AptaDB: a comprehensive database integrating aptamer-target interactions.

Aptamers have emerged as research hotspots of the next generation due to excellent performance benefits and application potentials in pharmacology, medicine, and analytical chemistry. Despite the numerous aptamer investigations, the lack of comprehensive data integration has hindered the development of computational methods for aptamers and the reuse of aptamers. A public access database named AptaDB, derived from experimentally validated data manually collected from the literature, was hence developed, integrating comprehensive aptamer-related data, which include six key components: (i) experimentally validated aptamer-target interaction information, (ii) aptamer property information, (iii) structure information of aptamer, (iv) target information, (v) experimental activity information, and (vi) algorithmically calculated similar aptamers. AptaDB currently contains 1350 experimentally validated aptamer-target interactions, 1230 binding affinity constants, 1293 aptamer sequences, and more. Compared to other aptamer databases, it contains twice the number of entries found in available databases. The collection and integration of the above information categories is unique among available aptamer databases and provides a user-friendly interface. AptaDB will also be continuously updated as aptamer research evolves. We expect that AptaDB will become a powerful source for aptamer rational design and a valuable tool for aptamer screening in the future. For access to AptaDB, please visit http://lmmd.ecust.edu.cn/aptadb/.

MetaPredictor: in silico prediction of drug metabolites based on deep language models with prompt engineering.

Metabolic processes can transform a drug into metabolites with different properties that may affect its efficacy and safety. Therefore, investigation of the metabolic fate of a drug candidate is of great significance for drug discovery. Computational methods have been developed to predict drug metabolites, but most of them suffer from two main obstacles: the lack of model generalization due to restrictions on metabolic transformation rules or specific enzyme families, and high rate of false-positive predictions. Here, we presented MetaPredictor, a rule-free, end-to-end and prompt-based method to predict possible human metabolites of small molecules including drugs as a sequence translation problem. We innovatively introduced prompt engineering into deep language models to enrich domain knowledge and guide decision-making. The results showed that using prompts that specify the sites of metabolism (SoMs) can steer the model to propose more accurate metabolite predictions, achieving a 30.4% increase in recall and a 16.8% reduction in false positives over the baseline model. The transfer learning strategy was also utilized to tackle the limited availability of metabolic data. For the adaptation to automatic or non-expert prediction, MetaPredictor was designed as a two-stage schema consisting of automatic identification of SoMs followed by metabolite prediction. Compared to four available drug metabolite prediction tools, our method showed comparable performance on the major enzyme families and better generalization that could additionally identify metabolites catalyzed by less common enzymes. The results indicated that MetaPredictor could provide a more comprehensive and accurate prediction of drug metabolism through the effective combination of transfer learning and prompt-based learning strategies.

admetSAR3.0: a comprehensive platform for exploration, prediction and optimization of chemical ADMET properties.

Absorption, distribution, metabolism, excretion and toxicity (ADMET) properties play a crucial role in drug discovery and chemical safety assessment. Built on the achievements of admetSAR and its successor, admetSAR2.0, this paper introduced the new version of the series, admetSAR3.0, as a comprehensive platform for chemical ADMET assessment, including search, prediction and optimization modules. In the search module, admetSAR3.0 hosted over 370 000 high-quality experimental ADMET data for 104 652 unique compounds, and supplemented chemical structure similarity search function to facilitate read-across. In the prediction module, we introduced comprehensive ADMET endpoints and two new sections for environmental and cosmetic risk assessments, empowering admetSAR3.0 to provide prediction for 119 endpoints, more than double numbers compared to the previous version. Furthermore, the advanced multi-task graph neural network framework offered robust and reliable support for ADMET prediction. In particular, a module named ADMETopt was added to automatically optimize the ADMET properties of query molecules through transformation rules or scaffold hopping. Finally, admetSAR3.0 provides user-friendly interfaces for multiple types of input data, such as SMILES string, chemical structure and batch molecule file, and supports various output types, including digital, chart displays and file downloads. In summary, admetSAR3.0 is anticipated to be a valuable and powerful tool in drug discovery and chemical safety assessment at http://lmmd.ecust.edu.cn/admetsar3/.

Identification of vital chemical information via visualization of graph neural networks.

