I'm really hoping that ZINC04777075 can be used in combination therapy with the standard medicine against the Brazilian strain of Leishmaniasis.

If I understand the chart correctly, it's showing thirty three percent inhibition at a low toxicity, whereas the current treatment is very toxic. Even a small decrease in Amphotericin B dosage would probably be much easier on the paitent, right?

Here's hoping for better news from the next set of compounds tested.
And if there's little else we can do, at least now we know for sure.

Crunch on!

Hi Jim1348,

it's an excellent question you are asking here. Long story short: yes, docking is still inaccurate, but it's still a very useful tool for finding new active molecules (see [1] at the bottom of the page).
I'll try to give you an overview of the current scenario of drug design and docking, I hope it will be useful to clarify a bit some of the aspects.

The issue of the accuracy is a constant subject of discussions and analyses in the field.
While many aspects are debated, a key point is widely accepted: with screenings, either virtual or experimental, we don't expect to find a drug, but a molecule with the minimum activity we can detect with a biological assay, commonly called a "hit". Hits, then can then be chemically modified increase their activity, turning them into "lead compounds". Although, in this process, other factors beside activity, such as the toxicity, or the capability of the molecules to be delivered in cells and tissues, determine their further development into drug candidates. For example, a molecule that kills an infection agent, a parasite, or a cancer cell is not particularly useful, if in the process it kills also half of the remaining cells or healthy organisms.

The path from a hit to an actual treatment usually involves 20+ years of research, several million dollars, and a good deal of trial and error. This kind of resources is usually available only to pharmaceutical companies. Even then, the failure rate is very high, and only one out of ten or more projects will deliver an actual drug to the market (imagine the development costs involved in that).

That said, for the more specific question about AutoDock Vina (or any predictive computational tool, for that matter), you are absolutely right, and there is a lot of room for improvement. The accuracy of docking results depends essentially on two factors: the complexity of the physics describing the event we're trying to simulate, and the 'shortcuts' we take for speeding up the calculation.

On one hand, the binding of two molecules is an intrinsically complex problem. The overall role of the forces involved is known, but there is still a lot of research trying assessing how the relative contribution of each of these forces changes depending on the molecular complexes considered.

On the other hand, if we would like to use the highest level of theory we have to date to simulate these interactions, we would spend months of computations just to evaluate a single molecule, even with all the WCG computing power.

As a consequence, the current status of molecular modeling in general (and docking in particular) is a compromise between speed and accuracy. And yet, despite all the issues mentioned above, computational methods are still an excellent way to identify new "hit" molecules [1], and able to compete with experimental methods. The unprecedented large scale virtual screenings made possible by WCG helped identifying unexpected problems and limitations, which where not apparent in our calculations on our local computational resources.

On our end, we are working hard to try improving both the physics and the search methods in order to increase the accuracy of the predictions with similar computational costs. Not an easy task, when even a single water molecule can make a difference between a correct result and a wrong one!

We are incredibly thankful to all you volunteers that donate your computing power to research, and IBM that makes possible to us to embark in such ambitious docking campaigns.

I hope I didn't scare you away with all these details.
Thank you for your interest and your support,

Stefano


[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4865415/