The H3ABioNet Seminar co-ordinating team on behalf of the H3ABioNet Research Working Group cordially invite you to join us for the July 2017 H3ABioNet seminar under the theme of: “Structural Bioinformatics”.

Seminar Format:  A seminar talk will be provided by Dr. Pamela Greenwell, from the Centre for Parallel Computing in the University of Westminster, London. The talk will be 40 to 50 minutes long followed by 10 to 15-minutes of discussion and Q/A.
 

Seminar Date: Thursday 20th July 2017
Seminar time: 1pm UTC / 2pm WAT / 3pm CAT / 4pm EAT

URL to join the seminar:  https://mconf.sanren.ac.za/webconf/h3abionet_seminars

This webinar explores the tools that can be used to investigate the binding of small molecules to proteins to identify novel drugs or potential off-target drug actions. Using Autodock 4, Autodock Vina, Racoon and Gromacs on computing grids, clusters and the Cloud, we have set up a system in which we can simulate either interactions of a single protein with thousands of potential small molecules to identify novel drugs or of multiple proteins with a single drug to determine off-target effects.
Our specific interest is investigating proteins whose gene sequences and biological activities are known but for which there are no crystal/NMR based structures available.  We will use as an example proteins of the protozoan Trichomonas vaginalis (TV), a pathogen that infects more than 200 million women worldwide and whose presence is associated with increased risk of spread of HIV. There is currently only one drug, metronidazole, that is used to treat the infection and hence new drugs are required. The genome of this parasite has been known for more than a decade but little has been done to investigate the structure of the encoded proteins.  However, for many proteins much is known about their biological activity.
The predicted proteins have been identified using BLASTp and show homology in the “twilight zone”; such proteins are not recommended for homology modelling. Nevertheless,  faced with a situation where we had no other option, we used homology modelling  to prepare 3D structures and validated these in docking experiments using known ligands (substrates and inhibitors) and biochemical analyses. Using Autodock 4 and Autodock Vina for docking and Gromacs for energy minimisation we developed Prosim; a workflow with a user-friendly interface that ran on the University grid of 2000 computers. From the first simulations, we were able to demonstrate the utility of the techniques using the sialidase enzyme of the organism.  This is an excellent model as the biological activity of the enzyme is easy to assess, there is data as to the substrate specificity and known inhibitors are available.  The enzyme is involved in the cytolysis seen in the pathogenesis of the disease, there is only one gene candidate, and inhibition of a related enzyme is used in the treatment of influenza. However, the known sugar-based inhibitors would be too expensive to contemplate for use as a drug and the drugs designed for the influenza enzyme are not good inhibitors of the TV protein. We have demonstrated that the Autodock results support the biological data available on the binding of known substrates and inhibitors to the model and progressed the work to successfully identify novel inhibitors using two libraries: one of small molecules and the other of known drugs.
The results to be presented will demonstrate the utility of this technique in identifying both novel drugs and potential off-target drug effects for organisms  where little is known in terms of  3D protein  structure and for which predicted proteins have little homology with other proteins in databases.