Comparative transcriptomic analysis of the tea plant (Camellia sinensis) reveals key genes involved in pistil deletion.

BACKGROUND: The growth process of the tea plant (Camellia sinensis) includes vegetative growth and reproductive growth. The reproductive growth period is relatively long (approximately 1.5 years), during which a large number of nutrients are consumed, resulting in reduced tea yield and quality, accelerated aging, and shortened economic life of the tea plant. The formation of unisexual and sterile flowers can weaken the reproductive growth process of the tea plant. To further clarify the molecular mechanisms of pistil deletion in the tea plant, we investigated the transcriptome profiles in the pistil-deficient tea plant (CRQS), wild tea plant (WT), and cultivated tea plant (CT) by using RNA-Seq. RESULTS: A total of 3683 differentially expressed genes were observed between CRQS and WT flower buds, with 2064 upregulated and 1619 downregulated in the CRQS flower buds. These genes were mainly involved in the regulation of molecular function and biological processes. Ethylene synthesis-related ACC synthase genes were significantly upregulated and ACC oxidase genes were significantly downregulated. Further analysis revealed that one of the WIP transcription factors involved in ethylene synthesis was significantly upregulated. Moreover, AP1 and STK, genes related to flower development, were significantly upregulated and downregulated, respectively. CONCLUSIONS: The transcriptome analysis indicated that the formation of flower buds with pistil deletion is a complex biological process. Our study identified ethylene synthesis, transcription factor WIP, and A and D-class genes, which warrant further investigation to understand the cause of pistil deletion in flower bud formation.

Population sequencing enhances understanding of tea plant evolution.

Tea is an economically important plant characterized by a large genome, high heterozygosity, and high species diversity. In this study, we assemble a 3.26-Gb high-quality chromosome-scale genome for the 'Longjing 43' cultivar of Camellia sinensis var. sinensis. Genomic resequencing of 139 tea accessions from around the world is used to investigate the evolution and phylogenetic relationships of tea accessions. We find that hybridization has increased the heterozygosity and wide-ranging gene flow among tea populations with the spread of tea cultivation. Population genetic and transcriptomic analyses reveal that during domestication, selection for disease resistance and flavor in C. sinensis var. sinensis populations has been stronger than that in C. sinensis var. assamica populations. This study provides resources for marker-assisted breeding of tea and sets the foundation for further research on tea genetics and evolution.

Genomic Variance and Transcriptional Comparisons Reveal the Mechanisms of Leaf Color Affecting Palatability and Stressed Defense in Tea Plant.

Leaves are one of the most important organs of plants, and yet, the association between leaf color and consumable traits remains largely unclear. Tea leaves are an ideal study system with which to investigate the mechanism of how leaf coloration affects palatability, since tea is made from the leaves of the crop Camellia sinensis. Our genomic resequencing analysis of a tea cultivar ZiJuan (ZJ) with purple leaves and altered flavor revealed genetic variants when compared with the green-leaf, wild type cultivar YunKang(YK). RNA-Seq based transcriptomic comparisons of the bud and two youngest leaves in ZJ and YK identified 93%, 9% and 5% expressed genes that were shared in YK- and ZJ-specific cultivars, respectively. A comparison of both transcript abundance and particular metabolites revealed that the high expression of gene UFGT for anthocyanin biosynthesis is responsible for purple coloration, which competes with the intermediates for catechin-like flavanol biosynthesis. Genes with differential expression are enriched in response to stress, heat and defense, and are casually correlated with the environmental stress of ZJ plant origin in the Himalayas. In addition, the highly expressed C4H and LDOX genes for synthesizing flavanol precursors, ZJ-specific CLH1 for degrading chlorophyll, alternatively spliced C4H and FDR and low photosynthesis also contributed to the altered color and flavor of ZJ. Thus, our study provides a better molecular understanding of the effect of purple coloration on leaf flavor, and helps to guide future engineering improvement of palatability.

Structural optimization on a virtual screening hit of smoothened receptor.

