Rheumatoid arthritis diagnosis system application and method based on high-entropy material assisted LDI MS combined with machine learning

Abstract

The invention relates to the technical field of metabonomics analysis, in particular to application and method of a rheumatoid arthritis diagnosis system based on high-entropy material assisted LDI MS combined with machine learning. On the basis of a high-entropy material TiVCrMoC3, a serum metabolism fingerprint spectrogram of a rheumatoid arthritis patient is extracted by utilizing a laser desorption ionization mass spectrometry technology; and analyzing the serum metabolism fingerprint spectrogram of the rheumatoid arthritis patient in combination with machine learning, screening and identifying differential metabolites, and realizing early-stage rapid and accurate diagnosis of the rheumatoid arthritis patient through combined diagnosis of the differential metabolites. The high-entropy material is used as a matrix for serum LDI MS analysis and has a good application prospect. According to the method, the high-throughput, high-sensitivity, high-signal-intensity and high-repeatability detection performance of the serum metabolism fingerprint spectrogram is remarkably improved, and the problems that in the traditional technology, the demand quantity of samples is large, and the pretreatment process is complex are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of metabolomics analysis, and in particular to an application and method of a rheumatoid arthritis diagnosis system combining high entropy material-assisted LDI MS with machine learning. Background Art

[0002] Rheumatoid arthritis (RA) is an autoimmune disease that originates in the synovium and involves multiple inflammatory cells, leading to joint dysfunction and even disability. Currently, serum markers for RA primarily include rheumatoid factor (RF) and anti-cyclic citrullinated antibodies (anti-CCP antibodies), but their low specificity makes early diagnosis difficult and often leads to misdiagnosis or missed diagnosis. Therefore, there is an urgent need to identify specific early diagnostic markers for RA, develop new diagnostic technologies, and establish an early diagnosis and early warning system for RA.

[0003] Metabolomics can reflect an individual's current physiological state by analyzing the metabolic characteristics of tissues and cells. Compared with proteomics and genomics, metabolomics has more advantages in clinical personalized diagnosis and treatment. In particular, blood metabolomics research can reveal systemic metabolic regulatory changes caused by the disease, providing important potential for the diagnosis of RA. However, due to the complex blood environment and low metabolite abundance, traditional blood testing methods have many limitations in practical applications. Therefore, the development of blood testing technology that can accurately reflect metabolic changes is of great significance for the diagnosis of RA.

[0004] Mass spectrometry (MS) has become an essential tool in metabolomics research due to its exceptional sensitivity and broad metabolite identification capabilities. Among the numerous mass spectrometry techniques, matrix-assisted laser desorption / ionization mass spectrometry (MALDI-MS) is widely used for the analysis of high-molecular-weight compounds (m / z > 1000 Da) due to its high throughput, high sensitivity, and high accuracy. MALDI-MS can obtain molecular information on a large amount of known and unknown substances from biological samples and distinguish subtle differences between molecules, opening the possibility of analyzing RA-specific small-molecule metabolites. However, traditional organic matrices present background interference in the analysis of low-molecular-weight compounds (m / z < 700 Da), severely limiting its application. Nanomaterial-assisted surface-assisted laser desorption / ionization mass spectrometry (SALDI-MS) significantly reduces background signal interference, making the analysis of low-molecular-weight compounds (m / z < 500 Da) possible. TiVCrMoC3, a novel high-entropy two-dimensional carbide MXene material composed of four transition metals (Ti, V, Cr, and Mo) in equal molecular ratios, exhibits excellent electrical conductivity, photothermal properties, and stability, enabling efficient laser desorption ionization (LDI) analysis and extraction of metabolic fingerprints from serum. Machine learning (ML) techniques can extract features and perform dimensionality reduction analysis from large amounts of mass spectrometry data to identify key metabolite differences. Leveraging pattern recognition and classification algorithms, ML can process high-dimensional, complex data, enabling efficient disease prediction and classification. Therefore, combining TiVCrMoC3-assisted LDI-MS analysis with machine learning is expected to be an effective tool for identifying and quantifying differential metabolites in the serum of RA patients. This technology offers new possibilities for rapid diagnosis and large-scale screening of RA, while also helping to develop personalized treatment plans for patients and promoting the development of precision medicine. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem of difficulty in early diagnosis of rheumatoid arthritis. The present invention provides a serum metabolic fingerprint extraction method assisted by LDI MS using a new high-entropy material TiVCrMoC3. This method is combined with machine learning technology to detect serum from rheumatoid arthritis patients and mine potential metabolic markers.

