Method for rapidly screening fentanyl compounds based on machine learning model

Abstract

The invention relates to a method for rapidly screening fentanyl compounds based on a machine learning model. The method comprises the following steps: 1) establishing a fentanyl compound MS / MS mass spectrum database; 2) constructing a fentanyl compound machine learning model; (3) extracting the fentanyl compounds in the sample to be detected, and collecting MS / MS mass spectrum data, and (4) rapidly screening and identifying the fentanyl compounds. According to the machine learning model disclosed by the invention, various fentanyl compounds (including known and unknown novel substitutes, metabolites and the like) can be quickly identified, screened and detected, and an effective technical support is provided for discovery and supervision of the fentanyl compounds.

Description

Technical Field

[0001] The present invention belongs to the field of analysis and detection, and specifically relates to a method for rapidly screening fentanyl compounds based on a machine learning model. Background Art

[0002] Fentanyl is a synthetic opioid receptor agonist that can selectively stimulate μ receptors to produce analgesic effects, with an analgesic effect 50 to 100 times that of morphine. However, due to its similar effects to heroin, fentanyl derivatives are constantly being "designed" from underground laboratories and have become a typical representative of contraband. my country defines fentanyl compounds as substances whose chemical structures meet one or more conditions compared to fentanyl: (1) the propionyl group is replaced by other acyl groups; (2) any substituted or unsubstituted monocyclic aromatic group is used to replace the phenyl group directly connected to the nitrogen atom; (3) the piperidine ring is replaced by alkyl, alkoxy, alkenyl, hydroxyl, ester, ether, halogen, alkyl halide, amino and nitro groups; (4) any other group (except hydrogen atoms) is used to replace the phenylethyl group. Due to its highly derived structural characteristics, even if existing drugs are listed as controlled drugs, criminals can still modify their skeletons and continuously design and synthesize new fentanyl compounds that are not on the control list. This has led to an increasing number of deaths from illegal use of fentanyl compounds, causing great harm to society.

[0003] my country has always attached great importance to the control of fentanyl-like substances. In the 1996 edition of the Catalogue of Narcotic Drugs, 12 fentanyl-like substances were included in the national control catalogue. Faced with the continuous update of fentanyl-like substances, my country has actively taken measures to control them. In 2017, my country included four fentanyl-like substances, including carfentanyl, furanyl fentanyl, acryl fentanyl and valeryl fentanyl, in the Supplementary Catalogue of Controlled Non-Medicinal Narcotic Drugs and Psychotropic Drugs. By the end of 2018, my country had listed 25 fentanyl-like substances and two fentanyl precursors. On May 1, 2019, my country included fentanyl-like substances in the Supplementary Catalogue of Controlled Non-Medicinal Narcotic Drugs and Psychotropic Drugs, which means that it will no longer control a single fentanyl-like substance, but will control all substances with similar chemical structures and effects to fentanyl, so as to effectively deal with the problem of rapid mutation and difficulty in combating fentanyl-like drugs.

[0004] For the detection of fentanyl compounds, targeted detection methods were used in the past. However, this detection method is mainly aimed at qualitative and quantitative detection of known fentanyl compounds, and it is still impossible to perform qualitative or quantitative detection on the emerging new fentanyl compounds. Therefore, the detection and supervision of these fentanyls have always been in a relatively backward state. In order to identify a wider range of fentanyl compounds and discover new fentanyl compounds that have not been reported before, researchers have gradually adopted non-targeted data collection methods, such as mass spectrometry detection technology, which has become the preferred non-targeted analysis and detection method for fentanyl compounds due to its fast analysis speed, high detection sensitivity and wide range of applications. However, many fentanyl compounds are not suitable for detection by gas chromatography-mass spectrometry, and the analysis effect is not good for thermally unstable or non-volatile compounds, so the detection effect and usage scenarios are limited. In addition, the data collected by non-targeted is massive and complex. How to find the required information in the complex data and how to parse the information are also the difficulties of non-targeted analysis at present.

[0005] In recent years, with the substantial increase in computer computing power, the field of machine learning has developed rapidly. Today, the application of machine learning models is no longer limited to classic problems such as computer vision, pattern recognition, and image segmentation, but has been widely infiltrated into data analysis in all walks of life. Compared with traditional data analysis methods, the biggest advantage of machine learning models is that they have the ability to automatically extract specific trends and features from massive data, fit the most appropriate function to the experimental data trend, and thus accurately predict new sample data. There are few studies on the use of machine learning methods for non-targeted rapid screening and identification of fentanyl compounds. Therefore, developing a method for detecting fentanyl compounds based on machine learning algorithms that can efficiently, accurately, and comprehensively detect all known and / or unknown fentanyl compounds, derivative compounds, and metabolites is a technical problem that needs to be solved in this field. Summary of the invention

[0006] In view of the problems existing in the above-mentioned prior art, the purpose of the present invention is to provide a method for rapidly detecting fentanyl compounds based on a machine learning model, thereby providing a basis and support for the supervision of fentanyl compounds. The present invention uses liquid chromatography quadrupole time-of-flight tandem mass spectrometry (LC-QTOF MS) to establish an MS / MS mass spectrogram database of fentanyl compounds, uses a machine learning model to study the database, and constructs a Binning-RF two-classification model of fentanyl compounds; taking fentanyl compounds in urine as the main research object, the urine is pre-treated by auxiliary automatic magnetic solid phase extraction technology, and combined with the machine learning model of fentanyl compounds, the purpose of rapidly screening various known and unknown fentanyl compounds is achieved. The present invention can be beneficial to strengthening the supervision of illegal fentanyl compounds and improving the ability to prevent drug abuse.

[0007] To achieve the above object, the present invention provides a method for rapidly screening fentanyl compounds based on a machine learning model, comprising the following steps:

[0008] Step 1) Establishing a MS / MS mass spectrometry database of fentanyl compounds

[0009] The secondary mass spectrometry data of fentanyl compounds were collected and preprocessed to obtain the MS / MS mass spectrometry database of fentanyl compounds. The data sources of the MS / MS database of fentanyl compounds include two parts: one is the secondary mass spectrometry data collected in the laboratory, and the other is the valid secondary mass spectrometry data of fentanyl compounds in the public database on the Internet.

[0010] The secondary mass spectrometry data collected by the laboratory: using a liquid chromatography quadrupole time-of-flight mass spectrometer (LC-QTOFMS mass spectrometer) to collect MS / MS mass spectrometry data of known standard fentanyl compounds.

[0011] The chromatographic conditions are as follows: mobile phase A is a combination of formic acid and an aqueous solution of ammonium formate, the volume ratio of the formic acid to the aqueous solution of ammonium formate is (1-5):1000, and the concentration of ammonium formate in the aqueous solution of ammonium formate is 10-15 mM; mobile phase B is a combination of acetonitrile and mobile phase A, wherein the volume percentage of acetonitrile is 90-95%; flow rate: 0.3-0.5 mL / min; gradient elution is adopted, and the injection volume is controlled to be 2-5 μL.

[0012] In one embodiment of the present invention, mobile phase A: 0.1% formic acid + 99.9 (v / v)% 10 mM ammonium formate aqueous solution, mobile phase B: a mixed solution of 95 (v / v)% acetonitrile + 5% mobile phase A (0.1% formic acid + 99.9 (v / v)% 10 mM ammonium formate aqueous solution).

[0013] In one embodiment of the present invention, the elution gradient is: 5% B (0 min), 5% B (0.5 min), 15% A (1.5 min), 30% A (4 min), 30% A (7.5 min), 90% A (8 min), 90% A (10 min), 5% A (10.1 min), 5% A (12 min).

