Combining Machine Learning Systems and Multiple Docking Simulation Packages to Improve Docking Prediction Reliability for Network Pharmacology

Hsin, Kun‐Yi; Ghosh, Samik; Kitano, Hiroaki

doi:10.1371/journal.pone.0083922

Cited by 331 publications

(224 citation statements)

References 43 publications

Supporting

Mentioning

224

Contrasting

Order By: Relevance

“…Overall, machine‐learning SFs have exhibited a substantial improvement over classical SFs in different binding affinity prediction benchmarks . Furthermore, a number of studies have shown that a classical SF can easily be improved by substituting their linear regression model with nonparametric machine‐learning regression, either using RF or SVR .…”

Section: Generic Machine‐learning Sfs To Predict Binding Affinitymentioning

confidence: 99%

“…It is noteworthy that this SVR model was trained on the docking poses of a set of known inhibitors, as crystal structures for these ligands were not available. This strategy has also been successfully employed with other SVR models and RF . Given these successes and the very large number of actives that are now known for a range of targets, training on the docking poses of these molecules should strongly increase the size of training sets and thus the performance of machine‐learning SFs.…”

Section: Conclusion and Future Prospectsmentioning

confidence: 99%

“…Overall, machine-learning SFs have exhibited a substantial improvement over classical SFs in different binding affinity prediction benchmarks. 20,57,58,[64][65][66] Furthermore, a number of studies have shown that a classical SF can easily be improved by substituting their linear regression model with nonparametric machinelearning regression, either using RF 20,63,66 or SVR. 67 On the other hand, Ashtawy and Mahatrapa presented 68 the first comprehensive assessment of machine-learning SFs for binding affinity prediction, which led to training and testing 42 SFs on each training-test data partition (six regression techniques and seven set of features).…”

Section: Generic Machine-learning Sfs To Predict Binding Affinitymentioning

confidence: 99%

See 2 more Smart Citations

Machine‐learning scoring functions to improve structure‐based binding affinity prediction and virtual screening

Ain

Aleksandrova

Roessler

et al. 2015

WIREs Comput Mol Sci

281

234

View full text Add to dashboard Cite

Docking tools to predict whether and how a small molecule binds to a target can be applied if a structural model of such target is available. The reliability of docking depends, however, on the accuracy of the adopted scoring function (SF). Despite intense research over the years, improving the accuracy of SFs for structure‐based binding affinity prediction or virtual screening has proven to be a challenging task for any class of method. New SFs based on modern machine‐learning regression models, which do not impose a predetermined functional form and thus are able to exploit effectively much larger amounts of experimental data, have recently been introduced. These machine‐learning SFs have been shown to outperform a wide range of classical SFs at both binding affinity prediction and virtual screening. The emerging picture from these studies is that the classical approach of using linear regression with a small number of expert‐selected structural features can be strongly improved by a machine‐learning approach based on nonlinear regression allied with comprehensive data‐driven feature selection. Furthermore, the performance of classical SFs does not grow with larger training datasets and hence this performance gap is expected to widen as more training data becomes available in the future. Other topics covered in this review include predicting the reliability of a SF on a particular target class, generating synthetic data to improve predictive performance and modeling guidelines for SF development. WIREs Comput Mol Sci 2015, 5:405–424. doi: 10.1002/wcms.1225For further resources related to this article, please visit the WIREs website.

show abstract

Section: Generic Machine‐learning Sfs To Predict Binding Affinitymentioning

confidence: 99%

Section: Conclusion and Future Prospectsmentioning

confidence: 99%

Section: Generic Machine-learning Sfs To Predict Binding Affinitymentioning

confidence: 99%

See 1 more Smart Citation

Machine‐learning scoring functions to improve structure‐based binding affinity prediction and virtual screening

Ain

Aleksandrova

Roessler

et al. 2015

WIREs Comput Mol Sci

281

234

View full text Add to dashboard Cite

show abstract

“…In spite of these issues, by focusing on the structural features, these docking simulations were able to correctly predict binding modes and the trends in kinase inhibition. Future efforts should focus on three fronts: a) Developing a receptor-based pharmacophore with spatial constraints to include receptor residues as exclusion areas to improve the estimation of binding affinities; b) Inclusion of the impact of conformational changes of key residues in the active site on the estimation of inhibition constant; c) A combined approach of using analytic learning methods along with docking simulations to model cross-reactivity and toxicity [73]. …”