[0006] The present invention is achieved through the following technical solutions: In a first aspect, the present invention provides a method for applying a high-entropy material-assisted LDI MS combined with machine learning rheumatoid arthritis diagnostic system, comprising the following steps: Step 1: prepare high entropy material TiVCrMoC3 and configure it into a suspension; Step 2: Using TiVCrMoC3 as a matrix-assisted LDI MS, the metabolic fingerprint of RA patient serum was extracted; Step 3: Perform machine learning on the extracted metabolic fingerprint to build a disease diagnosis model and biomarker identification.

[0007] Furthermore, step 1 is as follows: Step 1.1, TiVCrMoAlC3MAX powder is etched with hydrofluoric acid and centrifuged, and washed several times until the supernatant is neutral to obtain multilayer TiVCrMoC3; Step 1.2: Add TMAOH solution to the multilayer TiVCrMoC3 obtained in step 1.1 and stir at room temperature to allow sufficient intercalation; Step 1.3: Disperse the precipitate obtained in step 1.2 in deionized water, sonicate in an ice-water bath, centrifuge, collect the supernatant, concentrate, and freeze-dry to obtain a single-layer solid TiVCrMoC3.

[0008] Furthermore, in step 1.1, the TiVCrMoAlC3MAX powder is 2 g, the hydrofluoric acid concentration is 48 wt %, and the amount is 20 ml; the etching conditions are stirring at 55° C. for 4 days, and the centrifugation conditions after etching are 4200 rpm for 5 min.

[0009] Furthermore, in step 1.2, the concentration of the TMAOH solution was 5 wt %, the amount used was 20 ml, and the mixture was stirred at room temperature for 6 hours.

[0010] Furthermore, the TiVCrMoC3 monolayer solid in step 1.3 was prepared into a 1 mg / mL solution using deionized water. The centrifugation condition after ultrasonication was 4200 rpm for 5 min.

[0011] Furthermore, the specific method of step 2 is as follows: Step 2.1, collecting blood samples from RA patients and healthy volunteers as analytical samples; Step 2.2: Centrifuge the blood sample to obtain the supernatant (serum), dilute it 50-fold with methanol, and centrifuge to obtain the supernatant; Step 2.3: Take 1 μL of the serum sample diluted in step 2.2, spot it on the LDI MS target plate, and dry it at room temperature; Step 2.4: Prepare a solution of TiVCrMoC3, spot 1 μL of the solution on an LDI MS target plate, and dry at room temperature. Step 2.5, fingerprint spectrum extraction: extract the metabolic fingerprint spectrum of the sample in MALDI MS.

[0012] Furthermore, the laser desorption ionization mass spectrometry analysis in step 2.5 was performed as follows: all samples were analyzed in positive ion mode using a Rapiflex MALDITissuetyper™ TOF / TOF mass spectrometer. The mass spectrometer used a nitrogen laser equipped with a 355 nm Smartbeam II laser device; the laser operated at a repetition rate of 5000 Hz, the data acquisition range covered m / z 80-1000, and SALDI MS peak maps were generated using FlexAnalysis software.

[0013] Furthermore, the step 3 specifically includes: 3.1. Perform peak extraction, baseline calibration, smoothing, peak alignment and normalization on the metabolic fingerprint obtained in step 2.5; 3.2. Gradient boosting and component linear models, random forest, least absolute shrinkage and selection operator, NaiveBayes, and XGBoost algorithms were used to model the data, and the average performance of the models was evaluated by precision, recall, F1 score, accuracy, and AUC. 3.3. The model with satisfactory evaluation results is applied to the analysis of serum metabolic fingerprint.

[0014] Furthermore, the conditions for screening serum metabolic markers in step 3 were: Foldchange (FC) ≥ 2 or ≤ 0.5, pvalue < 0.05, and Variable important in projection (VIP) ≥ 1, which were used as serum metabolic markers.