[0014] The mass spectrometry conditions were as follows: ion source ESI+, curtain gas 30-35psi, ionization voltage 5000-5500kV, temperature 500-600°C, spray gas 50-55psi, auxiliary heating gas 50-55psi, declustering voltage 100-120V, cone voltage 60-65V, mass spectrometry acquisition mode was MS-IDA mode, primary mass spectrometry acquisition range was 100-1000m / z, and secondary mass spectrometry acquisition range was 50-1000m / z.

[0015] In one embodiment of the present invention, the ion source is ESI+, the curtain gas is 35psi, the ionization voltage is 5500kV, the temperature is 600°C, the spray gas is 55psi, the auxiliary heating gas is 55psi, the declustering voltage is 100V, and the cone voltage is 65V.

[0016] The secondary mass spectrometry data of the network public database include the MS / MS mass spectrometry data of known fentanyl compounds in the National Institute of Standards and Technology (NIST), MassBank of North America (MoNA) and HighResNPS network databases.

[0017] The chemical formula, InChIKey, precursor ion, collision energy, molecular weight and other information of the collected fentanyl compounds were entered into the self-built library. After data preprocessing steps such as normalization, data amplification, and data cleaning, they were used as positive data sets, and the construction of the MS / MS mass spectrometry database of fentanyl compounds was completed for machine learning.

[0018] The specific processing steps include:

[0019] 1-1) Data extraction: Use MS-DIAL 4.60 software to perform peak extraction on the wiff format file of the collected data. The parameters are set as follows: MS1 tolerance 0.01Da, MS2 tolerance 0.025Da, Minimum peak height 1000 amplitude, Mass slice width 0.1Da, Sigma window value 0.5, MS / MSabundance cut off 0 amplitude, Adduct ion included[M+H]+. After processing, the spectrum information is exported.

[0020] 1-2) Normalization processing: The absolute response intensity is converted into a relative response intensity, and normalized to 1000 (maximum relative response value). The normalization formula is as shown in (I), and smoothing or peak searching is performed as needed.

[0021] Relative response value = (absolute response value / maximum absolute response value) × maximum relative response value (I)

[0022] The public databases on the Internet have different acquisition instruments and parameter settings, so their formats need to be unified. First, the absolute response intensity is converted into relative response intensity according to the method described in the previous self-built library, and then the mass spectrometry data set is uniformly represented as L, L = {L1 / L2 / L3… / L N}, where each variable represents a mass spectrum data, N is the total number of sample data, and each Lx is composed of the fragment mass number m x and the relative response intensity i x Composition, that is, Lx=(m x i x ), where m x ={m1, m2, m3…, m n},i x ={i1,i2,i3…,i n}.

[0023] 1-3) Data augmentation: Data augmentation is used to reduce the problem of uneven positive and negative data sets and the resulting performance degradation of machine learning models.

[0024] For each original spectrum, the spectrum data with the maximum response intensity above x% (x=50, 60, 70, 80, 90, 100) were retained for data amplification to match the number of subsequent negative data sets.

[0025] 1-4) Data cleaning: Spectra with fragment numbers less than 5 after data amplification were eliminated to avoid the influence of weak spectral characteristics on the final accuracy of the model, and the establishment of the MS / MS mass spectrometry database of fentanyl compounds was completed.

[0026] Furthermore, blank urine, solvent blank, and high-quality spectra that were not classified as fentanyl in the Exposome-Explorer public database were collected to form a negative data set; both the positive data set and the negative data set were used as training sets of the mass spectrometry database to train the machine learning model in step 3).

[0027] The solvents are all solvents used in the experimental process of the present invention, such as methanol and acetonitrile.

[0028] Step 2) Build a machine learning model for fentanyl compounds

[0029] Fentanyl compounds were classified by feature extraction and using the Random Forest (RF) classification model.

[0030] 2-1) Feature extraction: Use Top N method, Binning method or Grid method to extract and optimize the features of MS / MS spectra.

[0031] Preferably, the Binning method is used to extract features and optimize parameters of the MS / MS spectrum.

[0032] The parameter optimization includes optimization of intensity threshold, m / z range of fentanyl compounds and bin width.

[0033] 2-2) Model construction: Build a random forest RF classification model for learning classification.

[0034] Based on the CART (Classification and Regression Trees) algorithm, the present invention independently develops a random forest RF classification model. By constructing and evaluating a decision tree, the selection of split points and pruning strategies are determined, thereby optimizing the structure and performance of the decision tree. The Gini index is used to select the optimal feature and determine the optimal binary split point of the feature. The Gini index calculation formula is as described in (II):

[0035]

[0036] Where Pi is the proportion of samples of the i-th category in the dataset, and k is the total number of categories.

[0037] The prediction of the RF model is the voting result of all decision tree prediction results. Assuming we have N decision trees, the probability of predicting category C can be expressed as formula (III):

[0038]

[0039] In the formula, I(C i =C) is an indicator function, indicating whether the prediction of the i-th decision tree is category C.

[0040] Furthermore, the Binning-RF model was evaluated using 5*3 nested cross validation to optimize the model hyperparameters, where the outer loop was used 5 times for algorithm evaluation and the inner loop was used 3 times for Bayesian optimization hyperparameter tuning.

[0041] Bayesian optimization is to update the posterior distribution (Gaussian process) of the objective function by continuously adding sample points until the posterior distribution basically fits the true distribution given the objective function to be optimized (a generalized function that only needs to specify the input and output without knowing the internal structure and mathematical properties).

[0042] In one embodiment of the present invention, the objective function is a performance index of the classifier, such as accuracy, precision, recall, F1 score, Matthews correlation coefficient (MCC), ROC-AUC, etc. These indicators, as output values ​​of the objective function, can reflect the performance of the model on the test data.

[0043] The Bayesian optimization hyperparameters include optimizing the hyperparameters of n_estimators, max_features, max_depth, min_samples_split, min_samples_leaf, and bootstrap of the random forest RF model.

[0044] 2-3) Model evaluation: Accuracy, precision, recall, F1 score, and MCC score (Matthews correlation coefficient) are used as evaluation indicators of the model.

[0045] Furthermore, since the training data set of the present invention is unbalanced data (i.e., negative data is greater than positive data), we calculate the Matthews correlation coefficient (MCC) score to examine the prediction accuracy of the model, and the calculation formula is as follows (IV):

[0046]

[0047] In the formula, TP is a true positive (TP): a positive sample predicted as positive by the model; FP is a false positive (FP): a negative sample predicted as positive by the model; FN is a false negative (FN): a positive sample predicted as negative by the model; TN is a true negative (TN): a negative sample predicted as negative by the model.

[0048] The present invention also compares the RF model with other machine learning algorithms such as LR, SVM, KNN, and the experimental results show that the RF classification model is significantly better than the support vector machine, K nearest neighbor, and logistic regression results. Therefore, it can be determined that the Binning-RF classification model of the present invention is effective and excellent.

[0049] In one embodiment of the present invention, the evaluation results of the Binning-RF classification model are: accuracy is 0.99±0.01, recall is 0.98±0.02, precision is 0.99±0.01, F1 is 0.99±0.01, and MCC is 0.98±0.01.

[0050] Step 3) Extracting fentanyl compounds in the sample to be tested and collecting MS / MS mass spectrometry data

[0051] 3-1) Extraction of fentanyl compounds:

[0052] The fentanyl compounds in the sample to be tested were extracted by automatic magnetic extraction method: pure methanol and 5v% methanol were used to activate HLB magnetic solid phase extraction material in sequence, and the material was treated with eluent and eluting solution.

[0053] In the activation step, 2-3 mg of HLB magnetic solid phase extraction material was added to every 400 uL of sample;

[0054] The elution process is to use two eluents for elution treatment respectively.