The Hedgehog (Hh) pathway plays a critical role during embryonic development by controlling cell patterning, growth and migration. In adults, the function of Hh pathway is curtailed to tissue repair and maintenance. Aberrant reactivation of Hh signaling has been linked to tumorigenesis in various cancers, such as basal cell carcinoma (BCC) and medulloblastoma. The Smoothened (Smo) receptor, a key component of the Hh pathway which is central to the signaling transduction, has emerged as an attractive therapeutic target for the treatment of human cancers. Taking advantage of the availability of several crystal structures of Smo in complex with different antagonists, we have previously conducted a molecular docking-based virtual screening to identify several compounds which exhibited significant inhibitory activity against the Hh pathway activation (IC50 < 10 µM) in a Gli-responsive element (GRE) reporter gene assay. The most potent compound (ChemDiv ID C794-1677: 47 nM) showed comparable Hh signaling inhibition to the marketed drug vismodegib (46 nM). Herein, we report our structural optimization based on the virtual screening hit C794-1677. Our efforts are aimed to improve potency, decrease cLogP, and remove potentially metabolic labile/toxic pyrrole and aniline functionalities presented in C794-1677. The optimization led to the identification of numerous potent compounds exemplified by 25 (7.1 nM), which was 7 folds more potent compared with vismodegib. In addition, 25 was much less lipophilic compared with C794-1677 and devoid of the potentially metabolic labile/toxic pyrrole and aniline functional groups. Furthermore, 25 exhibited promising efficacy in inhibiting Gli1 mRNA expression in NIH3T3 cells with either wildtype Smo or D473H Smo mutant. These results represented significant improvement over the virtual screening hit C794-1677 and suggested that compound 25 can be used as a good starting point to support lead optimization.

Discovery of Novel and Selective Adenosine A2A Receptor Antagonists for Treating Parkinson's Disease through Comparative Structure-Based Virtual Screening.

Among non-dopaminergic strategies for combating Parkinson's disease (PD), antagonism of the A2A adenosine receptor (AR) has emerged to show great potential. In this study, on the basis of two crystal structures of the A2A AR with the best capability to distinguish known antagonists from decoys, docking-based virtual screening (VS) was conducted to identify novel A2A AR antagonists. A total of 63 structurally diverse compounds identified by VS were submitted to experimental testing, and 11 of them exhibited substantial activity against the A2A AR (Ki < 10 µM), including two compounds with Ki below 1 µM (compound 43, 0.42 µM; compound 51, 0.27 µM) and good A2A/A1 selectivity (fold < 0.1). Compounds 43 and 51 demonstrated antagonistic activity according to the results of cAMP measurements (cAMP IC50 = 1.67 and 1.80 µM, respectively) and showed good efficacy in the haloperidol-induced catalepsy (HIC) rat model for PD at doses of up to 30 mg/kg. Further lead optimization based on a substructure searching strategy led to the discovery of compound 84 as an excellent A2A AR antagonist (A2A Ki = 54 nM, A2A/A1 fold < 0.1, cAMP IC50 = 0.3 µM) that exhibited significant improvement in anti-PD efficacy in the HIC rat model.

In Silico Exploration for Novel Type-I Inhibitors of Tie-2/TEK: The Performance of Different Selection Strategy in Selecting Virtual Screening Candidates.

The receptor tyrosine kinase Tie-2 is involved in vessel remodeling and maturation, and has been regarded as a potential target for the treatment of various solid tumors. The absence of novel, potent and selective inhibitors severely hampers the understanding of the therapeutic potential of Tie-2. In the present work, we describe the discovery of novel type-I inhibitors of Tie-2 by structure-based virtual screening. Preliminary SAR was also performed based on one active compound, and several novel inhibitors with low micro-molar affinity were discovered. To directly compare the efficiency between different filtering strategies in selecting VS candidates, two methods were separately carried out to screen the same chemical library, and the selected VS candidates were then experimentally assessed by in vitro enzymatic assays. The results demonstrate that the hit rate is improved when stricter drug-likeness criteria and less number of molecules for clustering analysis are used, and meanwhile, the molecular diversity of the compounds still maintains. As a case study of TIE-2, the information presented in this work underscores the importance of selecting an appropriate selection strategy in VS campaign, and the novel inhibitors identified and the detailed binding modes of action provide a starting point for further hit-to-lead optimization process.

