A method for constructing a dual-omics plasma fingerprint model and an infection diagnosis device of a fingerprint model constructed by using the method

Abstract

The present invention discloses a method for constructing a dual-omics plasma fingerprint model and an infection diagnosis device of the dual-omics plasma fingerprint model constructed by using this method, which relates to the technical field of infection disease diagnosis. The construction method includes preprocessing plasma samples, preparing matrix materials, spotting samples and matrixes on a mass spectrometry target plate, collecting plasma metabolite fingerprints in nanoparticle-enhanced laser desorption ionization mass spectrometry, and collecting plasma protein fingerprints in organic matrix material-assisted laser desorption ionization mass spectrometry. The infection diagnosis device performs data analysis on the dual-omics plasma fingerprint map to obtain a diagnosis result, and has advantages such as high detection reproducibility, fast analysis speed, and less sample consumption, realizing high-efficiency diagnostic performance for different infection types.

Description

Technical Field

[0001] The present invention relates to the technical field of infectious disease detection, and particularly to a method for constructing a dual-omics plasma fingerprint model and an infectious disease diagnosis device with a fingerprint model constructed by using the method. Background Art

[0002] Infection remains one of the leading causes of human disease and death, contributing to more than 20% of the global disease burden. Notably, the general classification of infections guides the selection of treatment plans, such as the use of antibiotics or inhibitors, which is crucial for the recovery and outcome of patients. However, it is very challenging to clinically distinguish between bacterial and viral infections early and accurately because the clinical symptoms of bacterial and viral infections are similar and there is a lack of high-performance technologies for differentiating between the two types of infections. Therefore, overuse or underuse of drugs usually brings serious health problems (such as antibiotic resistance) and economic consequences. For example, 23.1 - 51.4% of antibiotics are misused globally each year, and if no action is taken, antibiotic resistance will cause 10 million deaths per year by 2050. Therefore, as proposed by the World Health Organization's global initiative, there is an urgent need to develop new diagnostic tools aimed at rapidly differentiating between bacterial and non-bacterial infections. These tools are very important in selecting drugs for infectious diseases and customizing personalized treatments to reduce morbidity and mortality.

[0003] Traditionally, pathogen-centered diagnostic methods require prior knowledge of the pathogen to be tested and rely on complex and time-consuming sample pretreatment for target enrichment of the pathogen, such as bacterial culture (at least 1 - 2 days) and polymerase chain reaction amplification (about several hours), to address the limitations of low pathogen content to be tested and the complexity of the biological environment. Due to the host inflammatory response, the characterization of the host response is a valuable way to determine the presence and type of high-risk pathogen infections and is more amenable to clinical intervention. Host inflammatory markers, such as procalcitonin, serum C-reactive protein, and white blood cells, are commonly used to evaluate suspected infections. However, due to the heterogeneity of microbiological etiologies and individual differences in host responses, the large differences in a single biomarker among infected patients reduce the discriminative power and credibility of these markers. Notably, omics analysis is crucial for studying complex immune responses compared to single biomarkers.

[0004] Proteins and metabolites provide different understandings of the host immune response and are closer to the disease phenotype compared to nucleic acids. Proteomics analysis reflects the dynamic balance among transcription, translation, modification, etc., and metabolomics analysis studies the end products of pathways. The integration of metabolic and proteomics datasets will draw an overall picture of the human phenotype and explain the complexity of the biological system for infection classification.

[0005] The development of advanced dual-omics fingerprint acquisition methods is essential for realizing rapid and accurate clinical applications of infection classification. Mass spectrometry-based methods are the most commonly used techniques for large-scale studies of metabolic and proteomic biomarkers. However, the application of traditional mass spectrometry in the metabolic and proteomic analysis of complex biological fluids depends on separation methods, such as chromatography, which is used to reduce sample complexity. Recently, organic matrix (e.g., α-cyano-4-hydroxycinnamic acid)-assisted laser desorption ionization mass spectrometry can directly obtain plasma proteomic fingerprints of infected patients. Inorganic matrix (nanoparticle)-enhanced laser desorption ionization mass spectrometry can achieve high-performance metabolic fingerprint extraction of biological fluids. To date, although laser desorption ionization mass spectrometry has the potential to combine metabolic and proteomic analyses, reports on dual-omics fingerprints are still scarce. Therefore, a reasonable design is needed to achieve high-performance diagnostic applications, including but not limited to infection classification. Summary of the Invention

[0006] In view of the above-mentioned defects of the prior art, the present invention provides a method for constructing a dual-omics plasma fingerprint model and an infection diagnosis device using the dual-omics plasma fingerprint model constructed by this method, aiming to achieve efficient, accurate, and rapid classification of infections.