[0015] In a second aspect, the present invention provides an application of the high entropy material TiVCrMoC3, wherein the application is selected from any one of the following 1)-3): 1) Application in the analysis and detection of metabolites in biological samples for non-diagnostic and therapeutic purposes; 2) Application in the preparation of reagents or kits for analyzing and detecting metabolites in biological samples; 3) Application in screening metabolic markers of metabolic diseases.

[0016] Compared with the prior art, the present invention is beneficial in that 1. The performance of the novel high-entropy two-dimensional carbide MXenes material TiVCrMoC3-assisted LDI MS prepared in this invention is determined by four aspects: (1) the field-enhanced electron transport structure is suitable for photon-induced desorption; (2) the low-heat dissipation elemental composition is suitable for thermally driven desorption; (3) it has nanoscale surface roughness and good crystal stability, which is conducive to the selective ionization of metabolites; (4) it is easy to synthesize, low-cost, and suitable for industrial production. With these advantages, TiVCrMoC3-assisted LDI MS shows excellent performance in small molecule detection, with ultra-high sensitivity, excellent reproducibility, quantitative detection capability, clean background, and wide analyte coverage, which can realize the extraction of RA serum metabolic fingerprint spectrum; 2. The present invention identified 33 serum differential small molecule metabolites through machine learning of RA serum metabolic fingerprint spectra, which is expected to be applied to the early diagnosis of RA in clinical practice, solving the difficulties in early diagnosis of RA and establishing an early diagnosis and early warning system for RA. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below with reference to the accompanying drawings.

[0018] The accompanying drawing is a schematic structural diagram of the present invention.

[0019] Figure 1 This is a scanning transmission electron microscope photograph of the high entropy material TiVCrMoC3 according to Example 1 of the present invention.

[0020] Figure 2 This is an atomic force microscope photograph of the high entropy material TiVCrMoC3 according to Example 1 of the present invention.

[0021] Figure 3 This is the X-ray photoelectron spectrum of the high entropy material TiVCrMoC3 according to Example 1 of the present invention.

[0022] Figure 4 This is the Fourier transform infrared spectrum of the high entropy material TiVCrMoC3 of Example 1 of the present invention.

[0023] Figure 5 This is the UV-visible absorption spectrum of the high entropy material TiVCrMoC3 in Example 1 of the present invention and the temperature change curve under ultraviolet irradiation.

[0024] Figure 6 Thermal imaging image and local heating changes of the high entropy material TiVCrMoC3 of Example 1 of the present invention.

[0025] Figure 7 These are representative mass spectra of the serum of healthy people and rheumatoid arthritis patients in Example 2 of the present invention.

[0026] Figure 8 This is a heat map of serum markers screened out by machine learning analysis in Example 3 of the present invention. DETAILED DESCRIPTION

[0027] To make the present invention more clearly understood, preferred embodiments are described in detail below with reference to the accompanying drawings.

[0028] Example 1 Preparation method of high entropy material TiVCrMoC3: (1) First, in order to fully etch the material, 2g of TiVCrMoAlC3MAX powder was slowly added to 20ml of HF (48wt%) solution and stirred at 55°C for 4 days. The etching solution was then centrifuged (4200 rmp, 5 min) and washed several times until the supernatant was neutral. The resulting precipitate was multilayer TiVCrMoC3. The multilayer TiVCrMoC3 was added to 20ml of 5wt% TMAOH solution and stirred at room temperature for 6 hours to allow sufficient intercalation. It was then centrifuged and washed three times to remove excess TMAOH. Finally, the precipitate was dispersed in 20mL of deionized water, ultrasonicated in an ice water bath for 30 min, centrifuged, the supernatant collected, concentrated, and freeze-dried to obtain a single-layer solid TiVCrMoC3.

