptavidin magnetic beads were pre-blocked with salmon sperm DNA fragments (Life-Technologies) in buffer A (5 mM Tris pH 7.5, 0.5 mM EDTA, 1 M NaCl and 0.2% Tween 20); after the incubation, the magnetic beads were gently washed four times at room temperature, namely: washing with buffer A for 5 minutes, buffer B (buffer A does not contain NaCl) for 5 minutes, buffer C (buffer A, pH 9) and buffer D (buffer C, does not contain NaCl); the washed magnetic beads were suspended in 20 μl nuclease-free water;

[0086] (5) The captured free DNA was amplified and enriched according to the Phusion DNA Polymerase (NEB) operation manual, and the amplification cycle was set to 14 weeks; the amplified PCR product was purified using AMPure XP beads, and the purified library was sequenced according to the paired 75 base pairs in the NextSeq device of the Illumina sequencing platform.

[0087] S5. Perform a metabolomics test on the test sample. The steps of the metabolomics test are as follows:

[0088] (1) Take 0.5 μl of sample serum and place it on the polished steel target plate MTP384. After air drying, add 1 μl of mass spectrometry matrix of three-dimensional nanomaterials for ionization;

[0089] (2) Use MALDI-TOF-MS time-of-flight tandem mass spectrometer to perform mass spectrometry detection on the samples;

[0090] (3) Each sample was randomly measured 25 times at 20 different grating points to obtain 500 satisfactory measurement results;

[0091] (4) Use data analysis software (such as SIMCA-P) and machine learning algorithms to study the serum metabolomics peak spectra of patients with lung nodules and those diagnosed after surgery, determine the metabolomics fingerprint of lung nodules and the characteristic peaks with significant differences between benign and malignant nodules, and further analyze and discover the metabolic pathways in which the differential metabolites are mainly involved and the closely related enzymes

[0092] S6. Collect the above experimental data, conduct automated analysis of lung nodules, multi-omics data analysis and identification model construction; multi-omics data analysis and identification model construction include epigenetic data analysis preprocessing, protein and metabolomics analysis, multi-omics integrated analysis, such as Figure 2 As shown;

[0093] The automated analysis of lung nodules uses the data in S1, and is based on pre-developed artificial intelligence and deep learning lung nodule 3D segmentation technology for automatic analysis. It has the following features: the detection sensitivity of lung nodules larger than 3 mm reaches 97.1% (using the LIDC dataset, including 1013 test data); the false positive rate (FPS) is 2.5 / Scan; accurate anatomical positioning can be achieved; the clinical consistency rate of benign and malignant is high, based on the pathological results of nearly 10,000 cases in multiple centers for control testing;

[0094] The preprocessing steps for epigenetic data analysis are as follows:

[0095] (1) After the epigenetic sequencing data is obtained, the data is quality controlled by Illumina's own tools to obtain the fastq original file;

[0096] (2) The obtained raw files were quality assessed by fastqc (Version 0.11.9), and picard (v2.25.0) was used to remove adapter sequences and duplicate sequences;

[0097] (3) Bowtie2 (v2.4.2) was used to align HCSC / hg19 as the reference sequence. After alignment, samtools (v1.11) was used to deduplicate the sequences (samtools view -f 2 -F 1548 -q 30 | samtoolsrmdup), and only unique sequences were retained for subsequent analysis;

[0098] (4) Use the bedGraphToBigWig application in bedtools (v2.18) to convert the unique sequence tags (sequence pairs consisting of paired pairs) into bedgraph format, and normalize the converted sequences by dividing the number of paired sequences by the total number of paired sequences; use the UCSC Genome Browser to view the converted data;

[0099] (5) Capture efficiency evaluation;

[0100] (6) Through sequence alignment, locate the source region of cfDNA on the genome, thereby inferring the degree of 5hmC modification in the DNA region and performing stacking peak mapping; calculate the corresponding 5hmC levels of genes and chromosome fragments based on genes or chromosome fragments, and further compare the genes or chromosome fragments with different 5hmC modification levels between patients with benign and malignant pulmonary nodules; combine the construction of the 5hmC model and the input of clinical parameters to find potential biological markers (pathways / biological substances, etc.);

[0101] The steps of protein and metabolomics analysis are as follows:

[0102] (1) Collecting 100-5000Da mass spectrometry results, collecting the original mass spectrometry data in the range of 100-5000Da obtained by MALDI-TOF-MS instrument detection, and importing the generated MzML file using MALDIquantForeign to generate a mass spectrometry CSV result that is easy to analyze;