[0055] The composition of eluent 1 is: 5v% methanol aqueous solution, and the composition of eluent 2 is: 10%-20v% methanol aqueous solution;

[0056] The composition of the eluent is: 95v% methanol + 0.5v%-1v% formic acid in water;

[0057] The mixing time for each step is 30-60 s, and the magnetic adsorption time is 30-60 s.

[0058] After automatic magnetic extraction, the sample was reconstituted with the initial mobile phase, centrifuged, and the supernatant was taken, which was the sample to be tested.

[0059] In one embodiment of the present invention, nitrogen drying is performed at 40° C. and centrifugation is performed at 12,000 rpm for 10 min.

[0060] In the prior art, direct dilution method or MCX solid phase extraction method is commonly used for the extraction of fentanyl compounds. However, the above methods are only applicable to the extraction of known or target fentanyl compounds. In the case of numerous fentanyl derivatives or uncertain modification of the fentanyl molecular skeleton, the above conventional extraction methods cannot fully obtain the fragment fingerprint peaks of all fentanyl skeleton compounds, and the MS / MS spectrum is not accurate and comprehensive enough.

[0061] In the above conventional extraction method, in view of the certainty of the target compound, the mass spectrum peaks of some unknown fentanyl compounds are easily destroyed or "sacrificed", and effective extraction cannot be achieved. Therefore, the above conventional extraction method cannot identify all fentanyl compounds in the sample to be tested, and cannot solve the regulatory issues of various unknown fentanyl compounds. Even if the learning model of the present invention is used, it will not be able to quickly and effectively identify them (see Comparative Examples 4-5).

[0062] The present invention adopts the HLB magnetic solid phase extraction method. Through continuous optimization of the method, it is possible to achieve effective extraction of all fentanyl compounds in the sample to be tested, providing strong technical support for the identification and supervision of various fentanyl compounds. Moreover, on the basis of effective extraction and with the help of the machine learning model of the present invention, the specific structure and type of fentanyl derivatives can be quickly and accurately identified.

[0063] 3-2) Collecting MS / MS mass spectrum data of the fentanyl compounds to be tested:

[0064] A liquid chromatography quadrupole time-of-flight mass spectrometer (LC-QTOF MS mass spectrometer) was used to collect MS / MS mass spectrum data of fentanyl compounds.

[0065] Because fentanyl compounds cannot be exhaustively enumerated, their MS / MS mass spectrometry information directly affects the accuracy of the machine learning model. Therefore, the selection of chromatographic conditions and mobile phases is crucial. The present invention innovatively uses a specific combination of mobile phases A and B to obtain more comprehensive and accurate MS / MS mass spectrometry information, which is crucial to the accuracy and effectiveness of subsequent machine learning.

[0066] The chromatographic conditions are as follows: mobile phase A is a combination of formic acid and an aqueous solution of ammonium formate, the volume ratio of the formic acid to the aqueous solution of ammonium formate is (1-5):1000, and the concentration of ammonium formate in the aqueous solution of ammonium formate is 10-15 mM; mobile phase B is a combination of acetonitrile and mobile phase A, wherein the volume percentage of acetonitrile is 90-95%; flow rate: 0.3-0.5 mL / min; gradient elution is adopted, and the injection volume is controlled to be 2-5 μL.

[0067] The mass spectrometry conditions were as follows: ion source ESI+, curtain gas 30-35psi, ionization voltage 5000-5500kV, temperature 500-600°C, spray gas 50-55psi, auxiliary heating gas 50-55psi, declustering voltage 100-120V, cone voltage 60-65V, mass spectrometry acquisition mode was MS-IDA mode, primary mass spectrometry acquisition range was 100-1000m / z, and secondary mass spectrometry acquisition range was 50-1000m / z.

[0068] In one embodiment of the present invention, the chromatographic conditions are: mobile phase A: 0.1v% formic acid + 99.9v% 10mM ammonium formate aqueous solution, mobile phase B: a mixed solution of 95v% acetonitrile + 5v% mobile phase A (0.1v% formic acid + 99.9v% 10mM ammonium formate aqueous solution); flow rate: 0.3mL / min; elution gradient: 5%B (0min), 5%B (0.5min), 15%A (1.5min), 30%A (4min), 30%A (7.5min), 90%A (8min), 90%A (10min), 5%A (10.1min), 5%A (12min); injection volume: 5μL. The mass spectrometry conditions were as follows: ion source ESI+, curtain gas 35 psi, ionization voltage 5500 kV, temperature 600°C, spray gas 55 psi, auxiliary heating gas 55 psi, declustering voltage 100 V, cone voltage 65 V, collision energy 30 V, mass spectrometry acquisition mode was MS-IDA mode, the primary mass spectrometry acquisition range was 100-1000 m / z, and the secondary mass spectrometry acquisition range was 50-1000 m / z.

[0069] Furthermore, the chromatographic conditions and mass spectrometry conditions are the same as those in step 1), and comprehensive mass spectrometry data of fentanyl compounds (including known and to be tested) can be obtained.

[0070] Step 4) Rapid screening and identification of fentanyl compounds

[0071] The MS / MS mass spectrometry data of fentanyl compounds collected in step 3) are rapidly screened and classified using the machine learning model of step 2).

[0072] In one embodiment of the present invention, the structure of butyrylfentanyl in urine matrix can be quickly verified using the learning model of the present invention, indicating that the accuracy of the model is high.

[0073] In another embodiment of the present invention, fentanyl, an anesthetic, is incubated in vitro, and its MS / MS mass spectrometry data is collected after automatic magnetic extraction and processing, and its in vitro incubation metabolites are identified and classified using a machine learning model.

[0074] In a second aspect, the present invention provides a method for constructing a database of fentanyl compounds based on a machine learning model, comprising the following steps:

[0075] Step 1) establishing a fentanyl compound MS / MS database;

[0076] Step 2) Build a machine learning model for fentanyl compounds and their metabolites.

[0077] The specific methods of steps 1) and 2) are the same as those of the first aspect.

[0078] Through the method described in the present invention, a total of 772 MS / MS spectra of 277 fentanyl compounds, including 96 fentanyl compounds extracted in the laboratory, were collected and sorted, and a secondary mass spectrum database of fentanyl compounds was established using liquid chromatography-time of flight tandem mass spectrometry (LC-QTOFMS); fentanyl compounds and their metabolites were effectively extracted and purified by auxiliary rapid automatic magnetic extraction pretreatment technology, and fentanyl compounds were spectrally characterized by machine learning models, thereby achieving the purpose of rapid screening of unknown fentanyl compounds.

[0079] Beneficial effects of the present invention:

[0080] 1. The present invention comprehensively collects and organizes 772 MS / MS spectra of 277 fentanyl compounds for the first time, and establishes a secondary mass spectrum database of fentanyl compounds using liquid chromatography-time of flight tandem mass spectrometry (LC-QTOF MS), laying a comprehensive and powerful foundation for subsequent data identification and classification.

[0081] 2. The present invention uses a machine learning model to learn the spectral features of fentanyl compounds, extracts features from MS / MS, and establishes a Binning-RF optimal classification model, thereby achieving the purpose of rapid screening of unknown fentanyl compounds.

[0082] 3. The present invention constructs a machine learning model based on the mass spectrometry data characteristics of fentanyl compounds, and improves and customizes the RF model; the model is highly compatible with fentanyl compounds and has a high machine learning effect, thereby improving the performance in specific application scenarios.

[0083] 4. The present invention adopts rapid automatic magnetic extraction pretreatment technology to effectively extract and purify the fentanyl compounds to be tested. Through the optimization of effective pretreatment methods and liquid chromatography conditions, the solubility, separation effect and extraction efficiency of each fentanyl compound are improved. In the LC-QTOF MS detection process, the secondary mass spectra (MS / MS) of each fentanyl compound obtained are of higher quality, more comprehensive compound types and better peak shapes, thereby making the machine learning model training effect better and the prediction accuracy higher, thereby realizing efficient, accurate and comprehensive identification and screening of fentanyl compounds.