[0029] (2) The microstructure and elemental composition of the high entropy material were characterized. The results are as follows: Figure 1-3 As shown, Figure 1 This scanning transmission electron micrograph of the high-entropy material shows thin, irregular flakes with no apparent surface defects, pores, or oxide impurities. Ti, V, Cr, Mo, and C are evenly distributed across the surface. The edges of the sheet are slightly curled, indicating good flexibility. Figure 2 This is an atomic force microscope photo of the high-entropy material, showing that TiVCrMoC3 is a thin flake with a lateral size of 0.5 μm and a thickness of about 0.25 nm. Figure 3 The X-ray photoelectron spectrum of the high-entropy material shows that the c1s high-resolution spectrum shows the presence of characteristic peaks of C−Mo / Ti−Tx (282.3 eV), C−Ti / MoTx (282.7 eV), C−C−CHX (285.5 eV), C−O (286.3 eV) and COO (287.6 eV) in TiVCrMoC3. The ti2p spectrum shows two main peaks, which highlight that Ti exists mainly in the form of Ti−C (454.8 eV) and Ti−C (459.3 eV). 4+ The small peak at ) is due to the formation of a small amount of oxidized TiO2 phase on the surface of the thin film. The v2p spectrum shows that V is V 2+ (513.6eV) / V4+ (515.4eV) or V2O3 (517.1eV), while the Mo 3d region shows that Mo exists in the form of Mo 5+ (20.5 ev) / Mo 6+ (232eV), C−Mo−Tx (229.3eV) and Mo (228eV) exist in TiVCrMoC3. 2+ It exists in the form of Cr0 (574.8eV) and Cr0 (576.4eV).

[0030] (3) The photothermal properties of the high entropy material were characterized. The results are as follows: Figure 4-6 As shown, Figure 4 The Fourier transform infrared spectrum of the high entropy material is used to characterize the functional groups associated with the TiVCrMoC3 surface. The absorption peaks are 3434.71 cm −1 and 1632.95 cm −1 , which is caused by the stretching vibration of -OH. In addition, the methylene -CH2 (2923.04 cm −1 and 2853.07 cm −1 ), which may be caused by the stretching vibration of methylene on the sample surface. −1 The absorption peak at 1738.89 cm is attributed to the stretching of C-OH and CN. −1 The absorption peak at 1462 cm may be due to the stretching of the C=O of the acetaldehyde group. −1 The absorption peak at is caused by the bending of methyl and methylene groups. Figure 5 The UV-visible absorption spectrum of the high-entropy material and its temperature change curve under UV irradiation are shown. The UV-visible spectrum shows that the material exhibits strong absorption in the UV region, which gradually decreases as the wavelength shifts into the visible light range. Under UV irradiation, TiVCrMoC3 exhibits efficient photothermal conversion, with the temperature rapidly rising from room temperature to 65°C within 500 seconds, ultimately stabilizing at around 69°C. Figure 6 The thermal imaging image and local heating changes of the high-entropy material show a uniform temperature distribution on the sample surface, and the local heating phenomenon shows a distinct temperature peak. These results indicate that the high-entropy material has excellent performance and is conducive to LDI MS analysis and detection.

[0031] Example 2 The high-entropy material TiVCrMoC3 obtained in Example 1 was prepared into a 1 mg / mL solution in deionized water for laser desorption / ionization mass spectrometry (LDI MS) analysis of metabolites in serum samples from 20 patients (10 with rheumatoid arthritis and 10 healthy volunteers) at the First Affiliated Hospital of Shandong First Medical University (Qianfoshan Hospital, Shandong Province). After fasting overnight for at least 8 hours, the donors' skin was disinfected and 5 ml of peripheral venous blood was collected and stored in a purple vacuum anticoagulant collection tube. The blood was then transferred to a 15 ml centrifuge tube and centrifuged (4000 rpm for 10 min). The supernatant (serum) was collected and stored at -80°C in an epppendorf (EP) tube.

[0032] (1) The high entropy material TiVCrMoC3 obtained in Example 1 was added to deionized water and ultrasonically dissolved at room temperature to obtain a uniformly mixed TiVCrMoC3 solution.

[0033] (2) Dilute the serum sample 50 times with methanol, centrifuge and take the supernatant as the SALDI MS sample.

[0034] (3) SALDI mass spectrometry was performed on a Rapiflex MALDI Tissuetyper™ TOF / TOF mass spectrometer (CarbrookeDalton, Billerica, MA, USA), and all samples were analyzed in positive ion mode. The mass spectrometer uses a nitrogen laser and is equipped with a 355 nm Smartbeam II laser device. The laser operates at a repetition rate of 5000 Hz, and the data acquisition range covers m / z 80-1000. Mass spectrometry calibration was performed using a mixture of DHB and CHCA matrices provided by Brooke Dalton, USA. Before SALDI MS detection of serum samples, 1 μL of TiVCrMoC3 solution was dropped onto the target plate, dried in a vacuum pump, and 1 μL of analyte solution was dropped onto the TiVCrMoC3 precipitate and dried again. The dried target plate was then detected by SALDI MS. SALDI MS peak maps were generated using FlexAnalysis software (Bruker Daltonics), as shown in Figure 2. Figure 7 shown.