[0103] (2) Perform inter-group and intra-group quality inspection on the obtained CSV results. The qualified data are subjected to noise reduction and image smoothing, baseline adjustment, intensity calibration, peak detection and alignment (such as MassspecWavel and XCMS software) using the software in the MALDIquant package to obtain tens of thousands of metabolite characteristic peaks;

[0104] (3) Create a feature matrix based on the sample grouping, use machine learning or neural network methods to analyze and build models, and find substances with characteristic differences;

[0105] (4) Compare the characteristic differential substances with the proteome and metabolome databases to explore their metabolic pathways in lung cancer samples;

[0106] Multi-omics integrated analysis uses neural networks, deep machine learning methods and artificial intelligence algorithms to extract features, cluster and analyze the data of patients with pulmonary nodules based on the obtained epigenomics, protein metabolomics and original DICOM data of LDCT images, and establishes a corresponding multi-omics prediction model for benign and malignant pulmonary nodules.

[0107] The system in the present invention adopts the research approach of "omics to mechanism to application", uses medical imaging omics as pre-screening data for the research subjects, and performs epigenomics on the peripheral blood samples obtained, combining proteomics and metabolomics, as well as multi-omics research ideas of machine learning, to explore the molecular mechanism of the difference between benign and malignant lung nodules, screen differentially characteristic small molecules, and thus establish a highly sensitive and specific clinical lung cancer diagnosis scheme; the present invention combines imaging omics, self-developed small molecule enrichment nanomaterials, mass spectrometry technology, and artificial intelligence technology, with a data-driven methodological approach to distinguish benign and malignant lung nodules, providing a detailed and reliable methodological theoretical basis for establishing the molecular mechanism of the occurrence and development of the disease.

[0108] It should be understood that the above specific embodiments of the present invention are only used to illustrate or explain the principles of the present invention, and do not constitute a limitation of the present invention. Therefore, any modifications, equivalent substitutions, improvements, etc. made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention. In addition, the appended claims of the present invention are intended to cover all changes and modifications that fall within the scope and boundaries of the appended claims, or the equivalent forms of such scope and boundaries.