[0007] In an embodiment of the present invention, a method for constructing a dual-omics plasma fingerprint model includes the following steps:

[0008] Step 1: Pretreatment of plasma samples for metabolic fingerprint acquisition;

[0009] Step 2: Pretreatment of plasma samples for protein fingerprint acquisition;

[0010] Step 3: Preparation of inorganic nanoparticles for metabolic fingerprint acquisition;

[0011] Step 4: Sample preparation on a mass spectrometry target plate, spotting 1 µL of each pretreated sample, and drying at room temperature;

[0012] Step 5: Matrix preparation on a mass spectrometry target plate, spotting 1 µL of the matrix solution, and drying at room temperature;

[0013] Step 6: Collection of plasma metabolic fingerprints in nanoparticle-enhanced laser desorption ionization mass spectrometry;

[0014] Step 7: Collection of plasma protein fingerprints in organic matrix-assisted laser desorption ionization mass spectrometry;

[0015] Step 8: Machine learning of dual-omics plasma fingerprints.

[0016] Further, step 1 is specifically to mix 30 µL of plasma samples with 30 µL of metabolite extraction solution (a solution of equal volume mixture of methanol and acetonitrile) in equal volume, take the supernatant after centrifugation, and save it for later use.

[0017] Further, step 2 specifically involves diluting the plasma sample 20 times with deionized water and storing it for later use.

[0018] Further, step 3 specifically involves a method for preparing an iron oxide inorganic nanoparticle matrix, which includes the following steps: successively adding sodium citrate, ferric chloride, and sodium acetate to a solution of ethylene glycol, ultrasonically dispersing the mixture, transferring the mixed solution to a Teflon high-pressure reactor, reacting at 100 - 300 °C for 8 hours, rinsing the product with ethanol and deionized water, and finally drying it at 60 °C for use to obtain the iron oxide inorganic nanoparticles as the inorganic matrix material.

[0019] Further, step 5 specifically includes:

[0020] Step 5.1: Prepare a matrix solution of 1 mg / mL of the inorganic nanoparticles with deionized water;

[0021] Step 5.2: Dissolve the organic matrix (α-cyano-4-hydroxycinnamic acid) in a 50% aqueous acetonitrile solution to form a saturated solution as the matrix solution.

[0022] Further, step 6 specifically involves: For MALDI mass spectrometry detection of metabolic fingerprints, the reflection mode is used, positive ion detection is performed, the detection range is set to 100 - 1000 Da, the instrument model is Autoflex MALDI-TOF ( / TOF)-MS (Bruker AutoflexSpeed), and the specific parameters are: laser wavelength 355 nm, laser frequency 2 kHz; acceleration voltage 20 kV, repetition rate of delayed extraction 1 kHz; delay time 150 ns; 2000 laser irradiations are superimposed for each analysis.

[0023] Further, step 7 specifically involves: For MALDI mass spectrometry detection of metabolic fingerprints, the linear mode is used, positive ion detection is performed, the detection range is set to 1000 - 10000 Da, the instrument model is Autoflex MALDI-TOF ( / TOF)-MS (Bruker Autoflex Speed), and the specific parameters are: laser wavelength 355 nm, laser frequency 1 kHz; acceleration voltage 20 kV, repetition rate of delayed extraction 1 kHz; delay time 150 ns; 2000 laser irradiations are superimposed for each analysis.