[0035] Example 3 Perform machine learning on the extracted metabolic fingerprints to build disease diagnosis models and biomarker identification: (1) The raw data obtained by SALDI MS were imported into FlexAnalysis Version 4.0 (Bruker Daltonik GmbH) for peak extraction, baseline calibration, smoothing, peak alignment, and normalization. Unsupervised principal component analysis (PCA) was first used to convert the mass spectrometry data (m / z signals) into linearly uncorrelated variables to reduce the dimensionality of the data and identify natural clusters. All machine learning algorithms were written in the R environment. Gradient boosting and component linear model (glmBoost), random forest (RF), least absolute shrinkage and selection operator (LASSO), NaiveBayes, and XGBoost algorithms were used to model the data, and the average performance of the models was evaluated by precision, recall, F1 score, accuracy, and AUC.

[0036] (2) The data generated by machine learning were screened using the double difference method, t-test, and orthogonal partial least squares discriminant analysis (OPLS-DA). Statistical parameters such as Foldchange (FC) ≥ 2 or ≤ 0.5, pvalue < 0.05, and Variable importantin projection (VIP) ≥ 1 were considered. The top 100 m / z signals with significant differences in intensity between OA patients and RA patients were identified, and metabolites represented by upregulated and downregulated m / z signals were expressed as RA-specific serum metabolites, and a heat map was drawn, as shown in Figure 2. Figure 8 .

[0037] (3) The differential metabolites were annotated based on the Human Metabolite Database and MetFrag. Among them, 33 of the 100 metabolites were identified. Among these 33 differential metabolites, 18 metabolites were elevated and 15 metabolites were decreased in the serum of RA patients. The elevated metabolites were tryptophan (m / z 204.23), arginine (m / z 174.2), alanine (m / z 89.1), citrulline (m / z 175.19), histidine (m / z 155), glutamine (m / z 146.15), glycine-aspartic acid (m / z 190.15), glycine-alanine (m / z 146.14), citrulline, adenine (m / z 135.13), guanine (m / z 151.13), and hypoxanthine (m / z 136.11), uric acid (m / z 168.11), creatinine (m / z 113), lactic acid (m / z 90.08), pyruvate (m / z 88.06), taurine (m / z 125.15), and itaconic acid (m / z 137.11). The decreased metabolites were serine (m / z 105.1), leucine (m / z 131.2), asparagine (m / z 132.1), linolenic acid (m / z 278.4), dodecanoic acid (m / z 172.2), palmitic acid (m / z 256.4), glucose (m / z 180.1), sucrose (m / z 342.3), vitamin C (m / z 176.1), reduced glutathione (m / z 307.3), succinate (m / z 118.1), inositol (m / z 180.1), uracil (m / z 112.1), L-carnitine (m / z 161.2), and phosphatidylserine 40:4 (m / z 839.2).

[0038] The present invention is based on the high-entropy material TiVCrMoC3 and uses laser desorption ionization mass spectrometry to extract serum metabolic fingerprint spectra of rheumatoid arthritis patients. This method then combines machine learning to analyze the serum metabolic fingerprint spectra of rheumatoid arthritis patients, screen and identify differential metabolites, and achieve early, rapid, and accurate diagnosis of rheumatoid arthritis patients through combined diagnosis of differential metabolites. This high-entropy material has good application prospects as a matrix for serum LDI MS analysis. This method significantly improves the high-throughput, high-sensitivity, high-signal intensity, and high-repeatability detection performance of serum metabolic fingerprint spectra, effectively solving the problems of large sample requirements and complex pretreatment processes in traditional technologies.

[0039] The above embodiments are only preferred embodiments of the present invention and are not intended to limit the present invention in any form or substance. It should be pointed out that any technical solutions that can be obtained by technicians in this technical field through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of existing technologies should be within the scope of protection determined by the claims.