[0084] 5. Using the machine learning model of the present invention, in addition to achieving efficient and accurate rapid screening of fentanyl compounds in common urine matrices, fentanyl metabolites can also be accurately tested, reflecting the accuracy and universality of the machine learning model of the present invention. It can be expanded to the detection field of any detection matrix, any fentanyl compound and its metabolites, providing a guarantee for the control of fentanyl compounds. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 This is a flow chart of the detection method of Example 1 of the present invention.

[0086] Figure 2 This is the basic information of the 24 fentanyl compounds in Example 1.

[0087] Figure 3 The number of cut fragments is selected for Example 1.

[0088] Figure 4 This is the threshold selection of Example 1.

[0089] Figure 5 The mass range and bin number selection of Example 1.

[0090] Figure 6 This is an explanation of the SHAP model of Example 1.

[0091] Figure 7 The results of different feature extraction methods in Examples 1-3 are compared.

[0092] Figure 8 This is the extracted ion chromatogram (EIC) of butyrylfentanyl in urine samples in Example 4.

[0093] Fig. 9 This is the recognition result of butyrylfentanyl at different concentrations (1 ng / mL, 5 ng / mL) in urine using the machine learning model in Example 4.

[0094] Fig.10 This is the identification result of known metabolites of fentanyl using the machine learning model in Example 5.

[0095] Fig.11 This is the verification of the new metabolite of fentanyl in Example 5; wherein a is the EIC graph of the new metabolite standard; b is the EIC graph of the fentanyl metabolite sample; c is the MS / MS spectrum of the new metabolite standard; d is the MS / MS spectrum of the fentanyl metabolite sample.

[0096] Fig.12 This is a diagram showing the extraction effect of fentanyl compounds using MCX solid phase extraction material in Example 7-A; the left figure is the MS / MS spectrum of fentanyl compounds extracted at a low concentration point (0.5 ng / mL), and the right figure is the MS / MS spectrum of fentanyl compounds extracted at a medium and high concentration point (10 ng / mL).

[0097] Fig.13 This is a diagram showing the extraction effect of fentanyl compounds using HLB magnetic solid phase extraction material in Example 7-B.

[0098] Fig.14 The figures are diagrams of the extraction effects of Example 8 using different mobile phases, wherein a is the MS / MS spectrum of Example 8-A; b is the MS / MS spectrum of Example 8-B; and c is the MS / MS spectrum of Example 8-C.

[0099] Fig.15 It is a comparison chart of the random forest RF classification model of Example 1 and other models of Comparative Examples 1-3. DETAILED DESCRIPTION

[0100] In order to make the objectives, technical solutions and advantages of the present invention more clear, the specific implementation methods of the present invention are further described in detail below in conjunction with specific embodiments and drawings.

[0101] The instrument used in the present invention is: ExionLC AD high performance liquid chromatograph-7600 quadrupole time-of-flight mass spectrometer, AB SCIEX Company, USA.

[0102] The chemical reagents used in the experiments are all commercially available.

[0103] The source of urine matrix was blank urine samples from healthy volunteers.

[0104] Example 1

[0105] A method for rapid detection of fentanyl compounds based on a machine learning model. In this embodiment, 24 fentanyl compounds are selected, an MS / MS mass spectrometry database is established, and a machine learning model for fentanyl compounds is constructed. The entire process is as follows Figure 1 shown.

[0106] 1) Establishment of MS / MS database of fentanyl compounds

[0107] Laboratory-collected MS data: Figure 2 The names, chemical formula InChIKey, precursor ions, collision energy, MS / MS, adduct forms and retention times of 24 fentanyl compounds were entered into the self-built library as standards. The collected self-built library compound standards were then prepared into standard stock solutions and working solutions, respectively. The chromatographic conditions and mass spectrometry conditions were adjusted, and the secondary spectrum data of the working solutions of the standards were collected in IDA mode.

[0108] Liquid chromatography and mass spectrometry conditions:

[0109] Chromatographic conditions were as follows: mobile phase A: 0.1 v% formic acid + 99.9 v% 10 mM ammonium formate aqueous solution, mobile phase B: a mixed solution of 95 v% acetonitrile + 5 v% mobile phase A (0.1 v% formic acid + 99.9 v% 10 mM ammonium formate aqueous solution); flow rate: 0.3 mL / min; elution gradient: 5% B (0 min), 5% B (0.5 min), 15% A (1.5 min), 30% A (4 min), 30% A (7.5 min), 90% A (8 min), 90% A (10 min), 5% A (10.1 min), 5% A (12 min); injection volume: 5 μL.

[0110] The mass spectrometry conditions were as follows: ion source ESI+, curtain gas 35 psi, ionization voltage 5500 kV, temperature 600 °C, spray gas 55 psi, auxiliary heating gas 55 psi, declustering voltage 100 V, cone voltage 65 V, and collision energy of 10 V, 20 V, 30 V, and 40 V were used for fragmentation. The mass spectrometry acquisition mode was MS-IDA mode, the primary mass spectrometry acquisition range was 100-1000 m / z, and the secondary mass spectrometry acquisition range was 50-1000 m / z.

[0111] The mass spectrometry data of fentanyl compounds collected in the laboratory and the mass spectrometry data information in the public database on the Internet were entered into the self-built library to form an MS / MS mass spectrum database of fentanyl compounds for machine learning.

[0112] The data information in the database is sorted as a positive data set for training, and the sorting includes data extraction, normalization, data amplification, data cleaning and other data processing steps; blank urine, solvent blank and high-quality spectra that are not classified as fentanyl in the Exposome-Explorer public database are collected to form a negative data set. Both the positive data set and the negative data set are used as training sets to train the Binning-RF model in step 3).

[0113] 1-1) Data extraction: The obtained secondary mass spectrum data were input into the self-built library. For the data collected in the laboratory, the MS-DIAL 4.60 software was used to extract the peaks of the collected wiff format files. The parameters were set as follows: MS1tolerance 0.01Da, MS2 tolerance 0.025Da, Minimum peak height 1000amplitude, Mass slice width 0.1Da, Sigma window value 0.5, MS / MS abundance cut off 0amplitude, Adduct ion included[M+H]+. After processing, the spectrum information was exported.

[0114] 1-2) Normalization: The MSMS spectrum is a two-dimensional information, including the fragment mass number and the absolute response intensity. In order to facilitate the subsequent machine learning steps, the absolute response intensity needs to be converted into a relative response intensity and normalized to 1000. The normalization formula is as follows:

[0115] Relative response value = (absolute response value / maximum absolute response value) × maximum relative response value

[0116] Among them, the maximum relative response value is 1000.

[0117] Dynamic background subtraction was selected as the data acquisition step, so the data did not require any smoothing or peak finding.

[0118] Since the acquisition instruments and parameter settings of the network database are different, it is necessary to unify its format. First, the absolute response intensity is converted into relative response intensity according to the method described in the previous self-built library, and then the mass spectrometry data set is uniformly represented as L, L = {L1 / L2 / L3… / L N}, where each variable represents a mass spectrum data, N is the total number of sample data, and each Lx is composed of the fragment mass number m x and the relative response intensity i x Composition, that is, Lx=(m x i x ), where m x ={m1, m2, m3…, m n},ix={i1,i2,i3…,i n}.

[0119] 1-3) There are 772 spectra of 277 fentanyl compounds in the positive dataset (96 of which were collected in the laboratory and the rest were from the online database), but there are 4164 spectra in the negative dataset, which is much larger than the number of positive datasets. This imbalance will cause the model to be biased towards the majority class during training, thereby ignoring the characteristics of the minority class, resulting in a decrease in model performance. Therefore, in order to solve this data imbalance problem, we use data augmentation to increase the size of the fentanyl spectral pool.

