Funds aimed at establishing new ways to fight viral infections
Charles River Laboratories in Massachusetts will be conducting the experiments and sending the data back to ViQi to analyze and detect if their AI can differentiate between the groups of images sent.
The fight against COVID-19 is one that will continue years to come.
Even if a capable vaccine or competent medications to treat it are found, the novel coronavirus is a grim reminder that viruses are always threats to mutate and ruin the current way of life.
One company hoping to play a role in combating those future threats is right here in Santa Barbara, ViQi.
Earlier this month, ViQi received a special COVID-19 grant from the National Science Foundation in which the company will be using artificial intelligence to accelerate vaccine development.
“My reaction was surprised,” ViQi’s Chief Science Officer Dr. Ilya Goldberg told the News-Press.
“Not because I didn’t think our grant was worth funding, but because the competition for getting these grants funded, especially a targeted grant like this with the state of the world, the amount of competition is huge.”
Dr. Goldberg added that due to his experience, usually less than 10% of National Institutes of Health grants receive funding, making this “a great validation.”
“This grant really is for a very specific part of the process of developing antivirals, vaccines or drugs,” he said.
Specifically, ViQi will use the grant to design and run a series of experiments using machine learning to detect infection in individual cells prior to the formation of plaques.
If successful, it would prove that AI can detect differences in these virus-infected cells days or even weeks before a human could prove it.
Antivirals and vaccines typically act on the virus outside of the cells to lower infectivity of said cell. For example, vaccines are made to make antibodies, which would coat the virus and prevent it from causing infection.
Because of this, cells are used to measure the infectivity of a live virus.
“Part of the process is you need to determine the viability of the virus so if you’re attacking the virus with drugs or with a vaccine, you need to measure the effects which is a readout or a virus viability,” Dr. Goldberg explained.
“In the early stages, you can’t do this on humans or animals, you’re going to look at cells for the effectiveness, the grown and incubated cells.”
Knowing this, scientists use a process called Plaque Assays, in which they would have a live virus, one that is being treated with a possible vaccine or drug and one that is not in order to measure the infectivity of the virus.
“You measure what is the effective live virus count in the sample after treatment and basically you dilute out the virus to huge amounts so that you have individual viruses on land on individual cells and have somewhere between 10 and maybe a few hundred infections.”
Eventually, the virus will leave behind a “ring of death” or stain showing its infectivity, as scientists would then measure the live and dead cells.
This process could take anywhere from two to 14 days. What Dr. Goldberg and ViQi would like to do instead is detect the first infectious event.
“This is still an experiment, we haven’t shown an AI can tell this difference. There’s indications it would but we don’t know yet. The idea is that if it can tell the difference between a cell that has been turned into a virus factory versus a normal cell, then you can detect the initial infectious event and that will save you all of the time of these multiple rounds of infections,” Dr. Goldberg said.
Suddenly a process that takes days and even weeks would suddenly be reduced to taking just a few hours. Considering how many vaccines and different antivirals are being tested right now to combat COVID-19, this would be a huge breakthrough.
According to CEO Kathy Yeung, this breakthrough would also give scientists the opportunity to conduct High Content Analysis, a technology imaging process that is used heavily in the pharmaceutical industry.
“Currently they can’t use that technology because of the fact that assay plaque says the cells die because it takes too long. It takes us two to 14 days. Whereas if it can be done in an hour, suddenly you have this additional technology that opens up speed and the different variations that we can test at a much, much higher rate, and I think that is something that we’re particularly interested in because as a company here, we are looking at that and already doing so applying AI to that type of instrumentation,” Ms. Yeung said.
“It was an opportunity, because we’re already working in that area to apply it to virology, to not only make a faster assay but to make the plaque assay compatible with this HCA platform,” Dr. Goldberg added.
Dr. Goldberg said it would only take a few months to see if this AI experiment is a viable and effective option.
According to Ms. Yeung, the experiments will be conducted at Charles River Laboratories.
“They’re going to do the infections and take the images and send us the images and then we’re going to build the AI’s to see if they can differentiate themselves,” Dr. Goldberg said.
“And we’ll be able to tell fairly quickly because our AI tools are fairly mature, so we can run a routinely good, well-curated set of images through them and get an answer of whether or not our AI can differentiate between the groups.”
Dr. Goldberg has been working with HCA and believed in the role this technology could play nearly two decades ago.
For years, it was not understood, but now this grant proves that people believe in the potential.
“It’s definitely satisfying, but I would say I thought it would happen a lot sooner, like even sooner than 10 years,” Dr. Goldberg said with a laugh.
“But that’s the thing you always hear but this is five years in the future and five years in the future is almost as good as never right? It always takes longer than you think so actually getting there is very satisfying, because, you know, a lot of times you never get there.”
And again, the most important thing about this process, if successful, is that it can help combat COVID and future viral threats.
“If you take a broad view of this, the kinds of antivirals we have are nowhere near antibiotics, there really is no equivalent. To a large extent we conquered bacterial disease, with some exceptions, but viral disease is kind of an undiscovered country. The only tool we have really are vaccines, they are so highly tuned to a particular virus which is why you need a different one every year for the flu,” Dr. Goldberg said.
“A lot of people predicted respiratory viruses are a thing we really need to worry about so the tool development that we are in the infancy of is very much going to be a part of our future. We never got to the virus, but the development of antivirals tools and drugs is something we need to redouble our efforts in.”
email: jmercado@newspress.com
