Drug researchers are turning to computational chemistry to more quickly learn about potential drug targets as they develop treatments for COVID-19 and other pandemics
By David Pittman
It can take up to a decade for a drug to move from the discovery phase all the way through being approved for marketing by the Food and Drug Administration. In the face of a global pandemic, that is far too long to wait for a potential therapy.
Computation chemistry and artificial intelligence, which have been long used to aide drug researchers in the quest to find potential treatments for novel aliments and diseases, has helped in the fight against SARS-CoV-2, more commonly referred to as coronavirus, or the virus that causes COVID-19.
As explained in a recent American Association of Pharmaceutical Scientists eChalk Talk, entitled Computational Studies of SARS-CoV-2: Application of Computational Chemistry and A.I. Algorithms for Pandemic Response, modern computing science has been used to find drugs that can be repurposed to treat the novel coronavirus and to find potential drug targets so that therapies could be more quickly tested in clinic. Modern computing power is also being applied to better understand how the virus mutates during transmission, which effects how a potential vaccine could work.
“If there’s one thing that we can learn from the pandemic, it’s that we need to find a way to accelerate and improve the success rate of the drug development process in general,” Mike Bellucci, Ph.D., senior director of solid form design at XtalPi Inc. “One possible way that we believe that we can accomplish this is to leverage computation, A.I., and experiments to help guide experiments and make this process more efficient.”
Cambridge, Mass.-based XtalPi develops the algorithms that are used in a cloud-based platform that integrates computation chemistry and A.I. in both the drug discovery and preclinical, solid-state development phases. Because it is cloud-based and relies on computing clusters from across the globe, the platform can be quickly scaled to meet high-demand circumstances, such as pandemic response. Because XtalPi can assemble in under an hour 1 million cores, which do the actual calculations, analyses can also be turned around quickly.
XtalPi has been using its computation chemistry and A.I. platform to screen for drugs that can possibly be repurposed to retreat COVID-19, to design and help develop drugs with novel inhibitors, and to study mutation effects on COVID-19’s transmission. The company started many of its pandemic-related research studies in January, when less was known about the virus and how it transmitted.
XtalPi ran potential COVID-19 targets against 3,000 FDA-approved drugs and 10,000 natural compounds to see what a good fit for a potential treatment could be. The company found 96 marketed drugs and 87 natural compounds that could be good candidates based on the known genomic sequences of COVID-19.
And repurposing is working. One of the first compounds that drug researchers pinpointed as a potential therapy for COVID-19 was remdesivir, a broad-spectrum anti-viral medication. Remdesivir was first used to treat Ebola, which first arrived in 2014, but computational methods showed it could be repurposed as a potential COVID-19 therapy.
Drug researchers discovered from genomic sequencing that COVID-19 and the original SARS virus, which first appeared in 2003 to infect more than 8,000 people in 26 countries, share an 82 percent sequence identity overall. This suggested that drugs that treated the original SARS virus well could be applied to COVID-19. Furthermore, a repurposed SARS drug target that disrupts RNA-dependent RNA polymerase could be successful since SARS and COVID-19 share a 96 percent sequence identity.
“Drug repurposing is a viable strategy for immediate pandemic response, but for more successful treatments, I think it’s fair to say that there is a new, broad-spectrum for anti-viral drugs, specifically for viruses that have a high risk of transmission to humans,” Bellucci said.
To better develop those broad-spectrum drugs, Bellucci suggested leveraging the power of computational chemistry and A.I., which he explained during his talk. Libraries of molecules can be assembled based on various factors such as solubility and synthesizability. Machine learning can then be used to screen for different characteristics such as their likelihood to bind to certain receptors.
“Generally, when we do this, we do this in collaboration with experimentalists, so there’s a good amount of feedback in this loop where we essentially narrow down to a set of compounds,” Bellucci said. “Those are then synthesized and tested.”
Feedback from experimentalists is added back into the computational work to create better, more informed models.
With an eye on future pandemics, XtalPi is studying new, broad-spectrum anti-virals, but that work remains ongoing. Computation work, of course, needs to be combined with clinical experiments to translate the theoretical work to patient care. That works takes finding additional partners and collaborators. “This strategy will unfortunately not likely impact the current pandemic, but it could help mitigate future pandemics, which will inevitably happen,” Bellucci said.
Bellucci lastly explained XtalPi’s work using computation chemistry and artificial intelligence to understand how the COVID-19 virus mutates during transmission between different hosts and therefore how well potential vaccines might work over time.
Using crystal structures, drug researchers can pinpoint changes to the proteins that bind the virus with host cells. Then, computational methods can help understand the degree to which mutations over time can make a virus more or less likely to be transmitted.
“Currently, we’re using this information to try to understand how different mutations will in the future impact the viral transmission and also how these can potentially stabilize spike proteins’ use in vaccine therapeutics, specifically how to stabilize the protein to present to the immune system and perhaps to develop a vaccine,” Bellucci said.
To listen to the webinar in full, you may access the archive . Also, learn about upcoming webinars or to submit a proposal for a new webinar.
_____________________
David Pittman is a science and medical writer based in Washington, D.C.