A method for screening of anti-mrsa active molecules based on xgboost algorithm, storage medium and equipment

Abstract

The present application belongs to the technical field of artificial intelligence assisted drug screening, and particularly relates to an anti-MRSA active molecule screening method based on an XGBoost algorithm, a storage medium and an equipment. The method constructs an XGBoost integrated learning framework, and fuses ADME screening rules optimized for antibacterial drugs and multi-target molecule docking verification technology. Through the strategy, the data imbalance problem caused by the scarcity of known active samples and the huge compound library in drug screening is effectively overcome, and the screening hit rate is significantly improved. The present application also provides a computer readable storage medium and an electronic equipment for storing and executing the above method. The scheme realizes efficient and accurate anti-MRSA candidate drug discovery, provides a powerful intelligent screening tool for coping with super bacterial infection, and has important application value.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence-assisted drug screening technology, specifically relating to a method, storage medium, and device for screening active molecules against MRSA based on the XGBoost algorithm. Background Technology

[0002] Methicillin-resistant Staphylococcus aureus staphylococcus aureus MRSA (metastatic antibiotics) is a serious drug-resistant pathogen worldwide, causing a variety of severe infectious diseases such as skin and soft tissue infections, pneumonia, sepsis, and even septicemia. Its rapid spread, high pathogenicity, and limited clinical treatment options have made it a major public health challenge in the global healthcare field. With the irrational use and abuse of antibiotics in clinical practice, the drug resistance spectrum of MRSA continues to expand, and the efficacy of traditional first-line antibacterial drugs has significantly decreased. Therefore, the development of novel anti-MRSA drugs with entirely new mechanisms of action is urgently needed.

[0003] However, traditional high-throughput screening is costly and time-consuming. Although machine learning-based virtual screening has been widely used, it still faces two major challenges: First, known compounds with anti-MRSA activity are extremely scarce, while the size of the compound library to be screened often reaches millions or even tens of millions, resulting in an extremely unbalanced distribution of active and inactive samples. Machine learning models tend to favor the majority class, producing a high false positive rate and making it difficult to accurately identify truly active molecules. Second, existing screening methods lack targeted optimization. Conventional downsampling strategies are prone to losing important structural and feature information in inactive samples when balancing data distribution, and they do not incorporate specific screening rules based on the drug-likeness characteristics of antibacterial drugs. This results in candidate molecules with poor drug-likeness, high off-target risk, and insufficient affinity for bacterial targets.

[0004] In summary, given the shortcomings of existing technologies, there is an urgent need to develop a virtual screening method that is suitable for screening anti-MRSA active molecules, can overcome the bottleneck of extreme data imbalance, has both high accuracy and high interpretability, and is equipped with a complete validation system. Furthermore, it is necessary to complete the engineering transformation of this method and construct an intelligent screening system for anti-MRSA active molecules that is easy to operate, functionally integrated, and meets the actual needs of drug development. This would improve the screening hit rate of anti-MRSA active molecules and provide efficient and reliable technical support for the development of novel anti-MRSA antibacterial drugs. Summary of the Invention

[0005] This invention aims to solve the above-mentioned technical problems and provides a method, storage medium and device for screening active molecules against MRSA based on the XGBoost algorithm, which significantly improves the screening hit rate and reliability.

[0006] To achieve this objective, the present invention adopts the following technical solution: A method, storage medium, and device for screening MRSA-resistant active molecules based on the XGBoost algorithm include the following steps: S1. Construct a positive sample set P and a library of compounds to be screened U, and perform molecular standardization. S2. Extract features from the molecules obtained in step S1 to generate physicochemical property descriptors and molecular structure fingerprints, and perform feature fusion to obtain the numerical feature matrix X. S3. A training subset is formed by all positive samples and balanced negative samples randomly drawn from the compound library U to be screened. The base classifier is trained and iterated N times. The consensus score of the predicted probability output by each base classifier is obtained. S4. Based on the consensus score and the applicable domain model, a preliminary screening is conducted to select high-confidence candidate molecules. S5. Perform ADMET property screening on the candidate molecules obtained in step S4 to optimize them for antimicrobial drugs, and remove molecules that do not meet the drug-likeness and safety thresholds. S6. Perform multi-target molecular docking verification on the candidate molecules obtained in step S5, retain molecules with strong binding affinity to at least one anti-MRSA target, and obtain the final screening results. Furthermore, the positive sample set P in step S1 consists of 64 compounds that have been experimentally verified to have anti-MRSA activity; the compound library U to be screened is obtained from the ChEMBL database; the ratio of positive samples to unlabeled samples is 1:30,000.