Constructing and Validating High-Performance MIEC-SVM Models in Virtual Screening for Kinases: A Better Way for Actives Discovery.

The MIEC-SVM approach, which combines molecular interaction energy components (MIEC) derived from free energy decomposition and support vector machine (SVM), has been found effective in capturing the energetic patterns of protein-peptide recognition. However, the performance of this approach in identifying small molecule inhibitors of drug targets has not been well assessed and validated by experiments. Thereafter, by combining different model construction protocols, the issues related to developing best MIEC-SVM models were firstly discussed upon three kinase targets (ABL, ALK, and BRAF). As for the investigated targets, the optimized MIEC-SVM models performed much better than the models based on the default SVM parameters and Autodock for the tested datasets. Then, the proposed strategy was utilized to screen the Specs database for discovering potential inhibitors of the ALK kinase. The experimental results showed that the optimized MIEC-SVM model, which identified 7 actives with IC50 < 10 µM from 50 purchased compounds (namely hit rate of 14%, and 4 in nM level) and performed much better than Autodock (3 actives with IC50 < 10 µM from 50 purchased compounds, namely hit rate of 6%, and 2 in nM level), suggesting that the proposed strategy is a powerful tool in structure-based virtual screening.

The application of in silico drug-likeness predictions in pharmaceutical research.

The concept of drug-likeness, established from the analyses of the physiochemical properties or/and structural features of existing small organic drugs or/and drug candidates, has been widely used to filter out compounds with undesirable properties, especially poor ADMET (absorption, distribution, metabolism, excretion, and toxicity) profiles. Here, we summarize various approaches for drug-likeness evaluations, including simple rules/filters based on molecular properties/structures and quantitative prediction models based on sophisticated machine learning methods, and provide a comprehensive review of recent advances in this field. Moreover, the strengths and weaknesses of these approaches are briefly outlined. Finally, the drug-likeness analyses of natural products and traditional Chinese medicines (TCM) are discussed.

Near-infrared dye bound albumin with separated imaging and therapy wavelength channels for imaging-guided photothermal therapy.

Development of theranostic agent for imaging-guided photothermal therapy has been of great interest in the field of nanomedicine. However, if fluorescent imaging and photothermal ablation are conducted with the same wavelength of light, the requirements of the agent's quantum yield (QY) for imaging and therapy are controversial. In this work, our synthesized near-infrared dye, IR825, is bound with human serum albumin (HSA), forming a HSA-IR825 complex with greatly enhanced fluorescence under 600 nm excitation by as much as 100 folds compared to that of free IR825, together with a rather high absorbance but low fluorescence QY at 808 nm. Since high QY that is required for fluorescence imaging would result in reduced photothermal conversion efficiency, the unique optical behavior of HSA-IR825 enables imaging and photothermal therapy at separated wavelengths both with optimized performances. We thus use HSA-IR825 for imaging-guided photothermal therapy in an animal tumor model. As revealed by in vivo fluorescence imaging, HSA-IR825 upon intravenous injection shows high tumor uptake likely owing to the enhanced permeability and retention effect, together with low levels of retentions in other organs. While HSA is an abundant protein in human serum, IR825 is able to be excreted by renal excretion as evidenced by high-performance liquid chromatography (HPLC). In vivo tumor treatment experiment is finally carried out with HSA-IR825, achieving 100% of tumor ablation in mice using a rather low dose of IR825. Our work presents a safe, simple, yet imageable photothermal nanoprobe, promising for future clinical translation in cancer treatment.

Discovery of novel inhibitors targeting the macrophage migration inhibitory factor via structure-based virtual screening and bioassays.