[0024] The present invention further includes an infection diagnosis device using a dual-omics plasma fingerprint model constructed by the method described in any one of the foregoing, characterized in that it includes a preprocessing module, a data fusion module, a training module, and a testing module. The preprocessing module preprocesses the plasma metabolite fingerprint collected in step 6 of the method described in any one of the foregoing and the plasma protein fingerprint collected in step 7 of the method described in any one of the foregoing on Python, and respectively obtains metabolite m / z signals and protein m / z signals. The data fusion module performs dual-omics fingerprint data fusion on the metabolite m / z signals and protein m / z signals. The training module divides the samples into a training set and a testing set, and uses machine learning algorithms such as deep learning on Python to perform model training on the training set to obtain the diagnostic performance of the model on the training set. The testing module uses the trained model to make predictions on the testing set to obtain the diagnostic performance on the testing set.

[0025] Further, the preprocessing includes spectral smoothing, baseline correction, and spectral peak alignment.

[0026] Further, the samples include virus infection samples and bacterial infection samples.

[0027] The beneficial effects of the present invention compared with other infection detection methods are as follows:

[0028] 1. It has high detection reproducibility (the coefficient of variation of the characteristic intensity in the sample is <15% for about 95% of the samples), fast analysis speed (about 30 seconds per sample), and low sample consumption (about 550 nL per sample).

[0029] 2. Through the deep learning of dual-omics plasma fingerprints, we have achieved high-efficiency differential diagnostic performance for bacterial infections and virus infections, with an area under the curve of 0.796, which has a significant advantage over single-omics data (p < 0.05). In addition, high-performance diagnosis for novel coronavirus infection has been achieved, with an area under the curve of 0.917.

[0030] The technical solutions and the resulting technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose and effects of the present invention. Description of the Drawings

[0031] Figure 1 It is a representative spectrogram of the plasma metabolite fingerprint and protein fingerprint of an infected patient in an embodiment of the present invention;

[0032] Figure 2 It is a heat map corresponding to 146 m / z signals after preprocessing the dual-omics plasma fingerprints of 301 samples (including bacterial infections and virus infections) in an embodiment of the present invention;

[0033] Figure 3Schematic diagram for efficient differential diagnosis of bacterial and viral infections in an embodiment of the present invention;

[0034] Figure 4 Schematic diagram for high-performance diagnosis of novel coronavirus infection in an embodiment of the present invention. Detailed implementation manners

[0035] The preferred embodiments of the present invention are introduced below with reference to the accompanying drawings of the specification to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the protection scope of the present invention is not limited to the embodiments mentioned in the text.

[0036] Example 1: Use nanoparticle-enhanced laser desorption ionization mass spectrometry technology to collect plasma metabolite fingerprints, and use organic matrix-assisted laser desorption ionization mass spectrometry to collect plasma protein fingerprints. The specific steps are as follows:

[0037] Step 1: Pretreatment of plasma samples for metabolite fingerprint acquisition: Mix 30 μL of plasma samples with 30 μL of metabolite extraction solution (a solution of equal volume of methanol and acetonitrile) in equal volume, take the supernatant after centrifugation, and store it for later use;

[0038] Step 2: Pretreatment of plasma samples for protein fingerprint acquisition: Dilute the plasma samples 20 times with deionized water and store them for later use;

[0039] Step 3: Preparation of inorganic nanoparticles for metabolite fingerprint acquisition: Add sodium citrate, ferric chloride, and sodium acetate to the solution of ethylene glycol in sequence and disperse them by ultrasonic wave. Transfer the mixed solution to a Teflon high-pressure reaction kettle, react at 100 - 300 degrees Celsius for 8 hours, wash the product with ethanol and deionized water, and finally dry it at 60 degrees Celsius for later use to obtain the iron oxide inorganic nanoparticles as the inorganic matrix material;

[0040] Step 4: Perform sample preparation on the mass spectrometry target plate, spot 1 μL of each pretreated sample, and dry it at room temperature;

[0041] Step 5: Perform matrix preparation on the mass spectrometry target plate, spot 1 μL of the matrix solution, and dry it at room temperature. Among them, the inorganic nanoparticles are prepared into a matrix solution with a concentration of 1 mg / mL with deionized water, and the organic matrix (α-cyano-4-hydroxycinnamic acid) is dissolved in a 50% aqueous acetonitrile solution to form a saturated solution as the matrix solution;