[0007] Furthermore, the physicochemical property descriptors mentioned in step S2 include: molecular weight, lipid-water partition coefficient, topological polar surface area, number of hydrogen bond donors, and number of hydrogen bond acceptors; the molecular structure fingerprint is a MACCS Keys fingerprint.

[0008] Furthermore, in step S3, the number of iterations N is set to 50; In the i-th iteration, all 64 positive samples are retained, and 500 samples are randomly selected from the unlabeled pool U as negative samples for the current round to form the training set; After training the base classifier, predictions are made for all unlabeled samples; After the iteration is completed, the arithmetic mean of the predicted probabilities of each molecule in the 50 base classifiers is calculated as the final consensus score.

[0009] Furthermore, the consensus score threshold in step S4 is 0.9; the applicable domain filtering is as follows: calculate the Tanimoto similarity between the candidate molecule and 64 positive samples, and remove molecules with a maximum similarity of less than 0.7.

[0010] Furthermore, in the multi-target molecular docking verification described in step S6, the targets are selected from the following 10 anti-MRSA-related proteins: AgrA, CcrB, GdpP, IcaC, MenA, PBP2a, SarA, Sortase A, VraR, and VraS; the docking tool is Autodock Vina; and the binding energy threshold is -9.0 kcal / mol.

[0011] Furthermore, the method verifies the effectiveness and robustness of the model through one or more of the following methods: Chemical spatial PCA distribution analysis was used to verify the distinguishability between positive and inactive samples; ROC curve and AUC value analysis showed that the average AUC reached 0.92; Analysis of PR curve and AP value showed that the average AP was 0.71. Correlation analysis of the base classifiers showed correlation coefficients ranging from 0.67 to 0.88, demonstrating model diversity. SHAP interpretability analysis identifies key features including electrical topology state, LogP, and TPSA; The Y-permutation test showed that the AUC dropped to 0.5 after shuffling the labels, proving that the model was not overfitting.

[0012] Based on the same inventive concept, this invention also provides the application of the active molecules obtained by the method in the preparation of anti-MRSA products, wherein the active molecules include compounds as shown in the following CHEMBL ID: CHEMBL1619602, CHEMBL1689750, CHEMBL1739714, CHEMBL1739739, CHEMBL1782478, CHEMBL2057893, CHEMBL2057894, CHEMBL2104067, CHEMBL211 5054, CHEMBL276178, CHEMBL293043, CHEMBL29530, CHEMBL303949, CHEMBL3581926, CHEMBL3581933, CHEMBL4242024, CHEMBL54607, CHEMBL55252.

[0013] Based on the same inventive concept, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the above-mentioned virtual screening and activity prediction method for the membrane permeability of carbapenem-resistant Enterobacteriaceae is implemented.

[0014] Based on the same inventive concept, the present invention also provides an electronic device, comprising: a processor and a memory; the memory for storing a computer program; and the processor for executing the computer program stored in the memory to cause the electronic device to perform the above-described method.

[0015] The present invention has the following beneficial effects: (1) Breaking through the bottleneck of extreme data imbalance Under extreme conditions of only 64 positive samples and 1.92 million unlabeled samples (ratio 1:30,000), this invention achieves an average AUC of 0.92 and an average AP of 0.71 through an iterative random undersampling ensemble strategy, which is significantly better than traditional methods.