Qualitative or quantitative prediction models of structure-activity relationships based on graph neural networks (GNNs) are prevalent in drug discovery applications and commonly have excellently predictive power. However, the network information flows of GNNs are highly complex and accompanied by poor interpretability. Unfortunately, there are relatively less studies on GNN attributions, and their developments in drug research are still at the early stages. In this work, we adopted several advanced attribution techniques for different GNN frameworks and applied them to explain multiple drug molecule property prediction tasks, enabling the identification and visualization of vital chemical information in the networks. Additionally, we evaluated them quantitatively with attribution metrics such as accuracy, sparsity, fidelity and infidelity, stability and sensitivity; discussed their applicability and limitations; and provided an open-source benchmark platform for researchers. The results showed that all attribution techniques were effective, while those directly related to the predicted labels, such as integrated gradient, preferred to have better attribution performance. These attribution techniques we have implemented could be directly used for the vast majority of chemical GNN interpretation tasks.

Iridium-Catalyzed Enantioselective Transfer Hydrogenation of 1,1-Dialkylethenes with Ethanol: Scope and Mechanism.

Despite half a century's advance in the field of transition-metal-catalyzed asymmetric alkene hydrogenation, the enantioselective hydrogenation of purely alkyl-substituted 1,1-dialkylethenes has remained an unmet challenge. Herein, we describe a chiral PCNOx-pincer iridium complex for asymmetric transfer hydrogenation of this alkene class with ethanol, furnishing all-alkyl-substituted tertiary stereocenters. High levels of enantioselectivity can be achieved in the reactions of substrates with secondary/primary and primary/primary alkyl combinations. The catalyst is further applied to the redox isomerization of disubstituted alkenols, producing a tertiary stereocenter remote to the resulting carbonyl group. Mechanistic studies reveal a dihydride species, (PCNOx)Ir(H)2, as the catalytically active intermediate, which can decay to a dimeric species (κ3-PCNOx)IrH(µ-H)2IrH(κ2-PCNOx) via a ligand-remetalation pathway. The catalyst deactivation under the hydrogenation conditions with H2 is much faster than that under the transfer hydrogenation conditions with EtOH, which explains why the (PCNOx)Ir catalyst is effective for the transfer hydrogenation but ineffective for the hydrogenation. The suppression of di-to-trisubstituted alkene isomerization by regioselective 1,2-insertion is partly responsible for the success of this system, underscoring the critical role played by the pincer ligand in enantioselective transfer hydrogenation of 1,1-dialkylethenes. Moreover, computational studies elucidate the significant influence of the London dispersion interaction between the ligand and the substrate on enantioselectivity control, as illustrated by the complete reversal of stereochemistry through cyclohexyl-to-cyclopropyl group substitution in the alkene substrates.

Facilely Direct Construction, White-Light Emission, and Color-Adjustable Luminescence of LaF3 :Pr3+ @SiO2 Yolk-Shell Nanospheres with Moisture Resistance.

Poor water stability and single luminous color are the major drawbacks of the most phosphors reported. Therefore, it is important to realize multicolor luminescence in a phosphor with single host and single activator as well as moisture resistance. LaF3 :Pr3+ @SiO2 yolk-shell nanospheres are facilely obtained by a designing new technology of a simple and cost-effective electrospray ionization combined with a dicrucible fluorating technique without using protective gas. In addition, tunable photoluminescence, especially white-light emission, is successfully obtained in LaF3 :Pr3+ @SiO2 yolk-shell nanospheres by adjusting Pr3+ ion concentrations, and the luminescence mechanism of Pr3+ ion is advanced. Compared with the counterpart LaF3 :Pr3+ nanospheres, the water stability of LaF3 :Pr3+ @SiO2 yolk-shell nanospheres is improved by 15% after immersion in water for 72 h, and the fluorescence intensity can be maintained at 86% of the initial intensity. Furthermore, by treating the yolk-shell nanospheres with hydrofluoric acid, it is not only demonstrated that the shell-layer is SiO2 but also core-LaF3 :Pr3+ nanospheres are obtained. Particularly, only fluorination procedure among the halogenation can produce such special yolk-shell nanospheres, the formation mechanism of yolk-shell nanospheres is proposed detailedly based on the sound experiments and a corresponding new technology is built. These findings broaden practical applications of LaF3 :Pr3+ @SiO2 yolk-shell nanospheres.

Dual-Confinement Effect of Nanocages@Nanotubes Suppresses Polysulfide Shuttle Effect for High-Performance Lithium-Sulfur Batteries.