Claims

1. A multi-omics detection system for lung cancer, characterized in that: It includes a clinical data collection module, a clinical sample and information acquisition module, and a test data standardization and analysis module; the clinical data collection module collects the original DICOM data of the patient's CT; the clinical sample and information acquisition module includes a rule setting unit, a peripheral blood collection and processing unit, an epigenome test unit, and a proteome and metabolome test unit; the test data standardization and analysis module includes a lung nodule AI automated analysis unit, a multi-omics data analysis unit, and a lung nodule benign and malignant differentiation model construction unit; The working steps are as follows: S1. Patients who are scheduled for surgical pathological examination should undergo low-dose CT examination, with the scanning range from the apex of the lung to the lower edge of the posterior costophrenic angle, including the chest wall and axilla on both sides; S2. Set the detection object rules and establish the test group according to the rules; S3, collecting and processing peripheral blood to obtain test samples; S4. Perform epigenomic testing on the test samples; S5. Perform protein metabolomics testing on the test samples; S6. Collect the above test data, conduct AI automated analysis of lung nodules, multi-omics data analysis and construction of a model for distinguishing benign and malignant lung nodules; multi-omics data analysis and identification model construction include image AI processing and analysis, epigenetic data preprocessing and analysis, protein and metabolomics analysis, and multi-omics integrated analysis; In S2, the rules are as follows: collect clinical information related to patients, including age, gender, smoking history, clinical evaluation results at enrollment, and original DICOM information about lung nodules through CT imaging examination; In S3, peripheral blood collection and processing operations are as follows: (1) The patient's fasting arm venous blood was collected using one tube of EDTA anticoagulant tube and one tube of procoagulant tube respectively; (2) The collected blood was allowed to stand at room temperature for 30 minutes, and all blood samples were separated within 4 hours; blood samples collected from procoagulant tubes were centrifuged at 2000 g / min in a horizontal centrifuge for 15 minutes; blood samples collected from EDTA anticoagulant tubes were centrifuged at 2000 g / min at 4°C for 15 minutes; (3) Use a pipette to aspirate the blood supernatant after centrifugation and dispense it into pre-labeled nuclease-free cryopreservation tubes, 1000 µl per tube; (4) Store the cryopreserved tubes of plasma and serum samples after centrifugation in a -80°C refrigerator. Do not freeze-thaw before extraction to avoid repeated freezing and thawing. (5) Place the obtained samples in a special freezing box and use cold chain transportation to arrive at the laboratory for testing; In S4, the epigenome detection steps are as follows: (1) Extract and quantify free DNA in sample plasma; (2) Internal quality control preparation: Use dNTP mixture to amplify λ-DNA, prepare 3 sets of dNTP mix according to the proportion and use the corresponding amplification primers for amplification; (3) 5hmC library construction, labeling, capture and sequencing: Take 1-10ng of free DNA and add internal quality control amplification product at 0.01pg per 10ng of free DNA; then prepare the library; take 25µl of labeling solution to label the prepared library with 5hmC and incubate at 37°C for 2 hours; then take 2.5µl of DBCO-PEG4-labeled biotin, mix well, and continue incubating at 37°C for 2 hours to complete the labeling; add 10µg of salmon sperm DNA fragments to the labeled library, mix well, and purify using a Micro Bio-Spin 30 separation column; (4) Incubate the purified free DNA with 0.5 µl pre-treated M270 streptavidin magnetic beads at room temperature for 30 minutes. The M270 streptavidin magnetic beads were pre-blocked with salmon sperm DNA fragments in buffer A. After the incubation, the magnetic beads were gently washed four times at room temperature, namely: 5 minutes with buffer A, 5 minutes with buffer B, buffer C and buffer D. The washed magnetic beads were suspended in 20 µl nuclease-free water. (5) The captured free DNA was amplified and enriched, and the amplification cycle was set to 14 weeks; the amplified PCR product was purified using AMPure XP beads, and the purified library was sequenced according to the paired 75 base pairs; In S5, the steps of protein metabolomics experiment are as follows: (1) Take 0.5 µl of sample serum and place it on the polished steel target plate MTP384. After air drying, add 1 µl of the mass spectrometry matrix of the three-dimensional nanomaterial for ionization. (2) Analyze samples using matrix-assisted laser desorption / ionization time-of-flight mass spectrometry; (3) Each sample was randomly measured 25 times at 20 different grating points to obtain 500 satisfactory measurement results; (4) Use data analysis software and machine learning algorithms to study the peak spectra of serum protein metabolomics in patients with pulmonary nodules, determine the characteristic fingerprint of pulmonary nodule protein metabolomics and the characteristic peaks with significant differences between benign and malignant nodules, and further analyze and discover the metabolic pathways involved in the differential protein metabolites and the closely related enzymes; In S6, the AI ​​automated analysis of lung nodules uses the data in S1 and performs automatic analysis based on pre-developed artificial intelligence and deep learning lung nodule 3D segmentation technology and auxiliary diagnosis technology; In S6, the preprocessing steps for epigenetic data analysis are as follows: (1) After the epigenetic sequencing data is obtained, the data is quality controlled by Illumina's own tools to obtain the fastq original file; (2) The obtained raw files were evaluated for quality by FastQC, and Picard was used to remove adapter sequences and duplicate sequences; (3) Use Bowtie2 to align HCSC / hg19 as the reference sequence. After alignment, use samtools to remove duplicates and retain only unique sequences for subsequent analysis. (4) Use the bedGraphToBigWig application in bedtools to convert the unique sequence tags into bedgraph format, and normalize the converted sequences by dividing the number of paired sequences by the total number of paired sequences; use the UCSC Genome Browser to view the converted data; (5) Capture efficiency evaluation; (6) Through sequence alignment, locate the source region of cfDNA on the genome, thereby inferring the degree of 5hmC modification in the DNA region and performing stacking peak mapping; calculate the corresponding 5hmC levels of genes and chromosome fragments based on genes or chromosome fragments, and further compare genes or chromosome fragments with different 5hmC modification levels between patients with benign and malignant pulmonary nodules; combine 5hmC model construction and clinical parameter input to find potential biological markers; In S6, the steps of protein and metabolomics analysis are as follows: (1) Collect the mass spectrometry results of 100-5000Da. Collect the raw mass spectrometry data of 100-5000Da obtained by MALDI-TOF-MS instrument detection, and use MALDIquantForeign to import the generated MzML file to generate a mass spectrometry CSV result that is easy to analyze; (2) Perform inter-group and intra-group quality inspection on the obtained CSV results, and perform noise reduction, image smoothing, baseline adjustment, intensity calibration, peak detection and alignment on the data that pass the quality inspection, thereby obtaining tens of thousands of metabolite characteristic peaks; (3) Create a feature matrix based on the sample grouping, use machine learning or neural network methods to analyze and build models, and find substances with characteristic differences; (4) Compare the characteristic differential substances with the proteome and metabolome databases to explore their metabolic pathways in lung cancer samples; In S6, multi-omics integrated analysis targeted the acquired epigenomics, protein metabolomics, and LDCT image data, and used neural networks, deep machine learning methods, and artificial intelligence algorithms to extract features, cluster, and analyze the data of patients with lung nodules, and established a corresponding multi-omics model for differential diagnosis of benign and malignant lung nodules.

Images