[0016] (2) Significantly improves screening hit rate and reduces wet experiment cost. Through consensus scoring and application domain filtering, only 1,141 molecules were initially screened from 1.92 million molecules; after optimization with ADMET, this was reduced to 103; and further refined to 19 through multi-target docking. This stepwise convergence strategy significantly reduced false positives and avoided unnecessary wet experiments.

[0017] (3) Overcoming the limitations of the "five principles of drug-like formulation" in antibiotic screening This invention addresses the characteristics of antibacterial drugs by reasonably relaxing parameters such as molecular weight (≤800 Da), LogP (-2~6), and polar surface area, thereby avoiding the omission of potential antibiotics with macrocyclic skeletons or moderate lipophilic properties and improving the efficiency of antibacterial drug discovery.

[0018] (4) Revealing the multi-target synergistic mechanism Through systematic docking of 10 key MRSA targets, candidate compounds (such as CHEMBL3581933) were found to have strong binding affinity (<-9.5 kcal / mol) to both MenA and PBP2a, suggesting their potential for multi-target synergistic antibacterial activity.

[0019] (5) The model is highly interpretable and consistent with the understanding of medicinal chemistry. SHAP analysis shows that the model relies on physicochemical features such as LogP, TPSA, and electrotopological states to make decisions, which is highly consistent with the mechanism by which antimicrobial drugs penetrate the bacterial cell wall, proving that the model learns the true structure-activity relationship rather than data noise.

[0020] (6) A complete multi-level verification system This invention constructs a full-chain screening and verification process of "artificial intelligence prediction → ADMET → molecular docking". Among the 19 candidate molecules, one has been confirmed by literature to have real antibacterial activity, and the technical solution is complete and reliable.

[0021] (0) Provides a dedicated hardware execution environment for anti-MRSA drug screening. This invention also provides a computer-readable storage medium and an electronic device. The storage medium embeds the screening method of this invention into an executable program, which not only facilitates rapid deployment and standardized promotion in different computing environments (such as local servers or the cloud), but also effectively avoids human error and ensures high reproducibility of the screening results. The electronic device, by integrating a processor and memory, can automatically execute the entire process from massive data cleaning and XGBoost model prediction to multi-level rule filtering and molecular docking verification. This hardware-software combined solution compresses the high-throughput screening work that originally took weeks into a very short time, significantly improving computational efficiency and realizing the intelligent, automated, and large-scale discovery of anti-MRSA drugs, providing pharmaceutical companies with an efficient and low-cost engineering solution. Attached Figure Description

[0022] Figure 1 : Chemical spatial PCA distribution map, showing the clustering trend of positive samples; Figure 2 : ROC curve of the ensemble model, AUC=0.92; Figure 3 : PR curve of the ensemble model, AP=0.71; Figure 4 Correlation heatmap of base classifiers, correlation coefficient 0.67~0.88; Figure 5 SHAP Feature Importance Summary Diagram; Figure 6 Y-displacement test box plot; Figure 7 : Binding energy heatmap of docking between 19 candidate molecules and 10 target molecules. Detailed Implementation

[0023] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments, but this should not be construed as limiting the invention. Unless otherwise specified, the technical means used in the following embodiments are conventional means well known to those skilled in the art, and the materials, reagents, etc. used in the following embodiments are commercially available unless otherwise specified.

[0024] Example 1: A method, storage medium, and device for screening active molecules against MRSA based on the XGBoost algorithm.

[0025] The method flow of this embodiment includes the following steps: Step S1: Data Acquisition and Preprocessing Constructing the Positive Set: The inventors collected and organized data on 64 compounds with experimentally verified anti-MRSA activity by extensively reviewing existing literature and public databases (as shown in Table 1). The data was cleaned to remove duplicates and invalid entries, and the SMILES (Simplified Molecular Linear Input Canonical) strings of the compounds were extracted to form the positive set $P$.