Section: Resultsmentioning

confidence: 99%

Can structural features of kinase receptors provide clues on selectivity and inhibition? A molecular modeling study

Ravichandran

Luke

Collins

2015

Journal of Molecular Graphics and Modelling

View full text Add to dashboard Cite

Cancer is a complex disease resulting from the uncontrolled proliferation of cell signaling events. Protein kinases have been identified as central molecules that participate overwhelmingly in oncogenic events, thus becoming key targets for anticancer drugs. A majority of studies converged on the idea that ligand-binding pockets of kinases retain clues to the inhibiting abilities and cross-reacting tendencies of inhibitor drugs. Even though these ideas are critical for drug discovery, validating them using experiments is not only difficult, but in some cases infeasible. To overcome these limitations and to test these ideas at the molecular level, we present here the results of receptor-focused in-silico docking of nine marketed drugs to 19 different wild-type and mutated kinases chosen from a wide range of families. This investigation highlights the need for using relevant models to explain the correct inhibition trends and the results are used to make predictions that might be able to influence future experiments. Our simulation studies are able to correctly predict the primary targets for each drug studied in majority of cases and our results agree with the existing findings. Our study shows that the conformations a given receptor acquires during kinase activation, and their micro-environment, defines the ligand partners. Type II drugs display high compatibility and selectivity for DFG-out kinase conformations. On the other hand Type I drugs are less selective and show binding preferences for both the open and closed forms of selected kinases. Using this receptor-focused approach, it is possible to capture the observed fold change in binding affinities between the wild-type and disease-centric mutations in ABL kinase for Imatinib and the second-generation ABL drugs. The effects of mutation are also investigated for two other systems, EGFR and B-Raf. Finally, by including pathway information in the design it is possible to model kinase inhibitors with potentially fewer side-effects.

show abstract

“…Based on systems biology and pharmacology, network pharmacology leverages network analysis and screens nodes to identify proteins critical to a disease for drug design [13, 14]. Network pharmacology can be used to investigate synergism of multicomponent drugs to identify high efficacy and low toxicity agents with multiple targets [15, 16].…”

Section: Introductionmentioning

confidence: 99%

Bufei Huoxue Capsule Attenuates PM2.5‐Induced Pulmonary Inflammation in Mice

Yue

Zhang

Cai

et al. 2017

Evidence-Based Complementary and Alternative Medicine

View full text Add to dashboard Cite

Atmospheric fine particulate matter 2.5 (PM 2.5) may carry many toxic substances on its surface and this may pose a public health threat. Epidemiological research indicates that cumulative ambient PM2.5 is correlated to morbidity and mortality due to pulmonary and cardiovascular diseases and cancer. Mitigating the toxic effects of PM2.5 is therefore highly desired. Bufei Huoxue (BFHX) capsules have been used in China to treat pulmonary heart disease (cor pulmonale). Thus, we assessed the effects of BFHX capsules on PM2.5-induced pulmonary inflammation and the underlying mechanisms of action. Using Polysearch and Cytoscape 3.2.1 software, pharmacological targets of BFHX capsules in atmospheric PM2.5-related respiratory disorders were predicted and found to be related to biological pathways of inflammation and immune function. In a mouse model of PM2.5-induced inflammation established with intranasal instillation of PM2.5 suspension, BFHX significantly reduced pathological response and inflammatory mediators including IL-4, IL-6, IL-10, IL-8, TNF-α, and IL-1β. BFHX also reduced keratinocyte growth factor (KGF), secretory immunoglobulin A (sIgA), and collagen fibers deposition in lung and improved lung function. Thus, BFHX reduced pathological responses induced by PM2.5, possibly via regulation of inflammatory mediators in mouse lungs.

show abstract

Combining Machine Learning Systems and Multiple Docking Simulation Packages to Improve Docking Prediction Reliability for Network Pharmacology

Cited by 331 publications

References 43 publications

Machine‐learning scoring functions to improve structure‐based binding affinity prediction and virtual screening

Machine‐learning scoring functions to improve structure‐based binding affinity prediction and virtual screening

Can structural features of kinase receptors provide clues on selectivity and inhibition? A molecular modeling study

Bufei Huoxue Capsule Attenuates PM2.5‐Induced Pulmonary Inflammation in Mice

Contact Info

Product

Resources

About