The shuttle effect of lithium polysulfides (LiPSs) severely hinders the development and commercialization of lithium-sulfur batteries, and the design of high-conductive carbon fiber-host material has become a key solution to suppress the shuttle effect. In this work, a unique Co/CoN-carbon nanocages@TiO2-carbon nanotubes structure (NC@TiO2-CNTs) is constructed using an electrospinning and nitriding process. Lithium-sulfur batteries using NC@TiO2-CNTs as cathode host materials exhibit high sulfur utilization (1527 mAh g-1 at 0.2 C) and can still maintain a discharge capacity of 663 mAh g-1 at a high current density of 5 C, and the capacity loss is only 0.056% per cycle during 500 cycles at 1 C. It is worth noting that even under extreme conditions (sulfur-loading = 90%, surface-loading = 5.0 mg cm-2 (S), and E/S = 6.63 µL mg-1), the lithium-sulfur batteries can still provide a reversible capacity of 4 mAh cm-2. Throughdensity functional theory calculations, it has been found that the Co/CoN heterostructures can adsorb and catalyze LiPSs conversion effectively. Simultaneously, the TiO2 can adsorb LiPSs and transfer Li+ selectively, achieving dual confinement for the shuttle effect of LiPSs (nanocages and nanotubes). The new findings provide a new performance enhancement strategy for the commercialization of lithium-sulfur batteries.

ADENet: a novel network-based inference method for prediction of drug adverse events.

Identification of adverse drug events (ADEs) is crucial to reduce human health risks and improve drug safety assessment. With an increasing number of biological and medical data, computational methods such as network-based methods were proposed for ADE prediction with high efficiency and low cost. However, previous network-based methods rely on the topological information of known drug-ADE networks, and hence cannot make predictions for novel compounds without any known ADE. In this study, we introduced chemical substructures to bridge the gap between the drug-ADE network and novel compounds, and developed a novel network-based method named ADENet, which can predict potential ADEs for not only drugs within the drug-ADE network, but also novel compounds outside the network. To show the performance of ADENet, we collected drug-ADE associations from a comprehensive database named MetaADEDB and constructed a series of network-based prediction models. These models obtained high area under the receiver operating characteristic curve values ranging from 0.871 to 0.947 in 10-fold cross-validation. The best model further showed high performance in external validation, which outperformed a previous network-based and a recent deep learning-based method. Using several approved drugs as case studies, we found that 32-54% of the predicted ADEs can be validated by the literature, indicating the practical value of ADENet. Moreover, ADENet is freely available at our web server named NetInfer (http://lmmd.ecust.edu.cn/netinfer). In summary, our method would provide a promising tool for ADE prediction and drug safety assessment in drug discovery and development.

Profiling prediction of nuclear receptor modulators with multi-task deep learning methods: toward the virtual screening.

Nuclear receptors (NRs) are ligand-activated transcription factors, which constitute one of the most important targets for drug discovery. Current computational strategies mainly focus on a single target, and the transfer of learned knowledge among NRs was not considered yet. Herein we proposed a novel computational framework named NR-Profiler for prediction of potential NR modulators with high affinity and specificity. First, we built a comprehensive NR data set including 42 684 interactions to connect 42 NRs and 31 033 compounds. Then, we used multi-task deep neural network and multi-task graph convolutional neural network architectures to construct multi-task multi-classification models. To improve the predictive capability and robustness, we built a consensus model with an area under the receiver operating characteristic curve (AUC) = 0.883. Compared with conventional machine learning and structure-based approaches, the consensus model showed better performance in external validation. Using this consensus model, we demonstrated the practical value of NR-Profiler in virtual screening for NRs. In addition, we designed a selectivity score to quantitatively measure the specificity of NR modulators. Finally, we developed a freely available standalone software for users to make profiling predictions for their compounds of interest. In summary, our NR-Profiler provides a useful tool for NR-profiling prediction and is expected to facilitate NR-based drug discovery.

Computational models, databases and tools for antibiotic combinations.

Antibiotic combination is a promising strategy to extend the lifetime of antibiotics and thereby combat antimicrobial resistance. However, screening for new antibiotic combinations is both time-consuming and labor-intensive. In recent years, an increasing number of researchers have used computational models to predict effective antibiotic combinations. In this review, we summarized existing computational models for antibiotic combinations and discussed the limitations and challenges of these models in detail. In addition, we also collected and summarized available data resources and tools for antibiotic combinations. This study aims to help computational biologists design more accurate and interpretable computational models.