[0026] Table 1. CAS and SMILES of compounds with experimentally verified anti-MRSA activity. Constructing the Unlabeled Pool: The ChEMBL database was selected as the screening source, and all small molecule compounds, totaling 1,920,603, were extracted. Their SMILES strings and unique identifiers (ChEMBL IDs) were extracted to form a large-scale unlabeled sample set U. At this point, the ratio of positive samples to unlabeled samples was approximately 1:30,000, indicating a highly imbalanced data distribution.

[0027] Standardization: All the SMILES strings above are standardized using cheminformatics tools (RDKit is preferred in this embodiment), including desalting, neutralizing charges, and normalizing tautomers, to ensure the uniqueness of the molecular structure and machine readability.

[0028] Step S2: Molecular feature extraction and vectorization representation In order to convert the chemical structure into a numerical vector that can be processed by a computer, this embodiment performs feature extraction on all molecules in step S1: Physicochemical property feature extraction: Calculate the physicochemical property descriptors of molecules, including molecular weight (MolWt), lipid-water partition coefficient (MolLogP), topological polar surface area (TPSA), number of hydrogen bond donors, number of hydrogen bond acceptors, etc., to characterize the pharmacokinetic properties of molecules.

[0029] Structural fingerprint extraction: Calculate the MACCS Keys fingerprint (167 bits) of the molecule, encoding the molecular structure into a binary feature vector to capture specific chemical substructure fragments.

[0030] Feature fusion: The physicochemical property descriptor vector and the structural fingerprint vector are concatenated to form the final numerical feature matrix X.

[0031] Step S3: Construct an ensemble screening model based on iterative undersampling To address the data imbalance problem described in step S1, this invention employs an integration strategy based on the Bagging PU (Positive-Unlabeled) concept: Initialization settings: Set the number of iterations for ensemble learning to $N=50$, and select XGBoost (ExtremeGradient Boosting) as the base classifier.

[0032] Iterative training: In the i$-th iteration (i=1 to 50): Balanced sampling: All 64 positive samples are retained, and 500 samples are randomly selected from 1,920,603 unlabeled samples as "negative samples" to construct the balanced training set for the current round (a total of 564 samples).

[0033] Model training: Train the i-th XGBoost base classifier based on the balanced training set.

[0034] Full library prediction: The trained base classifier is used to predict the entire library of compounds to be screened (1,920,603 molecules), and the prediction probability is recorded.

[0035] Consensus Scoring: After the loop ends, the arithmetic mean of the predicted probabilities of each molecule in the 50 base classifiers is calculated as the final consensus score for that molecule.

[0036] Step S4: Primary Filtering and Applicable Domain Filtering High-confidence screening: Set the consensus score threshold to 0.9. Only retain molecules with a score > 0.9.

[0037] Application Domain (AD) Filtering: To ensure candidate molecules are within the reliable prediction range of the model, the Tanimoto structural similarity between the candidate molecules and 64 positive samples is calculated. A maximum similarity threshold of 0.7 is set, and molecules with similarity values ​​below this value are removed.

[0038] Screening results: Through this step, 1,141 high-confidence preliminary screening compounds were obtained from 1,920,603 initial molecules.

[0039] Step S5: ADMET screening optimized for antimicrobial drugs To address the limitations of the traditional "five principles of drug-likeness" in antimicrobial drug screening, this invention establishes a specific ADMET screening criterion. The 1,141 compounds obtained in step S4 were subjected to computational screening.

[0040] Physicochemical property filtering: Set the following thresholds: Molecular weight (MW): 800Da (Given that effective antibiotics often have large molecular skeletons, the restrictions are relaxed appropriately.) Lipid-water partition coefficient (LogP): -2.0~6.0 ; Hydrogen bond donor (HBD): 10 ; Hydrogen bond acceptor (HBA): 15 ; Topological polar surface area (TPSA): ; Rotatable Bonds: 15 .

[0041] PAINS Filtering: Utilizes a generalized assay interference compound (PAINS) filter to remove molecules containing structures that may contain false positives or toxicity warnings.