Macrophage migration inhibitory factor (MIF) is involved in regulation of both the innate and the adaptive immune responses and is regarded as an attractive anti-inflammatory pharmacological target. In this study, molecular docking-based virtual screening and in vitro bioassays were utilized to identify novel small-molecule inhibitors of MIF. The in vitro enzyme-based assay identified that ten chemically diverse compounds exhibited potent inhibitory activity against MIF in the micromolar regime, including three compounds with IC50 values below 10 µM and one with an IC50 value below 1 µM (0.55 µM); the latter is 26-fold more potent than the reference compound ISO-1. The structural analysis demonstrates that most of these active compounds possess novel structural scaffolds. Further in vitro cell-based glucocorticoid overriding, chemotaxis, and Western blotting assays revealed that the three compounds can effectively inhibit the biological functions of MIF in vitro, suggesting that these compounds could be potential agents for treating inflammatory diseases.

ADMET evaluation in drug discovery. 13. Development of in silico prediction models for P-glycoprotein substrates.

P-glycoprotein (P-gp) actively transports a wide variety of chemically diverse compounds out of cells. It is highly associated with the ADMET properties of drugs and drug candidates and, moreover, plays a major role in the multidrug resistance (MDR) phenomenon, which leads to the failure of chemotherapy in cancer treatments. Therefore, the recognition of potential P-gp substrates at the early stages of the drug discovery process is quite important. Here, we compiled an extensive data set containing 423 P-gp substrates and 399 nonsubstrates, which is the largest P-gp substrate/nonsubstrate data set yet published. Comparison of the distributions of eight important physicochemical properties for the substrates and nonsubstrates reveals that molecular weight and molecular solubility are the informative attributes differentiating P-gp substrates from nonsubstrates. Examination of the distributions of eight physicochemical properties for 735 P-gp inhibitors and 423 substrates gives the fact that inhibitors are significantly more hydrophobic than substrates while substrates tend to have more H-bond donors than inhibitors. Then, the classification models based on simple molecular properties, topological descriptors, and molecular fingerprints were developed using the naive Bayesian classification technique. The best naive Bayesian classifier yields a Matthews correlation coefficient of 0.824 and a prediction accuracy of 91.2% for the training set from a 5-fold cross-validation procedure, and a Matthews correlation coefficient of 0.667 and a prediction accuracy of 83.5% for the test set containing 200 molecules. Analysis of the important structural fragments given by the Bayesian classifier shows that the essential H-bond acceptors arranged in distinct spatial patterns and flexibility are quite essential for P-gp substrate-likeness, which affords a deeper understanding on the molecular basis of substrate/P-gp interaction. Finally, the reasons for mispredictions were discussed. It turns out that the presented classifier could be used as a reliable virtual screening tool for identifying potential substrates of P-gp.

Development and evaluation of an integrated virtual screening strategy by combining molecular docking and pharmacophore searching based on multiple protein structures.

In this study, we developed and evaluated a novel parallel virtual screening strategy by integrating molecular docking and complex-based pharmacophore searching based on multiple protein structures. First, the capacity of molecular docking or pharmacophore searching based on any single structure from nine crystallographic structures of Rho kinase 1 (ROCK1) to distinguish the known ROCK1 inhibitors from noninhibitors was evaluated systematically. Then, the naïve Bayesian classification or recursive partitioning technique was employed to integrate the predictions from molecular docking and complex-based pharmacophore searching based on multiple crystallographic structures of ROCK1, and the integrated protocol yields much better performance than molecular docking or complex-based pharmacophore searching based on any single ROCK1 structure. Finally, the well-validated integrated virtual screening protocol was applied to identify potential inhibitors of ROCK1 from traditional chinese medicine (TCM). The obtained potential active compounds from TCM are structurally novel and diverse compared with the known inhibitors of ROCK1, and they may afford valuable clues for the development of potent ROCK1 inhibitors.

Modeling compound-target interaction network of traditional Chinese medicines for type II diabetes mellitus: insight for polypharmacology and drug design.