Claims

1. A method for applying a high entropy material-assisted LDI MS combined with machine learning to a rheumatoid arthritis diagnostic system, characterized in that: The following steps are involved: Step 1: prepare high entropy material TiVCrMoC3 and configure it into a suspension; Step 2: Using TiVCrMoC3 as a matrix-assisted LDI MS, the metabolic fingerprint of RA patient serum was extracted; Step 3: Perform machine learning on the extracted metabolic fingerprint to build a disease diagnosis model and biomarker identification.

2. The method according to claim 1, characterized in that Step 1 is as follows: Step 1.1, TiVCrMoAlC3 MAX powder is etched with hydrofluoric acid and centrifuged, and washed several times until the supernatant is neutral to obtain multilayer TiVCrMoC3; Step 1.2: Add TMAOH solution to the multilayer TiVCrMoC3 obtained in step 1.1 and stir at room temperature to allow sufficient intercalation; Step 1.3: Disperse the precipitate obtained in step 1.2 in deionized water, sonicate in an ice-water bath, centrifuge, collect the supernatant, concentrate, and freeze-dry to obtain a single-layer solid TiVCrMoC3.

3. The method according to claim 2, characterized in that In step 1.1, the TiVCrMoAlC3 MAX powder is 2 g, the hydrofluoric acid concentration is 48 wt %, and the amount used is 20 ml. The etching conditions are stirring at 55°C for 4 days, and the centrifugation conditions after etching are 4200 rpm for 5 minutes.

4. The method according to claim 2, characterized in that The concentration of TMAOH solution in step 1.2 is 5 wt %, the amount used is 20 ml, and stirring is carried out at room temperature for 6 hours.

5. The method according to claim 2, characterized in that In step 1.3, the TiVCrMoC3 monolayer solid was prepared into a 1 mg / mL solution using deionized water. After ultrasonication, the centrifugation conditions were 4200 rpm for 5 min.

6. The method according to claim 1, wherein Step 2: Step 2.1, collecting blood samples from RA patients and healthy volunteers as analytical samples; Step 2.2: Centrifuge the blood sample to obtain the supernatant (serum), dilute it 50-fold with methanol, and centrifuge to obtain the supernatant; Step 2.3: Take 1 μL of the serum sample diluted in step 2.2, spot it on the LDI MS target plate, and dry it at room temperature; Step 2.4: Prepare a solution of TiVCrMoC3, spot 1 μL of the solution on an LDI MS target plate, and dry at room temperature. Step 2.5: Extraction of fingerprint spectrum: Extraction of metabolic fingerprint spectrum of the sample is performed in MALDI MS.

7. The method according to claim 6, characterized in that The laser desorption ionization mass spectrometry analysis in step 2.5 was performed using a Rapiflex MALDI Tissuetyper™ TOF / TOF mass spectrometer. All samples were analyzed in positive ion mode using a nitrogen laser equipped with a 355 nm Smartbeam II laser. The laser operated at a repetition rate of 5000 Hz, and the data acquisition range covered m / z 80-1000. SALDI MS peak maps were generated using FlexAnalysis software.

8. The method according to claim 1, characterized in that The step 3 specifically includes: 3.

1. Perform peak extraction, baseline calibration, smoothing, peak alignment and normalization on the metabolic fingerprint obtained in step 2.5; 3.

2. Gradient boosting and component linear models, random forest, least absolute shrinkage and selection operator, NaiveBayes, and XGBoost algorithms were used to model the data, and the average performance of the models was evaluated by precision, recall, F1 score, accuracy, and AUC. 3.

3. The model with satisfactory evaluation results is applied to the analysis of serum metabolic fingerprint.

9. The method according to claim 7, characterized in that The conditions for screening serum metabolic markers in step 3 were: Foldchange (FC) ≥ 2 or ≤ 0.5, pvalue < 0.05, and Variable important in projection (VIP) ≥ 1, which were used as serum metabolic markers.

10. An application of the high entropy material TiVCrMoC3 according to any one of claims 1 to 5, characterized in that: The application is selected from any one of the following 1)-3): 1) Application in the analysis and detection of metabolites in biological samples for non-diagnostic and therapeutic purposes; 2) Application in the preparation of reagents or kits for analyzing and detecting metabolites in biological samples; 3) Application in screening metabolic markers of metabolic diseases.

Images