[0042] Screening results: Through this step, 103 candidate compounds with good drug development potential were selected.

[0043] Step S6: Validation of molecular docking in multi-target systems To further confirm the mechanism of action and binding affinity of the candidate molecules, systematic molecular docking was performed on the 103 compounds obtained in step S5: Target selection: Ten classic anti-MRSA target proteins were selected, including: AgrA, CcrB, GdpP, IcaC, MenA, PBP2a, SarA, Sortase A, VraR, and VraS.

[0044] Docking implementation: A semi-flexible docking algorithm was used to simulate the binding of small molecules (Autodock vina) to the above-mentioned target sites.

[0045] Threshold screening: A binding affinity threshold of -9.0 kcal / mol is set. Only molecules with binding energies to at least one target site below this threshold are retained.

[0046] Screening results: A total of 19 lead compounds with extremely strong theoretical binding affinity were selected.

[0047] Literature review confirmed that candidate molecules, including CHEMBL1082, possess genuine antibacterial activity (as shown in Table 2), and the technical solution is complete and reliable. These results demonstrate the effectiveness of the model and indicate that the remaining compounds are highly likely to possess antibacterial activity and have the potential to be formulated into anti-MRSA products.

[0048] Table 2. Compounds with confirmed antibacterial activity based on literature review. To verify the effectiveness of the method of the present invention, the inventors conducted experimental verification and recorded relevant data: Figure 1 The PCA analysis plot shows the chemical spatial distribution of the original training dataset. Red dots represent the positive sample set (known active molecules), and gray dots represent the library of compounds to be screened (background molecules), showing their distribution in the PCA dimensionality reduction space.

[0049] analyze: Figure 1 Principal component analysis results showed that, although the library to be screened covered a wide chemical space, the known anti-MRSA active molecules (red dots) were not uniformly distributed, but rather exhibited a clear clustering tendency, concentrated in specific chemical space regions.

[0050] Conclusion: This distribution characteristic indicates that anti-MRSA active molecules possess specific structural or physicochemical property patterns, making them distinguishable from background molecules in the feature space. This verifies the effectiveness of the feature engineering method (physicochemical properties + fingerprint) selected in this invention, providing a solid data foundation for the subsequent construction of machine learning classification models.

[0051] Figure 2 The receiver operating characteristic (ROC) curve of the integrated model of this invention is shown. The solid red line in the figure represents the average ROC curve of the model after 50 iterations, the red shaded area represents the standard deviation range, and the gray dashed line is the random guess baseline.

[0052] Analysis: Experimental results show that the model's average AUC (area under the curve) is as high as 0.92. Meanwhile, the red shaded area is very narrow, indicating that the performance fluctuations of the 50 base classifiers are minimal, demonstrating a high degree of consistency.

[0053] Conclusion: This confirms that even with an extremely large ratio of positive to negative samples (1:30,000), the integrated undersampling strategy of this invention can still stably and accurately distinguish active molecules from background noise, and the model has extremely high classification robustness.

[0054] Figure 3 The precision-recall (PR) curves of the ensemble model of this invention are shown. The solid blue line in the figure represents the change in precision of the model under different recall rates.

[0055] Analysis: In extremely imbalanced datasets, the baseline precision-recall (PR) value is typically close to 0. However, this model achieved a mean precision-recall (AP) value of 0.71. This means that among the samples predicted as "active" by the model, a very high proportion are genuine active molecules, rather than false positives.

[0056] Conclusion: The results demonstrate the practical application value of the method of this invention in screening massive compound libraries, which can effectively reduce the "false positive rate" of subsequent wet experimental verification and significantly save research and development costs.

[0057] Figure 4 This shows a heatmap of Pearson correlations between base classifiers in the ensemble model. The heatmap displays the correlation coefficients between the predictions of each pair of the top 10 base classifiers (XGBoost), with darker colors indicating higher correlations.

[0058] Analysis: Statistical analysis shows that the correlation coefficients between different base classifiers range from 0.67 to 0.88. This indicates that the base classifiers maintain consistency in the overall prediction direction (positive correlation) while also retaining significant differences (not perfect correlation).