In this study, in order to elucidate the action mechanism of traditional Chinese medicines (TCMs) that exhibit clinical efficacy for type II diabetes mellitus (T2DM), an integrated protocol that combines molecular docking and pharmacophore mapping was employed to find the potential inhibitors from TCM for the T2DM-related targets and establish the compound-target interaction network. First, the prediction capabilities of molecular docking and pharmacophore mapping to distinguish inhibitors from noninhibitors for the selected T2DM-related targets were evaluated. The results show that molecular docking or pharmacophore mapping can give satisfactory predictions for most targets but the validations are still quite necessary because the prediction accuracies of these two methods are variable across different targets. Then, the Bayesian classifiers by integrating the predictions from molecular docking and pharmacophore mapping were developed, and the well-validated Bayesian classifiers for 15 targets were utilized to find the potential inhibitors from TCM and establish the compound-target interaction network. The analysis of the compound-target network demonstrates that a small portion (18.6%) of the predicted inhibitors can interact with multitargets. The pharmacological activities for some potential inhibitors have been experimentally confirmed, highlighting the reliability of the Bayesian classifiers. Besides, it is interesting to find that a considerable number of the predicted multitarget inhibitors have free radical scavenging/antioxidant activities, which are closely related to T2DM. It appears that the pharmacological effect of the TCM formulas is determined not only by the compounds that interact directly with one or more T2DM-related targets, but also by the compounds with other supplementary bioactivities important for relieving T2DM, such as free radical scavenging/antioxidant effects. The mechanism uncovered by this study may offer a deep insight for understanding the theory of the classical TCM formulas for combating T2DM. Moreover, the predicted inhibitors for the T2DM-related targets may provide a good source to find new lead compounds against T2DM.

Platelet aggregation pathway network-based approach for evaluating compounds efficacy.

Traditional Chinese medicines (TCMs) contain a large quantity of compounds with multiple biological activities. By using multitargets docking and network analysis in the context of pathway network of platelet aggregation, we proposed network efficiency and network flux model to screen molecules which can be used as drugs for antiplatelet aggregation. Compared with traditional single-target screening methods, network efficiency and network flux take into account the influences which compounds exert on the whole pathway network. The activities of antiplatelet aggregation of 19 active ingredients separated from TCM and 14 nonglycoside compounds predicated from network efficiency and network flux model show good agreement with experimental results (correlation coefficient = 0.73 and 0.90, resp.). This model can be used to evaluate the potential bioactive compounds and thus bridges the gap between computation and clinical indicator.

Recent developments in computational prediction of HERG blockage.

The blockage of the voltage dependent ion channel encoded by human ether-a-go-go related gene (hERG) may lead to drug-induced QT interval prolongation, which is a critical side-effect of non-cardiovasular therapeutic agents. Therefore, identification of potential hERG channel blockers at the early stage of drug discovery process will decrease the risk of cardiotoxicity-related attritions in the later and more expensive development stage. Computational approaches provide economic and efficient ways to evaluate the hERG liability for large-scale compound libraries. In this review, the structure of the hERG channel is briefly outlined first. Then, the latest developments in the computational predictions of hERG channel blockers and the theoretical studies on modeling hERG-blocker interactions are summarized. Finally, the challenges of developing reliable prediction models of hERG blockers, as well as the strategies for surmounting these challenges, are discussed.

Discovery of Rho-kinase inhibitors by docking-based virtual screening.

Rho kinases (ROCK1 and ROCK2) belong to serine/threonine (Ser/Thr) protein kinase family, and play the central roles in the organization of the actin cytoskeleton. Therefore, Rho kinases have become attractive targets for the treatments of many diseases, such as cancer, renal disease, hypertension, ischemia, and stroke. In order to develop small-molecule inhibitors of ROCK1, molecular docking was utilized to virtually screen two chemical databases and identify molecules that interact with ROCK1. A small set of virtual hits was purchased and submitted to a series of experimental assays. The in vitro enzyme-based and cell-based assays reveal that 12 compounds have good inhibitory activity against ROCK1 in the micro molar regime (IC50 values between about 7 and 28 µM) and antitumor activity against lung cancer, breast cancer or/and myeloma cell lines. The structural analysis shows that two active compounds present novel scaffolds and are potential leads for the development of novel anti-cancer drugs. We then characterized the interaction patterns between ROCK1 and two inhibitors with novel scaffolds by molecular dynamics (MD) simulations and free energy decomposition analysis. In addition, the pharmacological effect of the two ROCK1 inhibitors with novel scaffolds on atorvastatin-induced cerebral hemorrhage was evaluated by using zebrafish model, and one compound candidate is able to prevent atorvastatin-induced cerebral hemorrhage effectively.