[0120] For each original spectrum, data above x% (x=50, 60, 70, 80, 90, 100) of its maximum response intensity were retained, thus obtaining 3860 positive spectra.

[0121] 1-4) Data cleaning: For positive and negative data sets, if the number of fragments in the spectrum is too small, the spectrum will not be characteristic enough, affecting the final accuracy of the model, such as Figure 3Therefore, the number of fragments was checked, and spectra with fragment numbers less than 5 were removed; thus, the MS / MS mass spectrum database of fentanyl compounds was established.

[0122] 2) Construction of machine learning model for fentanyl compounds

[0123] Use the self-developed Random Forest (RF) classification model to learn and classify the positive and negative data sets in the fentanyl compound database:

[0124] 2-1) Feature extraction: Use the Binning method to extract features from MS / MS mass spectrometry data, and optimize the various parameters of the Binning method. Because mass spectrometry data contains a large amount of background noise and low-intensity signals, noise may interfere with real and effective signals. Therefore, it is necessary to set an intensity threshold to filter out noise below the threshold and retain more meaningful high-intensity signals, thereby improving the quality of the data. At the same time, it can also reduce the amount of data that needs to be processed and reduce the complexity and time of calculations, which is very important in machine learning training. Finally, because our data involves different experimental conditions and different instrumentation, this may lead to large differences in signal intensity in mass spectrometry data. By setting a unified intensity threshold, the data can be standardized to reduce uncertain differentiated features caused by external factors such as experimental conditions.

[0125] Therefore, this embodiment examines the performance indicators of the model when the intensity threshold is 10, 20, 30, and 40, and the results are as follows: Figure 4 As shown. It shows that when the intensity threshold is 10, the performance of the model is optimal. The secondary mass spectrometry scanning range of fentanyl compounds in the self-built library is 50-1000Da, so the data range is initially determined to be 50-1000Da. However, after statistics on the m / z range of the fragment ions of fentanyl compounds, it was found that 99.75% of the fentanyl compounds have an m / z range of 50-400Da. Therefore, considering the amount of calculation, this embodiment also examines the m / z range: 50-1000 and 50-400. The following two investigations were conducted on the box width: 0.1Da and 1Da. The results for the four combinations of m / z range and box width are shown as follows: Figure 5 Finally, the results of the training set showed that the m / z range of 50-400 Da and the bin width of 0.1 Da were the optimal combination.

[0126] 2-2) Model construction: The RF classification model is used to learn and classify the MS / MS mass spectrometry data after feature extraction.

[0127] By building and evaluating the decision tree, we can determine the selection of split points and pruning strategies, thereby optimizing the structure and performance of the decision tree. The Gini index is used to select the optimal feature and determine the optimal binary split point of the feature. The Gini index calculation formula is as described in (II):

[0128]

[0129] Where Pi is the proportion of samples of the i-th category in the dataset, and k is the total number of categories.

[0130] The prediction of the RF model is the voting result of all decision tree prediction results. Assuming we have N decision trees, the probability of predicting category C can be expressed as formula (III):

[0131]

[0132] In the formula, I(C i =C) is an indicator function, indicating whether the prediction of the i-th decision tree is category C.

[0133] Furthermore, 5*3 nested cross validation was used to evaluate the performance of the model and optimize the model hyperparameters. The outer loop was used 5 times for algorithm evaluation, and the inner loop was used 3 times for Bayesian optimization hyperparameter tuning, including the optimization of n_estimators, max_features, max_depth, min_samples_split, min_samples_leaf and bootstrap hyperparameters of the random forest model.

[0134] 2-3) Model evaluation: The accuracy of the Binning-RF classification model is 0.99±0.01, the recall rate is 0.98±0.02, the precision is 0.99±0.01, the F1 is 0.99±0.01, and the MCC is 0.98±0.01; finally, SHAP (SHapleyAdditive exPlanations) is used to interpret the prediction results of the Binning-RF model. SHAP is a model interpretation method that assigns importance values ​​(Shap Values) to each feature of a specific prediction and identifies which feature is the most important, thereby promoting the understanding of the decision-making process of the machine learning model. It aims to provide transparency and explainability for the predictions of machine learning models.

[0135] The results are as follows Figure 6It shows that the two fragment bins with the highest SHAP values ​​are m / z 188.14 and m / z 105.07, and these two fragment ions also happen to correspond to the characteristic fragments of fentanyl compounds. This not only explains why the Binning-RF model has a better classification effect on fentanyl compounds, but also proves from another perspective that this model has better results in learning the fragmentation characteristics of fentanyl compounds.

[0136] Example 2

[0137] In step 2-1), the Top N method is used to replace the Binning method to extract features from the MS / MS mass spectrum data. The other steps and methods are the same as those in Example 1. Finally, the SHAP model is used to interpret the prediction results of the Top N-RF model.

[0138] Example 3

[0139] In step 2-1), Grid-1D and Grid-2D methods are used to replace the Binning method to extract features from the MS / MS mass spectrum data. The other steps and methods are the same as those in Example 1. Finally, the SHAP model is used to interpret the prediction results of the Grid-RF model.

[0140] Figure 7 The results of feature extraction of MS / MS mass spectrometry data using the Binning method, Top N method, Grid-1D and Grid-2D methods in Examples 1-3 are shown. By comparing the accuracy, precision, F1 score and MCC score, it can be seen that the Binning method has higher accuracy, precision, F1 score and MCC score, and is the optimal feature extraction method. In summary, the Binning-RF machine learning model is the best model for the identification and detection of fentanyl compounds.

[0141] Example 4 Screening and Identification of Butyrylfentanyl in Urine

[0142] The machine learning model constructed in Example 1 was used to quickly screen fentanyl compounds - butyrylfentanyl in urine, providing a guarantee for the control of fentanyl compounds.

[0143] Wherein, step 1-2) is the same as in Example 1;

[0144] 3) Extracting fentanyl compounds from the sample to be tested and collecting MS / MS mass spectrometry data:

[0145] 3-1) Extraction of fentanyl compounds

[0146] The butyrylfentanyl standard solution (100 μg / mL) was added to the urine matrix with a final concentration of 5 ng / mL as a positive urine sample. 100uL of HLB magnetic extraction material (20 mg / mL) was added to the 1st to 6th columns of the automatic magnetic extraction device, respectively. The activated magnetic extraction solution: 100uL of 5v% methanol, 400uL of urine sample; eluent 1: 400uL of 5v% methanol, eluent 2: 400uL of 20v% methanol, elution solution: 95v% methanol + 0.5v% formic acid. Each step was mixed for 60s, magnetic adsorption was performed for 30s, and the magnetic extraction process was automatically completed. After the magnetic extraction was completed, the eluate was transferred to another clean centrifuge tube, dried with nitrogen at 40°C, re-dissolved with 100uL of the initial mobile phase, centrifuged at 12000rpm for 10min, and the supernatant was sampled to obtain the extract of fentanyl compounds in the urine matrix.

[0147] 3-2) Collecting MS / MS mass spectrometry data of fentanyl compounds

[0148] A liquid chromatography quadrupole time-of-flight mass spectrometer (LC-QTOF MS mass spectrometer) was used to collect MS / MS spectral data of fentanyl compounds.

[0149] Chromatographic conditions were as follows: mobile phase A: 0.1 v% formic acid + 99.9 v% 10 mM ammonium formate aqueous solution, mobile phase B: a mixed solution of 95 v% acetonitrile + 5 v% mobile phase A (0.1 v% formic acid + 99.9 v% 10 mM ammonium formate aqueous solution); flow rate: 0.3 mL / min; elution gradient: 5% B (0 min), 5% B (0.5 min), 15% A (1.5 min), 30% A (4 min), 30% A (7.5 min), 90% A (8 min), 90% A (10 min), 5% A (10.1 min), 5% A (12 min); injection volume: 5 μL.