[0059] Conclusion: This diversity stems from the "random undersampling" strategy of this invention, which allows each base classifier to learn features from different subsets of the library to be screened. This diversity is key to ensemble learning's ability to reduce variance and prevent overfitting.

[0060] Figure 5 This displays a summary plot of model feature importance based on SHAP values. Each row in the plot represents a feature, the color of the dots indicates the feature value (red for high, blue for low), and the position of the dots indicates the direction of their influence on the model output (right for positive contribution, left for negative contribution).

[0061] Analysis: Key features ranking highly include Electrotopological State, LogP (lipid-water partition coefficient), and TPSA (topological polar surface area). In particular, the distribution of LogP shows that moderate lipophilicity makes a significant positive contribution to predicting anti-MRSA activity.

[0062] Conclusion: This result is highly consistent with the pharmacological mechanism: anti-MRSA drugs must possess specific lipid solubility and polarity characteristics to penetrate the thick bacterial cell wall. This proves that the model is not a "black box," but has successfully captured a structure-activity relationship (SAR) with physicochemical significance.

[0063] Figure 6The results of the Y-Scrambling (Y permutation test) model robustness test are shown. The red box plot represents the performance distribution of the model trained with the true labels, and the gray box plot represents the performance distribution of the model after randomly shuffling the labels (Y-Scrambling).

[0064] Analysis: The AUC of the true label model is stable at around 0.92, while after shuffling the labels, the model's AUC drops sharply to around 0.5 (i.e., at the level of random guessing). There is a huge performance gap between the two.

[0065] Conclusion: This comparison strongly demonstrates that the high performance of the model in this invention stems from learning the true relationship between molecular structure and activity, rather than randomly fitting the dataset noise, thus confirming the statistical significance of the screening results.

[0066] Figure 7 This chart displays a heatmap of molecular docking binding energies between selected candidate molecules and multiple target sites. The vertical axis represents the IDs of the 19 selected candidate compounds, and the horizontal axis represents the 10 key anti-MRSA targets. Colors from red to blue indicate a gradual decrease in binding energy (increased binding strength).

[0067] Analysis: Several compounds exhibited broad-spectrum, strong binding affinity. In particular, for MenA (menadione biosynthetic enzyme) and PBP2a (a key protein for drug resistance), the binding energies of some lead compounds (such as CHEMBL3581933) were significantly lower than -9.5 kcal / mol (the dark blue area in the figure).

[0068] Conclusion: This not only validated the potential activity of the candidate molecules at the atomic level, but also revealed that these molecules may exert their anti-drug-resistant effects through a "multi-target synergistic mechanism," and have the potential to be developed into broad-spectrum anti-MRSA drugs.

[0069] Example 2: Computer Program Product This embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in Embodiment 1.

[0070] The computer-readable storage medium may be a storage medium such as ROM, RAM, disk, or optical disk.

[0071] Example 3: Electronic Equipment This embodiment provides an electronic device, including: a processor and a memory; the memory is used to store a computer program; the processor is used to execute the computer program stored in the memory to cause the electronic device to perform the method as described in Embodiment 1.

[0072] The electronic device may be a personal computer, server, workstation, high-performance computing cluster, or cloud computing platform.

[0073] It should be noted that when numerical ranges are mentioned in the claims of this invention, it should be understood that the two endpoints of each numerical range and any value between the two endpoints can be selected. To avoid redundancy, the present invention describes preferred embodiments.

