Dataset Information

A comparative study on the molecular descriptors for predicting drug-likeness of small molecules.

ABSTRACT: Screening of " drug-like" molecule from the molecular database produced through high throughput techniques and their large repositories requires robust classification. In our work, a set of heuristically chosen nine molecular descriptors including four from Lipinski's rule, were used as classification parameter for screening "drug-like" molecules. The robustness of classification was compared with four fundamental descriptors of Lipinski. Back propagation neural network based classifier was applied on a database of 60000 molecules for classification of, " drug-like" and "non drug-like" molecules. Classification result using nine descriptors showed high classification accuracy of 96.1% in comparison to that using four Lipinski's descriptors which yielded an accuracy of 82.48%. Also a significant decrease of false positives resulted while using nine descriptors causing a sharp 18% increase of specificity of classification. From this study it appeared that Lipinski's descriptors which mainly deal with pharmacokinetic properties of molecules form the basis for identification of "drug-like" molecules that can be substantially improved by adding more descriptors representing pharmaco-dynamics properties of molecules.

SUBMITTER: Mishra H

PROVIDER: S-EPMC2728118 | biostudies-literature | 2009 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A comparative study on the molecular descriptors for predicting drug-likeness of small molecules.

Mishra Hrishikesh H Singh Nitya N Lahiri Tapobrata T Misra Krishna K

Bioinformation 20090613 9

Screening of " drug-like" molecule from the molecular database produced through high throughput techniques and their large repositories requires robust classification. In our work, a set of heuristically chosen nine molecular descriptors including four from Lipinski's rule, were used as classification parameter for screening "drug-like" molecules. The robustness of classification was compared with four fundamental descriptors of Lipinski. Back propagation neural network based classifier was appl ...[more]

PMID: 19707563

Similar Datasets

Project description:Computer aided toxicity and pharmacokinetic prediction studies attracted the attention of pharmaceutical industries as an alternative means to predict potential drug candidates. In the present study, in-silico pharmacokinetic properties (ADME), drug-likeness, toxicity profiles of sixteen antidiabetic flavonoids that have ideal bidentate chelating sites for metal ion coordination were examined using SwissADME, Pro Tox II, vNN and ADMETlab web tools. Density functional theory (DFT) calculations were also employed to calculate quantum chemical descriptors of the compounds. Molecular docking studies against human alpha amylase were also conducted. The results were compared with the control drugs, metformin and acarbose. The drug-likeness prediction results showed that all flavonoids, except myricetin, were found to obey Lipinski's rule of five for their drug like molecular nature. Pharmacokinetically, chrysin, wogonin, genistein, baicalein, and apigenin showed best absorption profile with human intestinal absorption (HIA) value of ≥ 30%, compared to the other flavonoids. Baicalein, butein, ellagic acid, eriodyctiol, Fisetin and quercetin were predicted to show carcinogenicity. The flavonoid derivatives considered in this study are predicted to be suitable molecules for CYP3A probes, except eriodyctiol which interacts with P-glycoprotein (p-gp). The toxicological endpoints prediction analysis showed that the median lethal dose (LD50) values range from 159-3919 mg/Kg, of which baicalein and quercetin are found to be mutagenic whereas butein is found to be the only immunotoxin. Molecular docking studies showed that the significant interaction (-7.5 to -8.3 kcal/mol) of the studied molecules in the binding pocket of the α-amylase protein relative to the control metformin with the crucial amino acids Asp 197, Glu 233, Asp 197, Glu 233, Trp 59, Tyr 62, His 101, Leu 162, Arg 195, His 299 and Leu 165. Chrysin was predicted to be a ligand with high absorption and lipophilicity with 84.6% absorption compared to metformin (78.3%). Moreover, quantum chemical, ADMET, drug-likeness and molecular docking profiles predicted that chrysin is a good bidentate ligand.

Project description:Various toxicity and pharmacokinetic evaluations as screening experiments are needed at the drug discovery stage. Currently, to reduce the use of animal experiments and developmental expenses, the development of high-performance predictive models based on quantitative structure-activity relationship analysis is desired. From these evaluation targets, we selected 50% lethal dose (LD50), blood-brain barrier penetration (BBBP), and the clearance (CL) pathway for this investigation and constructed predictive models for each target using 636-11,886 compounds. First, we constructed predictive models using the DeepSnap-deep learning (DL) method and images of compounds as features. The calculated area under the curve (AUC) and balanced accuracy (BAC) were, respectively, 0.887 and 0.818 for LD50, 0.893 and 0.824 for BBBP, and 0.883 and 0.763 for the CL pathway. Next, molecular descriptors (MDs) of compounds were calculated using Molecular Operating Environment, alvaDesc, and ADMET Predictor to construct predictive models using the MD-based method. Using these MDs, we constructed predictive models using DataRobot. The calculated AUC and BAC were, respectively, 0.931 and 0.805 for LD50, 0.919 and 0.849 for BBBP, and 0.900 and 0.807 for the CL pathway. In this investigation, we constructed predictive models combining the DeepSnap-DL and MD-based methods. In ensemble models using the mean predictive probability of the DeepSnap-DL and MD-based methods, the calculated AUC and BAC were, respectively, 0.942 and 0.842 for LD50, 0.936 and 0.853 for BBBP, and 0.908 and 0.832 for the CL pathway, with improved predictive performance observed for all variables compared with either single method alone. Moreover, in consensus models that adopted only compounds for which the results of the two methods agreed, the calculated BAC for LD50, BBBP, and the CL pathway were 0.916, 0.918, and 0.847, respectively, indicating higher predictive performance than the ensemble models for all three variables. The predictive models combining the DeepSnap-DL and MD-based methods displayed high predictive performance for LD50, BBBP, and the CL pathway. Therefore, the application of this approach to prediction targets in various drug discovery screenings is expected to accelerate drug discovery.

Dataset Information

A comparative study on the molecular descriptors for predicting drug-likeness of small molecules.

Publications

A comparative study on the molecular descriptors for predicting drug-likeness of small molecules.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets