Specifically, for a virtual screen of 50,000~200,000 compounds, how many binding modes (poses) per compound do you usually ask VINA to generate? And how do you analyze these results? Visually inspect each pose individually, e.g. in PyMOL? Just pick the top pose for each compound?
Feel free to share any other tips for analyzing VINA results. Thanks!
Dear Mr. Fong,
Autodock Vina usually provides nine binding modes (poses) per compound (although it varies). The docked poses can be visualized by using the PyMOL software, is a good choice.
You may follow these articles for better understanding.
https://www.researchgate.net/publication/319993784_Integrating_regression_and_classification-based_QSARs_with_molecular_docking_analyses_to_explore_the_structure-antiaromatase_activity_relationships_of_letrozole-based_analogs
https://www.researchgate.net/publication/317750016_Possible_Binding_Mode_Analysis_of_Pyrazolo-triazole_Hybrids_as_Potential_Anticancer_Agents_through_Validated_Molecular_Docking_and_3D-QSAR_Modeling_Approaches
https://www.researchgate.net/publication/303202679_Insight_into_the_Structural_Requirements_of_Theophylline-Based_Aldehyde_Dehydrogenase_lAl_ALDHlAl_Inhibitors_Through_Multi-QSAR_Modeling_and_Molecular_Docking_Approaches
However, it is quite hard to use Autodock Vina for a virtual screen of 50,000~200,000 compounds. I recommend you to use other softwares like GOLD, GLIDE, etc for high-throughput docking. Let me know if you have any questions.
Thanking you,
Regards,
Sk. Abdul Amin
Jadavpur University, India.
Hi, autodock vina generate only possible 09 binding poses. Practically it is impossible to virtually screen 50,000 to 2,00,000 lead molecules against single target molecule at a time using autodock vina. You can visualize ligand-receptor interaction (binding poses) using pymol. So, you can use pyrx or dock blaster server for VS.
Filter the small molecule libraries using ADME/Lipinsky properties. So that false hits were reduced. Then you do docking with Autodock. There are few other commercial docking softwares available. Though, Auto-dock is the best for protein-ligand docking. You can have options to change the default settings.
After docking you can visualise the protein-ligand docked complex using Pymol/Chimera. Ligplot can be used to plot the interactions.
As an alternative to pure visual inspection (which is a little primitive and not particularly well suited for a beginner, who might neglect many relevant interactions), I would like to recommend the Arpeggio webserver.
It is free, fast, and less prone to human error or bias:
http://biosig.unimelb.edu.au/arpeggioweb/
Of course, many other sophisticated codes offer interaction analysis possibilities.
Dear Fong,
The visual examination of each database dock pose from a Vina-guided virtual screen is beyond human subjectivity. Although genetic algorithms are better pose generators but the docking score computed by its empirical scoring function is cumbersome. I developed a re-scoring method PLHINT which works on the dock poses of the database molecules. The method has overcome poor enrichment by Vina with better performance. The method just need the H-bonds and it can be calculated easily with open source programs. The details can be found here: https://www.sciencedirect.com/science/article/pii/S1093326317306356