[0074] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0075] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for screening MRSA-resistant active molecules based on the XGBoost algorithm, characterized in that, Includes the following steps: S1. Construct a positive sample set P and a library of compounds to be screened U, and perform molecular standardization. S2. Extract features from the molecules obtained in step S1 to generate physicochemical property descriptors and molecular structure fingerprints, and perform feature fusion to obtain the numerical feature matrix X. S3. A training subset is formed by all positive samples and balanced negative samples randomly drawn from the compound library U to be screened. The base classifier is trained and iterated N times. The consensus score of the predicted probability output by each base classifier is obtained. XGBoost is selected as the base classifier. S4. Based on the consensus score and the applicable domain model, a preliminary screening is conducted to select high-confidence candidate molecules. S5. The candidate molecules obtained in step S4 are screened for ADMET properties optimized for antibacterial drugs. The screening criteria include: molecular weight ≤ 800 Da; LogP: -2.0 ~ 6.0; hydrogen bond donors ≤ 10; hydrogen bond acceptors ≤ 15; TPSA ≤ 180 Ų; number of rotatable bonds ≤ 15; and molecules containing PAINS warning structures are excluded. S6. Perform multi-target molecular docking verification on the candidate molecules obtained in step S5, retain molecules with strong binding affinity to at least one anti-MRSA target, and obtain the final screening results.

2. The method according to claim 1, characterized in that, The positive sample set P in step S1 consists of 64 compounds that have been experimentally verified to have anti-MRSA activity; the compound library U to be screened is obtained from the ChEMBL database; the ratio of positive samples to unlabeled samples is 1:30,000.

3. The method according to claim 1, characterized in that, The physicochemical property descriptors mentioned in step S2 include: molecular weight, lipid-water partition coefficient, topological polar surface area, number of hydrogen bond donors, and number of hydrogen bond acceptors; the molecular structure fingerprint is a MACCSKeys fingerprint.

4. The method according to claim 1, characterized in that, In step S3, the number of iterations is set to N = 50; In the i-th iteration, all 64 positive samples are retained, and 500 samples are randomly selected from the compound library U to be screened as negative samples for the current round to form the training set. After training the base classifier, predictions are made for all unlabeled samples; After the iteration is completed, the arithmetic mean of the predicted probabilities of each molecule in the 50 base classifiers is calculated as the final consensus score.

5. The method according to claim 1, characterized in that, The consensus score threshold in step S4 is 0.9; the applicable domain filtering is: calculating the Tanimoto similarity between the candidate molecule and 64 positive samples, and removing molecules with a maximum similarity of less than 0.

7.

6. The method according to claim 1, characterized in that, In the multi-target molecular docking verification described in step S6, the targets were selected from the following 10 anti-MRSA-related proteins: AgrA, CcrB, GdpP, IcaC, MenA, PBP2a, SarA, Sortase A, VraR, and VraS; the docking tool was Autodock Vina; and the binding energy threshold was -9.0 kcal / mol.

7. The method according to any one of claims 1 to 6, characterized in that, This method verifies the effectiveness and robustness of the model through one or more of the following methods: Chemical spatial PCA distribution analysis was used to verify the distinguishability between positive and inactive samples; ROC curve and AUC value analysis showed that the average AUC reached 0.92; Analysis of PR curve and AP value showed that the average AP was 0.

71. Correlation analysis of the base classifiers showed correlation coefficients ranging from 0.67 to 0.88, demonstrating model diversity. SHAP interpretability analysis identifies key features including electrical topology state, LogP, and TPSA; The Y-permutation test showed that the AUC dropped to 0.5 after shuffling the labels, proving that the model was not overfitting.

8. The use of the active molecules screened by the method according to any one of claims 1 to 6 in the preparation of anti-MRSA products, characterized in that, The active molecules include compounds as shown in the following CHEMBL ID: CHEMBL1619602, CHEMBL1689750, CHEMBL1739714, CHEMBL1739739, CHEMBL1782478, CHEMBL2057893, CHEMBL2057894, CHEMBL2104067, CHEMBL211 5054, CHEMBL276178, CHEMBL293043, CHEMBL29530, CHEMBL303949, CHEMBL3581926, CHEMBL3581933, CHEMBL4242024, CHEMBL54607, CHEMBL55252.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-7.

10. An electronic device, characterized in that, include: Processor and memory; The memory is used to store computer programs; The processor is configured to execute a computer program stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-7.

Images