[0150] The mass spectrometry conditions were as follows: ion source ESI+, curtain gas 35 psi, ionization voltage 5500 kV, temperature 600 °C, spray gas 55 psi, auxiliary heating gas 55 psi, declustering voltage 100 V, cone voltage 65 V, collision energy 30 V, mass spectrometry acquisition mode was MS-IDA mode, primary mass spectrometry acquisition range 100-1000 m / z, secondary mass spectrometry acquisition range 50-1000 m / z. The extracted ion chromatogram (EIC) of butyrylfentanyl in urine samples is shown in Figure 1. Figure 8 shown.

[0151] 4) Screening and identification of fentanyl compounds

[0152] The machine learning model in step 2) is used to perform non-targeted recognition of the mixture of butyrylfentanyl standard and urine matrix. The recognition results are as follows: Fig. 9As shown in the figure, it can be seen that the machine learning model can effectively identify butyrylfentanyl at different concentrations (1ng / mL, 5ng / mL) ( Fig. 9 The Label is 1, which means it is a fentanyl compound), indicating that the model has high accuracy and precision.

[0153] Example 5 Screening and Identification of Fentanyl and Its Metabolites by In vitro Incubation

[0154] Using the machine learning model of Example 1, this example screens and detects fentanyl and its in vitro metabolites, providing a guarantee for the control of fentanyl compounds and their metabolites.

[0155] Wherein, step 1-2) is the same as in Example 1;

[0156] 3) Extract fentanyl compounds and their metabolites in the sample to be tested, and collect MS / MS mass spectrometry data:

[0157] 3-1) Extraction of fentanyl compounds and their metabolites:

[0158] Human liver microsomes (final concentration 0.5 mg protein / mL) were used to incubate fentanyl standard solution (100 μg / mL, final concentration 5 μM) in vitro. The incubation system was 200 μL, including uridine 5′-diphosphate-glucuronic acid (UDPGA) (final concentration 5 mM), nicotinamide adenine dinucleotide phosphate (NADPH) (final concentration 5 mM), and glutathione (GSH) (final concentration 10 μM). After incubation for 60 minutes, 600 μL of ice acetonitrile was added to terminate the reaction, vortexed for 1 minute, centrifuged at 16000g for 20 minutes, and the supernatant was evaporated under a nitrogen flow at 30°C, blown dry, and reconstituted with 400 uL of water to obtain the incubation sample.

[0159] Add 100uL of HLB magnetic extraction material (20mg / mL) to the 1st to 6th columns of the automatic magnetic extraction device, activate the magnetic extraction solution: 100uL of 5% methanol, incubate the sample 400uL; eluent 1: 400uL of 5v% methanol, eluent 2: 400uL of 20v% methanol, and elution solution: 95v% methanol + 0.5v% formic acid. Mix for 60s in each step, magnetic adsorption for 30s, and automatically complete the magnetic extraction process. After the magnetic extraction is completed, the eluate is transferred to another clean centrifuge tube, dried with nitrogen at 40℃, re-dissolved with 100uL of the initial mobile phase, centrifuged at 12000rpm for 10min, and the supernatant is sampled to obtain the extract of fentanyl compounds and their metabolites in the incubated sample.

[0160] 3-2) Collecting MS / MS mass spectrometry data of fentanyl compounds and their metabolites

[0161] A quadrupole time-of-flight mass spectrometer (LC-QTOF MS) was used to collect MS / MS spectral data of fentanyl compounds and their metabolites using chromatographic and mass spectrometric conditions and different collision energies in the Information Dependent Analysis (IDA) mode.

[0162] Chromatographic conditions were as follows: mobile phase A: 0.1 v% formic acid + 99.9 v% 10 mM ammonium formate aqueous solution, mobile phase B: a mixed solution of 95 v% acetonitrile + 5 v% mobile phase A (0.1 v% formic acid + 99.9 v% 10 mM ammonium formate aqueous solution); flow rate: 0.3 mL / min; elution gradient: 5% B (0 min), 5% B (0.5 min), 15% A (1.5 min), 30% A (4 min), 30% A (7.5 min), 90% A (8 min), 90% A (10 min), 5% A (10.1 min), 5% A (12 min); injection volume: 5 μL.

[0163] The mass spectrometry conditions were as follows: ion source ESI+, curtain gas 35 psi, ionization voltage 5500 kV, temperature 600°C, spray gas 55 psi, auxiliary heating gas 55 psi, declustering voltage 100 V, cone voltage 65 V, collision energy 30 V, mass spectrometry acquisition mode was MS-IDA mode, the primary mass spectrometry acquisition range was 100-1000 m / z, and the secondary mass spectrometry acquisition range was 50-1000 m / z.

[0164] 4) Application of screening and identification models for fentanyl compounds and their metabolites

[0165] The machine learning model in step 2) is used to perform non-targeted identification of fentanyl compounds and in vitro incubation metabolites. The identification results are as follows: Fig.10 As shown in the figure, it can be seen that the model successfully identified and classified the known metabolites of fentanyl into fentanyl-like compounds, and also identified a new fentanyl metabolite that has not been reported in the literature (i.e. Fig.10 compounds marked as “Unknown” in the

[0166] In order to further verify the metabolites of this new fentanyl compound, we will obtain the corresponding standard samples for compound verification. The results are as follows Fig.11 As shown in the figure, a is the EIC graph of the new metabolite standard; b is the EIC graph of the fentanyl metabolite sample. It is inferred that the metabolite with a retention time of RT of 4.503 is the new fentanyl compound metabolite, and its chemical structure is inferred.

[0167] Compounds were synthesized according to the inferred chemical structure of the metabolites as metabolite standards. The metabolite standards and the novel metabolites in the incubation samples were detected by the method of step 3) to obtain MS / MS spectra (Figures c and d). By comparing the metabolic standard in Figure c with the MS / MS spectrum of the fentanyl in vitro incubation metabolite in Figure d, it can be found that the mass spectra of the two are basically consistent, indicating that the novel metabolite successfully predicted by the machine learning model of the present invention is determined to be a new fentanyl compound, and the model is also very accurate in predicting unknown fentanyl compounds.

[0168] The present invention identifies new fentanyl compounds for the first time, expands the exploration and identification of unknown fentanyl compounds, enriches the identification scope of fentanyl compounds, provides a strong technical basis for the control of such compounds, and is conducive to the promotion and use in related fields.

[0169] Example 7 Investigating the effects of different methods on the extraction of fentanyl compounds

[0170] Example 7-A MCX solid phase extraction method

[0171] 1) The 24 fentanyl compounds in Example 1 were added to the urine matrix to obtain a solution with a final concentration of 100 ng / ml, and extracted using MCX solid phase extraction material: pure methanol and 5v% methanol were used to activate the MCX magnetic solid phase extraction material in sequence, and treated with eluent and eluting liquid, and re-dissolved with the initial mobile phase (95% mobile phase A + 5% mobile phase B, the composition of mobile phases A and B is the same as the mobile phase in Example 1), centrifuged, and the supernatant was taken, i.e., the sample to be tested.

[0172] The eluent, elution solution, mixing and treatment time are the same as those in step 3-1) of Example 4.

[0173] 2) Collecting MS / MS mass spectrometry data:

[0174] A liquid chromatography quadrupole time-of-flight mass spectrometer (LC-QTOF MS mass spectrometer) was used to collect MS / MS mass spectral data of fentanyl compounds.