Drug-likeness analysis of traditional Chinese medicines: 2. Characterization of scaffold architectures for drug-like compounds, non-drug-like compounds, and natural compounds from traditional Chinese medicines.

BACKGROUND: In order to better understand the structural features of natural compounds from traditional Chinese medicines, the scaffold architectures of drug-like compounds in MACCS-II Drug Data Report (MDDR), non-drug-like compounds in Available Chemical Directory (ACD), and natural compounds in Traditional Chinese Medicine Compound Database (TCMCD) were explored and compared. RESULTS: First, the different scaffolds were extracted from ACD, MDDR and TCMCD by using three scaffold representations, including Murcko frameworks, Scaffold Tree, and ring systems with different complexity and side chains. Then, by examining the accumulative frequency of the scaffolds in each dataset, we observed that the Level 1 scaffolds of the Scaffold Tree offer advantages over the other scaffold architectures to represent the scaffold diversity of the compound libraries. By comparing the similarity of the scaffold architectures presented in MDDR, ACD and TCMCD, structural overlaps were observed not only between MDDR and TCMCD but also between MDDR and ACD. Finally, Tree Maps were used to cluster the Level 1 scaffolds of the Scaffold Tree and visualize the scaffold space of the three datasets. CONCLUSION: The analysis of the scaffold architectures of MDDR, ACD and TCMCD shows that, on average, drug-like molecules in MDDR have the highest diversity while natural compounds in TCMCD have the highest complexity. According to the Tree Maps, it can be observed that the Level 1 scaffolds present in MDDR have higher diversity than those presented in TCMCD and ACD. However, some representative scaffolds in MDDR with high frequency show structural similarities to those in TCMCD and ACD, suggesting that some scaffolds in TCMCD and ACD may be potentially drug-like fragments for fragment-based and de novo drug design.

Drug-likeness analysis of traditional Chinese medicines: 1. property distributions of drug-like compounds, non-drug-like compounds and natural compounds from traditional Chinese medicines.

BACKGROUND: In this work, we analyzed and compared the distribution profiles of a wide variety of molecular properties for three compound classes: drug-like compounds in MDL Drug Data Report (MDDR), non-drug-like compounds in Available Chemical Directory (ACD), and natural compounds in Traditional Chinese Medicine Compound Database (TCMCD). RESULTS: The comparison of the property distributions suggests that, when all compounds in MDDR, ACD and TCMCD with molecular weight lower than 600 were used, MDDR and ACD are substantially different while TCMCD is much more similar to MDDR than ACD. However, when the three subsets of ACD, MDDR and TCMCD with similar molecular weight distributions were examined, the distribution profiles of the representative physicochemical properties for MDDR and ACD do not differ significantly anymore, suggesting that after the dependence of molecular weight is removed drug-like and non-drug-like molecules cannot be effectively distinguished by simple property-based filters; however, the distribution profiles of several physicochemical properties for TCMCD are obviously different from those for MDDR and ACD. Then, the performance of each molecular property on predicting drug-likeness was evaluated. No single molecular property shows good performance to discriminate between drug-like and non-drug-like molecules. Compared with the other descriptors, fractional negative accessible surface area (FASA-) performs the best. Finally, a PCA-based scheme was used to visually characterize the spatial distributions of the three classes of compounds with similar molecular weight distributions. CONCLUSION: If FASA- was used as a drug-likeness filter, more than 80% molecules in TCMCD were predicted to be drug-like. Moreover, the principal component plots show that natural compounds in TCMCD have different and even more diverse distributions than either drug-like compounds in MDDR or non-drug-like compounds in ACD.