In Silico Prediction of Skin Sensitization for Compounds via Flexible Evidence Combination Based on Machine Learning and Dempster-Shafer Theory.

Skin sensitization is increasingly becoming a significant concern in the development of drugs and cosmetics due to consumer safety and occupational health problems. In silico methods have emerged as alternatives to traditional in vivo animal testing due to ethical and economic considerations. In this study, machine learning methods were used to build quantitative structure-activity relationship (QSAR) models on five skin sensitization data sets (GPMT, LLNA, DPRA, KeratinoSens, and h-CLAT), achieving effective predictive accuracies (correct classification rates of 0.688-0.764 on test sets). To address the complex mechanisms of human skin sensitization, the Dempster-Shafer theory was applied to merge multiple QSAR models, resulting in an evidence-based integrated decision model. Various evidence combinations and combination rules were explored, with the self-defined Q3 rule showing superior balance. The combination of evidence such as GPMT and KeratinoSens and h-CLAT achieved a correct classification rate (CCR) of 0.880 and coverage of 0.893 while maintaining the competitiveness of other combinations. Additionally, the Shapley additive explanations (SHAP) method was used to interpret important features and substructures related to skin sensitization. A comparative analysis of an external human test set demonstrated the superior performance of the proposed method. Finally, to enhance accessibility, the workflow was implemented into a user-friendly software named HSkinSensDS.

AttentiveSkin: To Predict Skin Corrosion/Irritation Potentials of Chemicals via Explainable Machine Learning Methods.

Skin Corrosion/Irritation (Corr./Irrit.) has long been a health hazard in the Globally Harmonized System (GHS). Several in silico models have been built to predict Skin Corr./Irrit. as an alternative to the increasingly restricted animal testing. However, current studies are limited by data amount/quality and model availability. To address these issues, we compiled a traceable consensus GHS data set comprising 731 Corr., 1283 Irrit., and 1205 negative (Neg.) samples from 6 governmental databases and 2 external data sets. Then, a series of binary classifiers were developed with five machine learning (ML) algorithms and six molecular representations. For 10-fold cross-validation, the best Corr. vs Neg. classifier achieved an Area Under the Receiver Operating Characteristic Curve (AUC) of 97.1%, while the best Irrit. vs Neg. classifier achieved an AUC of 84.7%. Compared with existing in silico tools on external validation, our Attentive FP classifiers showed the highest metrics on Corr. vs Neg. and the second highest accuracy on Irrit. vs Neg. The SHapley Additive exPlanation approach was further applied to figure out important molecular features, and the attention weights were visualized to perform interpretable prediction. Structural alerts associated with Skin Corr./Irrit. were also identified. The interpretable Attentive FP classifiers were integrated into the software AttentiveSkin at https://github.com/BeeBeeWong/AttentiveSkin. The conventional ML classifiers are also provided on our platform admetSAR at http://lmmd.ecust.edu.cn/admetsar2/. Considering the data deficiency and the limited model availability of Skin Corr./Irrit., we believe that our data set and models could facilitate chemical safety assessment and relevant studies.

Prediction of Cytochrome P450 Substrates Using the Explainable Multitask Deep Learning Models.

Cytochromes P450 (P450s or CYPs) are the most important phase I metabolic enzymes in the human body and are responsible for metabolizing ∼75% of the clinically used drugs. P450-mediated metabolism is also closely associated with the formation of toxic metabolites and drug-drug interactions. Therefore, it is of high importance to predict if a compound is the substrate of a given P450 in the early stage of drug development. In this study, we built the multitask learning models to simultaneously predict the substrates of five major drug-metabolizing P450 enzymes, namely, CYP3A4, 2C9, 2C19, 2D6, and 1A2, based on the collected substrate data sets. Compared to the single-task model and conventional machine learning models, the multitask fingerprints and graph neural networks model achieved superior performance with the average AUC values of 90.8% on the test set. Notably, the multitask model demonstrated its good performance on the small amount of substrate data sets such as CYP1A2, 2C9, and 2C19. In addition, the Shapley additive explanation and the attention mechanism were used to reveal specific substructures associated with P450 substrates, which were further confirmed and complemented by the substructure mining tool and the literature.

In Silico Prediction of Chemical Acute Dermal Toxicity Using Explainable Machine Learning Methods.