[0042] Step 6: Collection of plasma metabolic fingerprints in nanoparticle-enhanced laser desorption / ionization mass spectrometry. The metabolic fingerprints were detected by MALDI mass spectrometry in reflectron mode, with positive ion detection. The detection range was set to 100 - 1000 Da, and the instrument model was Autoflex MALDI-TOF ( / TOF)-MS (Bruker Autoflex Speed). The specific parameters were as follows: laser wavelength 355 nm, laser frequency 2 kHz; acceleration voltage 20 kV, repetition rate of delayed extraction 1 kHz; delay time 150 ns; 2000 laser irradiations were superimposed for each analysis;

[0043] Step 7: Collection of plasma protein fingerprints in organic matrix-assisted laser desorption / ionization mass spectrometry. The metabolic fingerprints were detected by MALDI mass spectrometry in linear mode, with positive ion detection. The detection range was set to 1000 - 10000 Da, and the instrument model was Autoflex MALDI-TOF ( / TOF)-MS (Bruker Autoflex Speed). The specific parameters were as follows: laser wavelength 355 nm, laser frequency 1 kHz; acceleration voltage 20 kV, repetition rate of delayed extraction 1 kHz; delay time 150 ns; 2000 laser irradiations were superimposed for each analysis;

[0044] Example 2: Machine learning was performed on the dual-omics plasma fingerprints of patients with bacterial infections and viral infections to achieve infection classification. Specifically, the infection classification was achieved through an infection diagnosis device of the dual-omics plasma fingerprint model constructed by the method described in Example 1. The device included a preprocessing module, a data fusion module, a training module, and a testing module. The preprocessing module preprocessed the plasma metabolic fingerprints collected in Step 6 and the plasma protein fingerprints collected in Step 7 on Python, and respectively obtained metabolic m / z signals and protein m / z signals; the data fusion module performed dual-omics fingerprint data fusion on the metabolic m / z signals and protein m / z signals; the training module divided the samples into a training set and a testing set, and used machine learning algorithms such as deep learning to perform model training on the training set on Python to obtain the diagnostic performance of the model on the training set; the testing module used the trained model to make predictions on the testing set to obtain the diagnostic performance on the testing set. The specific steps were as follows:

[0045] Step 1: Through Example 1, dual-omics plasma fingerprints were collected from 301 infected patients (175 with viral infections and 126 with bacterial infections), and the results were as shown in the appendix Figure 1 as follows;

[0046] Step 2: Preprocessing was performed on Python, including spectral smoothing, baseline correction, and spectral peak alignment, to obtain 123 metabolic m / z signals and 23 protein m / z signals, and the results were as shown in the appendix Figure 2as shown;

[0047] Step 3: Divide 301 patients into a training set (201 samples, including 116 virus infections and 85 bacterial infections) and a test set (100 samples, including 59 virus infections and 41 bacterial infections);

[0048] Step 4: Use a deep learning algorithm on Python to train a model on 146 m / z signals in the training set to obtain the diagnostic performance of the model on the training set;

[0049] Step 5: Use the model trained in Step 4 to make predictions on the test set to obtain the diagnostic performance on the test set ( Figure 3 ).

[0050] Example 3: Perform machine learning on the dual-omics plasma fingerprints of COVID-19 patients and healthy controls to achieve COVID-19 diagnosis.

[0051] The specific steps are as follows:

[0052] Step 1: Through Example 1, collect dual-omics plasma fingerprints from 138 research subjects (31 COVID-19 infections and 107 control groups);

[0053] Step 2: Perform preprocessing on Python, including spectral line smoothing, baseline correction, and spectral peak alignment, to obtain 123 metabolic m / z signals and 23 protein m / z signals;

[0054] Step 3: Divide the research subjects into a training set (77 samples, including 22 COVID-19 infections and 55 control groups) and a test set (57 samples, including 9 COVID-19 infections and 48 control groups);

[0055] Step 4: Use a deep learning algorithm on Python to train a model on 146 m / z signals in the training set to obtain the diagnostic performance of the model on the training set;

[0056] Step 5: Use the model trained in Step 4 to make predictions on the test set to obtain the diagnostic performance on the test set ( Figure 4 ).

