As the vaccination process advances in most countries, just as we have been in the coronavirus pandemic for more than a year, it seems more and more likely to achieve the long-awaited herd immunity , which would help reduce or stop the spread of SARS- CoV-2 (the virus that causes COVID-19).
However, scientists continue to search for effective medical treatments for people who continue to be ill from the new coronavirus, especially when the question has arisen as to what would happen if the expected herd immunity were not fully achieved, or if the virus, at mutating leads to hypervirulent variants that end up diminishing the benefits of vaccination. And, although some existing treatments show many benefits, there is an urgent need to find new therapies.
In a scientific article published a few days ago in the specialized journal ‘Nature Machine Intelligence’, researchers from the University of New Mexico have presented a set of online, open source computational models that would help experts quickly detect small molecules in Search for possible properties to treat and combat COVID , baptized with the name REDIAL-2020 .
According to experts at the National Center for the Advancement of Translational Sciences (NCATS), this could help replace laboratory experiments, thanks to the fact that today there is so much data available that scientists can build robust machine learning models.
The results of the tests carried out by NCTAS scientists measured the ability of each molecule to inhibit viral entry, infectivity and reproduction, as well as the cytopathic effect, the ability of a cell to protect itself from death caused by the action of the virus.
Unlike what biomedical researchers tend to do, who tend to focus mostly on the positive findings of their studies, in this case the scientists also reported the presence of molecules that had no effect on fighting viruses. However, including negative data would actually help improve the accuracy of machine learning .
As experts say, “the coronavirus is a cunning adversary”, but they hope to design a cocktail of multiple drugs that are capable of attacking the virus on multiple fronts.
What is REDIAL-2020 and how does it work?
The set of computational models designed by the experts is based on machine learning algorithms capable of rapidly processing large amounts of data, in addition to detecting hidden patterns that a human researcher might not perceive.
Subsequently, the team of scientists validated the machine learning predictions from data from the National Center for the Advancement of Translational Sciences, comparing them with the known effects of drugs already tested in the DrugCentral database of the University of New Mexico.
One of its main advantages is the flexibility of the computational workflow, so it could be trained to evaluate compounds against other pathogens, in addition to evaluating chemicals that have not yet been tested for human use.
And it is that, although the experts pursue the objective of reusing drugs, they are also focusing on any other small molecule. And it wouldn’t have to be an approved drug; In short, “anyone who tests your molecule could find something important.”
It would therefore be a unique and extremely useful and interesting tool for identifying molecules capable of attacking the virus before it invades human cells . Or even disable or “disable” it in the early stages of infection.
Reference: Govinda B. KC et al, A machine learning platform to estimate anti-SARS-CoV-2 activities, Nature Machine Intelligence (2021). DOI: 10.1038 / s42256-021-00335-w.