The research on acute dermal toxicity has consistently been a crucial component in assessing the potential risks of human exposure to active ingredients in pesticides and related plant protection products. However, it is difficult to directly identify the acute dermal toxicity of potential compounds through animal experiments alone. In our study, we separately integrated 1735 experimental data based on rabbits and 1679 experimental data based on rats to construct acute dermal toxicity prediction models using machine learning and deep learning algorithms. The best models for the two animal species achieved AUC values of 78.0 and 82.0%, respectively, on 10-fold cross-validation. Additionally, we employed SARpy to extract structural alerts, and in conjunction with Shapley additive explanation and attentive FP heatmap, we identified important features and structural fragments associated with acute dermal toxicity. This approach offers valuable insights for the detection of positive compounds. Moreover, a standalone software tool was developed to make acute dermal toxicity prediction easier. In summary, our research would provide an effective tool for acute dermal toxicity evaluation of pesticides, cosmetics, and drug safety assessment.

Effects of grassland degradation on diversity of arbuscular mycorrhizal fungi of a pioneer plant.

Arbuscular mycorrhizal fungi (AMF) are obligate symbionts that engage in crucial interactions with plants, playing a vital role in grassland ecology. Our study focuses on the pioneer plant Agropyron cristatum, and we collected soil samples from four degraded grasslands in Yudaokou to investigate the response of community composition to the succession of degraded grasslands. We measured the vegetation status, soil physical and chemical properties, AMF colonization, and spore density in different degraded grasslands. High-throughput sequencing was employed to analyze AMF in soil samples. Correlations among community composition, soil characteristics, and plant factors were studied using principal component and regression analyses. The distribution of AMF in grasslands exhibited variation with different degrees of degradation, with Glomus, Scutellospora, and Diversispora being the dominant genera. The abundance of dominant genera in AMF also varied, showing a gradual increase in the relative abundance of the genus Diversispora with higher degradation levels. AMF diversity decreased from 27.7% to 12.4% throughout the degradation process. Among 180 samples of Agropyron cristatum plants, AMF hyphae and vesicles displayed the highest infection status in non-degraded grasslands and the lowest in severely degraded ones. Peak AMF spore production occurred in August, with maximum values in the 0-10-cm soil layer, and the highest spore densities were found in lightly degraded grasslands. Apart from pH, soil factors exhibited a positive correlation with AMF infection during grassland degradation. Furthermore, changes in AMF community composition were jointly driven by vegetation and soil characteristics, with vegetation coverage and soil organic carbon significantly impacting AMF distribution. Significant differences in AMF variables (spore number and diversity index) were also observed at different soil depths. Grassland successional degradation significantly influences AMF community structure and composition. Our future focus will be on understanding response mechanisms and implementing improvement methods for AMF during grassland degradation and subsequent restoration efforts.

Unraveling the Ligand-Binding Sites of CYP3A4 by Molecular Dynamics Simulations with Solvent Probes.

Cytochrome P450 3A4 (CYP3A4) is one of the most important drug-metabolizing enzymes in the human body and is well known for its complicated, atypical kinetic characteristics. The existence of multiple ligand-binding sites in CYP3A4 has been widely recognized as being capable of interfering with the active pocket through allosteric effects. The identification of ligand-binding sites other than the canonical active site above the heme is especially important for understanding the atypical kinetic characteristics of CYP3A4 and the intriguing association between the ligand and the receptor. In this study, we first employed mixed-solvent molecular dynamics (MixMD) simulations coupled with the online computational predictive tools to explore potential ligand-binding sites in CYP3A4. The MixMD approach demonstrates better performance in dealing with the receptor flexibility compared with other computational tools. From the sites identified by MixMD, we then picked out multiple sites for further exploration using ensemble docking and conventional molecular dynamics (cMD) simulations. Our results indicate that three extra sites are suitable for ligand binding in CYP3A4, including one experimentally confirmed site and two novel sites.

Network-Based Methods and Their Applications in Drug Discovery.

Drug discovery is time-consuming, expensive, and predominantly follows the "one drug → one target → one disease" paradigm. With the rapid development of systems biology and network pharmacology, a novel drug discovery paradigm, "multidrug → multitarget → multidisease", has emerged. This new holistic paradigm of drug discovery aligns well with the essence of networks, leading to the emergence of network-based methods in the field of drug discovery. In this Perspective, we initially introduce the concept and data sources of networks and highlight classical methodologies employed in network-based methods. Subsequently, we focus on the practical applications of network-based methods across various areas of drug discovery, such as target prediction, virtual screening, prediction of drug therapeutic effects or adverse drug events, and elucidation of molecular mechanisms. In addition, we provide representative web servers for researchers to use network-based methods in specific applications. Finally, we discuss several challenges of network-based methods and the directions for future development. In a word, network-based methods could serve as powerful tools to accelerate drug discovery.