[0175] Chromatographic conditions were as follows: mobile phase A: 0.1 v% formic acid + 99.9 v% 10 mM ammonium formate aqueous solution, mobile phase B: a mixed solution of 95 v% acetonitrile + 5 v% mobile phase A (0.1 v% formic acid + 99.9 v% 10 mM ammonium formate aqueous solution); flow rate: 0.3 mL / min; elution gradient: 5% B (0 min), 5% B (0.5 min), 15% A (1.5 min), 30% A (4 min), 30% A (7.5 min), 90% A (8 min), 90% A (10 min), 5% A (10.1 min), 5% A (12 min); injection volume: 5 μL.

[0176] The mass spectrometry conditions were as follows: ion source ESI+, curtain gas 35 psi, ionization voltage 5500 kV, temperature 600°C, spray gas 55 psi, auxiliary heating gas 55 psi, declustering voltage 100 V, cone voltage 65 V, collision energy 30 V, mass spectrometry acquisition mode was MS-IDA mode, the primary mass spectrometry acquisition range was 100-1000 m / z, and the secondary mass spectrometry acquisition range was 50-1000 m / z.

[0177] Example 7-A Spectrum data collection results are shown in Fig.12 As can be seen from the left figure, when using the MCX solid phase extraction cartridge, some fentanyl compounds (such as valerylfentanyl and α-methylthiofentanyl) at low concentrations (0.5 ng / mL) cannot be effectively extracted; in the right figure, when using the MCX solid phase extraction cartridge, some medium and high concentrations (10 ng / mL) have a tailing effect. Therefore, the MCX solid phase extraction material is not suitable for the comprehensive and effective extraction of fentanyl compounds, which is not conducive to the data collection, training and learning of the subsequent machine learning model, and reduces the effectiveness of fentanyl compound identification.

[0178] Example 7-B HLB magnetic solid phase extraction method

[0179] 1) The 24 fentanyl compounds in Example 1 were added to the urine matrix to obtain a solution with a final concentration of 100 ng / ml, and extracted using an HLB magnetic solid phase extraction material. The other steps were the same as in Example 7-A. Fig.13 As shown, it can be seen that the MS / MS spectra of 24 fentanyl compounds have more complete peak shapes, high quality, and no tailing phenomenon, indicating that the HLB magnetic solid phase extraction method is more suitable for the extraction of fentanyl compounds. When the peak shapes are comprehensive and accurate, it is more conducive to the subsequent training and learning of machine models, so that an accurate and efficient learning model can be obtained to identify and screen various fentanyl compounds.

[0180] Example 7-C Direct Dilution Method

[0181] The 24 fentanyl compounds in Example 1 were extracted using a direct dilution method, specifically, corresponding volumes of 24 fentanyl standards were added to 1 mL of urine, and the samples were diluted step by step to concentrations of 10 ng / mL, 5 ng / mL, 4 ng / mL, 3 ng / mL, 2 ng / mL, 1 ng / mL, and 0.5 ng / mL, 2 mL of water was added for dilution, mixed, dried with nitrogen, and 250 uL of initial mobile phase (95% mobile phase A + 5% mobile phase B, the composition of mobile phases A and B was the same as the mobile phase in Example 1) was added, mixed, centrifuged at 12000 rpm for 10 min, and the supernatant was taken for sampling and analysis, wherein the test conditions of the liquid chromatography quadrupole time-of-flight mass spectrometer (LC-QTOFMS mass spectrometer) were the same as those in Example 7-A.

[0182] The results show that some fentanyl compounds in urine, such as acetylfentanyl, ocfentanil, remifentanil, and cyclopropylfentanyl, cannot be effectively extracted. In addition, Table 1 also shows the detection limits (LODs) at which various fentanyl compounds can be accurately identified by the Binning-RF machine learning model in Example 1 after using the dilution method of Example 7-C and the HLB magnetic solid phase extraction method of Example 1, respectively. The results show that the use of the magnetic solid phase extraction method of Example 1 can significantly reduce the detection limit of the machine learning model, and has better compatibility with the machine learning model Binning-RF of the present invention, and can fully cope with the detection of trace or minute amounts of fentanyl compounds; while the dilution rule of Example 7-C has greater limitations and cannot guarantee the accuracy and effectiveness of the detection of minute amounts of fentanyl compounds.

[0183] Table 1

[0184]

[0185]

[0186] Example 8 Investigating the extraction effects of different chromatographic mobile phases on fentanyl compounds

[0187] Example 8-A

[0188] 1) Take 24 methanol standard solutions of fentanyl and mix them to a concentration of 100 ng / mL;

[0189] 2) Chromatographic conditions:

[0190] Mobile phase A: Mobile phase A is an aqueous solution containing 0.1v% formic acid, and mobile phase B is methanol; flow rate: 0.3mL / min; elution gradient: 5%B (0min), 5%B (0.5min), 15%A (1.5min), 30%A (4min), 30%A (7.5min), 90%A (8min), 90%A (10min), 5%A (10.1min), 5%A (12min); injection volume: 5μL.

[0191] The mass spectrometry conditions were:

[0192] Ion source ESI+, air curtain gas 35psi, ionization voltage 5500kV, temperature 600℃, spray gas 55psi, auxiliary heating gas 55psi, declustering voltage 100V, cone voltage 65V, collision energy of 10V, 20V, 30V, and 40V were used for fragmentation. The mass spectrometry acquisition mode was MS-IDA mode, the primary mass spectrometry acquisition range was 100-1000m / z, and the secondary mass spectrometry acquisition range was 50-1000m / z. The extraction effect diagram of each fentanyl compound is shown in Fig.14 a.

[0193] Example 8-B

[0194] Only the composition of the chromatographic conditions mobile phases A and B in step 2) was changed, and the other steps were the same as in Example 8-A.

[0195] Wherein, mobile phase A is 0.1v% formic acid + 99.9v% 10mM ammonium formate aqueous solution, and mobile phase B is acetonitrile. The extraction effect diagram of each fentanyl compound is shown in Fig.14 b.

[0196] Example 8-C

[0197] Only the composition of the chromatographic conditions mobile phases A and B in step 2) was changed, and the other steps were the same as in Example 8-A.

[0198] Wherein, mobile phase A is 0.1v% formic acid + 99.9v% 10mM ammonium formate aqueous solution, and mobile phase B is a mixed solution of 95v% acetonitrile + 5v% mobile phase A (0.1v% formic acid + 99.9v% 10mM ammonium formate aqueous solution). The extraction effect diagram of each fentanyl compound is shown in Fig.14 c.

[0199] according to Fig.14It can be seen that, compared with Example 8-A and Example 8-B, Example 8-C is more comprehensive and of higher quality for the extraction of various fentanyl compounds, such as fentanyl, sufentanyl, 4-fluoroisobutyrlfentanyl, 3-methylthiofentanyl, and butyrfentanyl. The peak shape is better; correspondingly, the secondary mass spectra (MS / MS) of each fentanyl compound obtained are of higher quality, thereby making the machine learning model training effect better and the prediction accuracy higher. Therefore, it can be determined that the specific combination of mobile phases used in the present invention can obtain more comprehensive and high-quality MS / MS mass spectrum information, whether in the construction of the self-built library of fentanyl compounds or in the detection process of the compounds to be tested, and can optimize the identification and screening of various fentanyl compounds with an accurate machine learning model.

[0200] Comparative Example 1

[0201] Step 2-2) The LR classification model is used to replace the RF classification model in Example 1. The other steps are the same as Example 1.

[0202] Comparative Example 2

[0203] Step 2-2) The RF classification model in Example 1 is replaced by the SVM classification model, and the other steps are the same as those in Example 1.

[0204] Comparative Example 3

[0205] Step 2-2) The RF classification model in Example 1 is replaced by the KNN classification model, and the other steps are the same as those in Example 1.