[0057] The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those of ordinary skill in the art can make many modifications and variations based on the concept of the present invention without creative labor. Therefore, all technical solutions that can be obtained by those skilled in the art in the technical field based on the concept of the present invention through logical analysis, reasoning, or limited experiments on the basis of the prior art should fall within the protection scope determined by the claims.

Claims

1. A method for constructing a dual-omics plasma fingerprint model, characterized in that, It includes the following steps: Step 1, Pretreatment of plasma samples for metabolic fingerprint acquisition; Step 2, Pretreatment of plasma samples for protein fingerprint acquisition; Step 3, Preparation of inorganic nanoparticles for metabolic fingerprint acquisition; Step 4, Sample preparation on a mass spectrometry target plate, spotting 1 µL of each pretreated sample and drying at room temperature; Step 5, Matrix preparation on a mass spectrometry target plate, spotting 1 µL of the matrix solution and drying at room temperature; Step 6, Collection of plasma metabolic fingerprints in nanoparticle-enhanced laser desorption ionization mass spectrometry; Step 7, Collection of plasma protein fingerprints in organic matrix-assisted laser desorption ionization mass spectrometry; Step 8, Pretreat the plasma metabolic fingerprints collected in Step 6 and the plasma protein fingerprints collected in Step 7 to obtain metabolic m / z signals and protein m / z signals respectively; perform dual-omics fingerprint data fusion on the metabolic m / z signals and protein m / z signals; divide the samples into a training set and a test set, use deep learning and machine learning algorithms on Python to train the model on the training set, and obtain the diagnostic performance of the model on the training set; use the trained model to make predictions on the test set to obtain the diagnostic performance on the test set, and achieve the classification of viral infections and bacterial infections; The plasma samples include viral infection samples and bacterial infection samples; Step 3 is specifically a preparation method of an iron oxide inorganic nanoparticle matrix, which includes the following steps: sequentially add sodium citrate, ferric chloride, and sodium acetate into a solution of ethylene glycol and ultrasonically disperse them, transfer the mixed solution to a Teflon autoclave, react at 100 - 300 degrees Celsius for 8 hours, wash the product with ethanol and deionized water, and finally dry it at 60 degrees Celsius for use to obtain the iron oxide inorganic nanoparticles as an inorganic matrix material; Step 5 is specifically: Step 5.1, Prepare a 1 mg / mL matrix solution of the inorganic nanoparticles with deionized water; Step 5.2, Dissolve the organic matrix in a 50% acetonitrile aqueous solution to form a saturated solution as the matrix solution, and the organic matrix is α-cyano-4-hydroxycinnamic acid.

2. The method for constructing a dual-omics plasma fingerprint model according to claim 1, characterized in that, Step 1 is specifically to mix 30 µL of plasma sample and 30 µL of metabolite extraction solution in equal volume, take the supernatant after centrifugation, and save it for later use. The metabolite extraction solution is a mixed solution of methanol and acetonitrile in equal volume.

3. The construction method of the dual-omics plasma fingerprint model according to claim 1, characterized in that, Step 2 is specifically to dilute the plasma sample 20 times with deionized water and save it for later use.

4. The method for constructing a dual-omics plasma fingerprint model according to claim 1, characterized in that, Step 6 is specifically: For MALDI mass spectrometry detection of metabolic fingerprints, the reflection mode is used, positive ion detection, and the detection range is set to 100 - 1000 Da.

5. The method for constructing the dual-omics plasma fingerprint model according to claim 1, characterized in that Step 7 is specifically: For MALDI mass spectrometry detection of protein fingerprints, the linear mode is used, positive ion detection, and the detection range is set to 1000 - 10000 Da.

6. The method for constructing the dual-omics plasma fingerprint model according to claim 1, characterized in that, The pretreatment in Step 8 includes spectral smoothing, baseline correction, and spectral peak alignment.

Images