NADPHnet: a novel strategy to predict compounds for regulation of NADPH metabolism via network-based methods.

Identification of compounds to modulate NADPH metabolism is crucial for understanding complex diseases and developing effective therapies. However, the complex nature of NADPH metabolism poses challenges in achieving this goal. In this study, we proposed a novel strategy named NADPHnet to predict key proteins and drug-target interactions related to NADPH metabolism via network-based methods. Different from traditional approaches only focusing on one single protein, NADPHnet could screen compounds to modulate NADPH metabolism from a comprehensive view. Specifically, NADPHnet identified key proteins involved in regulation of NADPH metabolism using network-based methods, and characterized the impact of natural products on NADPH metabolism using a combined score, NADPH-Score. NADPHnet demonstrated a broader applicability domain and improved accuracy in the external validation set. This approach was further employed along with molecular docking to identify 27 compounds from a natural product library, 6 of which exhibited concentration-dependent changes of cellular NADPH level within 100 µM, with Oxyberberine showing promising effects even at 10 µM. Mechanistic and pathological analyses of Oxyberberine suggest potential novel mechanisms to affect diabetes and cancer. Overall, NADPHnet offers a promising method for prediction of NADPH metabolism modulation and advances drug discovery for complex diseases.

Changes in the soil fungal communities of steppe grasslands at varying degradation levels in North China.

The grasslands in North China are rich in fungal resources. However, the knowledge of the structure and function of fungal communities and the role of microbial communities in vegetation restoration and succession are limited. Thus, we used an Illumina HiSeq PE250 high-throughput sequencing platform to study the changing characteristics of soil fungal communities in degraded grasslands, which were categorized as non-degraded (ND), lightly degraded, moderately degraded, and severely degraded (SD). Moreover, a correlation analysis between soil physical and chemical properties and fungal communities was completed. The results showed that the number of plant species, vegetation coverage, aboveground biomass, and diversity index decreased significantly with increasing degradation, and there were significant differences in the physical and chemical properties of the soil among the different degraded grasslands. The dominant fungal phyla in the degraded grassland were as follows: Ascomycota, 44.88%-65.03%; Basidiomycota, 12.68%-29.91%; and unclassified, 5.51%-16.91%. The dominant fungi were as follows: Mortierella, 6.50%-11.41%; Chaetomium, 6.71%-11.58%; others, 25.95%-36.14%; and unclassified, 25.56%-53.0%. There were significant differences in the microbial Shannon-Wiener and Chao1 indices between the ND and degraded meadows, and the composition and diversity of the soil fungal community differed significantly as the meadows continued to deteriorate. The results showed that pH was the most critical factor affecting soil microbial and fungal communities in SD grasslands, whereas soil microbial and fungal communities in ND grasslands were mainly affected by water content and other environmental factors.

Evaluation of machine learning models for cytochrome P450 3A4, 2D6, and 2C9 inhibition.

Cytochrome P450 (CYP) enzymes are involved in the metabolism of approximately 75% of marketed drugs. Inhibition of the major drug-metabolizing P450s could alter drug metabolism and lead to undesirable drug-drug interactions. Therefore, it is of great significance to explore the inhibition of P450s in drug discovery. Currently, machine learning including deep learning algorithms has been widely used for constructing in silico models for the prediction of P450 inhibition. These models exhibited varying predictive performance depending on the use of machine learning algorithms and molecular representations. This leads to the difficulty in the selection of appropriate models for practical use. In this study, we systematically evaluated the conventional machine learning and deep learning models for three major P450 enzymes, CYP3A4, CYP2D6, and CYP2C9 from several perspectives, such as algorithms, molecular representation, and data partitioning strategies. Our results showed that the XGBoost and CatBoost algorithms coupled with the combined fingerprint/physicochemical descriptor features exhibited the best performance with Area Under Curve (AUC)  of 0.92, while the deep learning models were generally inferior to the conventional machine learning models (average AUC reached 0.89) on the same test sets. We also found that data volume and sampling strategy had a minor effect on model performance. We anticipate that these results are helpful for the selection of molecular representations and machine learning/deep learning algorithms in the P450 model construction and the future model development of P450 inhibition.