Undoubtedly, the latest “Covid-19” coronavirus strain has caught us by surprise, as the development of an effective anti-viral drug as well as a vaccine is still underway. Insilico Medicine, an American startup, is joining the effort to develop an effective treatment against this novel virus by deploying a pioneering Artificial Intelligence (A.I.) system. Their A.I. can identify thousands of new molecules, synthesize the most promising ones, and test them against the Covid-19 virus.
Sure, all of this mixing and matching happens on a theoretical level, but it is still very useful in the hands of researchers who can exclude a large number of possible drug compositions and only focus only on those that have higher chances to be effective. Even if Gilead’s “Remdesivir” proves to be the answer to the coronavirus after all, Insilico is in the game to see if A.I. systems can really help develop drugs to hinder pandemics in just a couple of weeks.
Insilico lost precious time in the beginning, as they didn’t realize how serious Covid-19 was when the outbreak in Wuhan happened. They only decided to help on January 28, 2020, a whole month after the initial infections. Quickly, they set their A.I. to find ways to target the 3C-like protease enzyme that Covid-19 uses for its reproduction, and to inhibit its functionality. They set up 28 individual machine learning models to work on this purpose, and by day four, they already had thousands of promising enough molecule candidates to test on the lab.
Now, the company has selected 100 molecules to be synthesized and tested, so they have openly shared them with researchers and labs so as to expedite the discovery of an effective treatment. In the meantime, another team from the Michigan State University is also using machine learning technologies to help find promising cures for Covid-19, taking into account the molecule’s synthesizability, effectiveness, solubility, and its bonding affinity.