[0206] Fig.15 The results of the machine learning models LR, SVM, and KNN in comparison examples 1-3 are compared with the random forest model RF in Example 1. By comparing the results of support vector machine, K nearest neighbor, logistic regression, F1 score, MCC score, etc., it is determined that the result of the random forest model is optimal, and combined with the feature extraction method, the optimal classification model Binning-RF is established.

[0207] It can be seen from the above embodiments and comparative examples that the present invention utilizes a specific fentanyl compound extraction process and liquid chromatography detection conditions to effectively extract various fentanyl compounds (including known and unknown derivatives, metabolites, and skeleton-modified compounds, etc.), and cooperates with a specific Binning-RF model to perform machine learning and classification on MS / MS mass spectrometry data. By means of nested cross-validation and Bayesian optimization, the accuracy of the model is evaluated while optimizing hyperparameters, and an efficient and accurate machine learning model is constructed for the first time. By comparing with the model, the present invention can quickly identify, screen and detect various fentanyl compounds (including known and unknown derivatives, metabolites, and skeleton-modified compounds, etc.), and provides effective technical support for the discovery, identification and supervision of fentanyl compounds.

[0208] The above contents are only preferred embodiments of the present invention and are not intended to limit the implementation scheme of the present invention. A person skilled in the art can easily make corresponding changes or modifications based on the main concept and spirit of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope required by the claims.

Claims

1. A method for rapid screening of fentanyl compounds based on a machine learning model, characterized in that: The steps include: 1) Establish a MS / MS mass spectrometry database of fentanyl compounds; 2) Constructing a machine learning model for fentanyl compounds: 2-1) Feature extraction: Use Top N method, Binning method or Grid method to extract and optimize the features of MS / MS mass spectrometry data; 2-2) Model construction: Build a random forest RF classification model for learning classification; 2-3) Model evaluation: Use accuracy, precision, recall, F1 score and MCC score as evaluation indicators of the model; 3) Extracting fentanyl compounds from the sample to be tested and collecting MS / MS mass spectrometry data: 3-1) extracting fentanyl compounds from the sample using an automatic magnetic extraction method; 3-2) using a liquid chromatography quadrupole time-of-flight mass spectrometer to collect MS / MS mass spectrum data of fentanyl compounds; 4) Rapid screening and identification of fentanyl compounds.

2. The method according to claim 1, characterized in that In step 1), the secondary mass spectrometry data of fentanyl compounds are collected and preprocessed to obtain a MS / MS mass spectrometry database of fentanyl compounds; The secondary mass spectrometry data of fentanyl compounds include secondary mass spectrometry data collected in the laboratory and valid secondary mass spectrometry data on fentanyl compounds in the online public database; Secondary mass spectrometry data collected by the laboratory: using a liquid chromatography quadrupole time-of-flight mass spectrometer to collect MS / MS mass spectrometry data of known fentanyl compounds; The secondary mass spectrometry data of the online public database include the MS / MS mass spectrometry data of known fentanyl compounds in the National Institute of Standards and Technology (NIST), MassBank of North America (MoNA) and HighResNPS online databases.

3. The method according to claim 2, characterized in that The chromatographic conditions are as follows: mobile phase A is a combination of formic acid and an aqueous solution of ammonium formate, wherein the volume ratio of the formic acid to the aqueous solution of ammonium formate is (1-5):1000, and the concentration of ammonium formate in the aqueous solution of ammonium formate is 10-15 mM; mobile phase B is a combination of acetonitrile and mobile phase A, wherein the volume percentage of acetonitrile is 90-95%; The mass spectrometry conditions were as follows: ion source ESI+, curtain gas 30-35psi, ionization voltage 5000-5500kV, temperature 500-600°C, spray gas 50-55psi, auxiliary heating gas 50-55psi, declustering voltage 100-120V, cone voltage 60-65V; the mass spectrometry acquisition mode was MS-IDA mode, the primary mass spectrometry acquisition range was 100-1000m / z, and the secondary mass spectrometry acquisition range was 50-1000m / z.

4. The method according to claim 2 or 3, characterized in that: The chemical formula, InChIKey, precursor ion, collision energy, molecular weight, MS / MS mass spectrum data, adduct form and retention time of the collected fentanyl compounds were entered into the self-built library. After normalization, data amplification and data cleaning preprocessing steps, an MS / MS mass spectrum database of fentanyl compounds was formed.

5. The method according to claim 4, characterized in that 1-1) Data extraction: Use MS-DIAL 4.60 software to extract peaks from the acquired wiff format files; 1-2) Normalization: The absolute response intensity is converted into a relative response intensity, and normalized to the maximum relative response value. The normalization formula is as follows (I): Relative response value = (absolute response value / maximum absolute response value) × maximum relative response value (I); 1-3) Data amplification: for each original spectrum, retain the spectrum data above the maximum response intensity x% (x=50, 60, 70, 80, 90, 100) and perform data amplification to match the number of negative data sets; 1-4) Data cleaning: After data amplification, spectra with fragment numbers less than 5 were removed, and the MS / MS mass spectrum database of fentanyl compounds was established.

6. The method according to claim 4, characterized in that 2-1) The Binning method is used to extract features and optimize parameters of MS / MS mass spectrometry data; the parameters include intensity threshold, m / z range of fentanyl compound precursor ions and bin width.

7. The method according to claim 4, characterized in that 2-2) Based on the CART (Classification and Regression Trees) algorithm, develop the RF classification model, determine the selection of split points and pruning strategies by building and evaluating decision trees; use the Gini index to select the optimal feature and determine the optimal binary split point of the feature; The Gini index calculation formula is as described in (II): Where Pi is the proportion of samples of the i-th category in the data set, and k is the total number of categories; The prediction of the RF model is the voting result of the prediction results of all decision trees. When there are N decision trees, the probability of predicting category C can be expressed as formula (III): In the formula, I(C i =C) is an indicator function, indicating whether the prediction of the i-th decision tree is category C; A 5*3 nested cross validation was used to evaluate the performance of the model and optimize the model hyperparameters.

8. The method according to claim 7, characterized in that The outer loop is used for 5 times to evaluate the model algorithm, and the inner loop is used for 3 times to optimize the hyperparameters in Bayesian way. The hyperparameters include: n_estimators, max_features, max_depth, min_samples_split, min_samples_leaf and bootstrap of the random forest RF model.

9. The method according to claim 1, characterized in that: Step 3-1) using pure methanol and 5% methanol to activate the magnetic solid phase extraction material in sequence, and using eluent and eluting liquid to obtain an extract of fentanyl compounds; 3-2) Chromatographic conditions are as follows: mobile phase A is a combination of formic acid and an aqueous solution of ammonium formate, wherein the volume ratio of formic acid to the aqueous solution of ammonium formate is (1-5):1000, and the concentration of ammonium formate in the aqueous solution of ammonium formate is 10-15 mM; mobile phase B is a combination of acetonitrile and mobile phase A, wherein the volume percentage of acetonitrile is 90-95%; flow rate: 0.3-0.5 mL / min; gradient elution is adopted, and the injection volume is controlled to be 2-5 μL; The mass spectrometry conditions were as follows: ion source ESI+, curtain gas 30-35psi, ionization voltage 5000-5500kV, temperature 500-600°C, spray gas 50-55psi, auxiliary heating gas 50-55psi, declustering voltage 100-120V, cone voltage 60-65V; the mass spectrometry acquisition mode was MS-IDA mode, the primary mass spectrometry acquisition range was 100-1000m / z, and the secondary mass spectrometry acquisition range was 50-1000m / z.

10. Use of the method according to any one of claims 1 to 9 in the detection and identification of fentanyl compounds.

Images