Drug-likeness analysis of traditional Chinese medicines: prediction of drug-likeness using machine learning approaches.

Quantitative or qualitative characterization of the drug-like features of known drugs may help medicinal and computational chemists to select higher quality drug leads from a huge pool of compounds and to improve the efficiency of drug design pipelines. For this purpose, the theoretical models for drug-likeness to discriminate between drug-like and non-drug-like based on molecular physicochemical properties and structural fingerprints were developed by using the naive Bayesian classification (NBC) and recursive partitioning (RP) techniques, and then the drug-likeness of the compounds from the Traditional Chinese Medicine Compound Database (TCMCD) was evaluated. First, the impact of molecular physicochemical properties and structural fingerprints on the prediction accuracy of drug-likeness was examined. We found that, compared with simple molecular properties, structural fingerprints were more essential for the accurate prediction of drug-likeness. Then, a variety of Bayesian classifiers were constructed by changing the ratio of drug-like to non-drug-like molecules and the size of the training set. The results indicate that the prediction accuracy of the Bayesian classifiers was closely related to the size and the degree of the balance of the training set. When a balanced training set was used, the best Bayesian classifier based on 21 physicochemical properties and the LCFP_6 fingerprint set yielded an overall leave-one-out (LOO) cross-validated accuracy of 91.4% for the 140,000 molecules in the training set and 90.9% for the 40,000 molecules in the test set. In addition, the RP classifiers with different maximum depth were constructed and compared with the Bayesian classifiers, and we found that the best Bayesian classifier outperformed the best RP model with respect to overall prediction accuracy. Moreover, the Bayesian classifier employing structural fingerprints highlights the important substructures favorable or unfavorable for drug-likeness, offering extra valuable information for getting high quality lead compounds in the early stage of the drug design/discovery process. Finally, the best Bayesian classifier was used to predict the drug-likeness of 33,961 compounds in TCMCD. Our calculations show that 59.37% of the molecules in TCMCD were identified as drug-like molecules, indicating that traditional Chinese medicines (TCMs) are therefore an excellent source of drug-like molecules. Furthermore, the important structural fingerprints in TCMCD were detected and analyzed. Considering that the pharmacology of TCMCD and MDDR (MDL Drug Data Report) was linked by the important common structural features, the potential pharmacology of the compounds in TCMCD may therefore be annotated by these important structural signatures identified from Bayesian analysis, which may be valuable to promote the development of TCMs.

ADME evaluation in drug discovery. 10. Predictions of P-glycoprotein inhibitors using recursive partitioning and naive Bayesian classification techniques.

P-Glycoprotein (P-gp), an efflux transporter, plays a crucial role in drug pharmacokinetic properties (ADME), and is critical for multidrug resistance (MDR) by mediating the active transport of anticancer drugs from the intracellular to the extracellular compartment. Here we reported an original database of 1273 molecules that are categorized into P-gp inhibitors and noninhibitors. The impact of various physicochemical properties on P-gp inhibition was examined. We then built the decision trees from a training set of 973 compounds using the recursive partitioning (RP) technique and validated by an external test set of 300 compounds. The best decision tree correctly predicted 83.5% of the inhibitors and 67.0% of the noninhibitors in the test set. Finally, we applied naive Bayesian categorization modeling to establish classifiers for P-gp inhibitors. The Bayesian classifier gave average correct prediction for 81.7% of 973 compounds in the training set with leave-one-out cross-validation procedure and 81.2% of 300 compounds in the test set. By establishing multiple decision trees and Bayesian classifiers, we evaluated the impact of molecular fingerprints on classification by the prediction accuracy for the test set, and we found that the inclusion of molecular fingerprints improves the prediction obviously. As an unsupervised learner without tuning parameters, the Bayesian classifier employing fingerprints highlights the important structural fragments favorable or unfavorable for P-gp transport, which provides critical information for designing new efficient P-gp inhibitors.