Covid-19 is challenging to treat. Researchshowsthat there can be six distinct "types" of the disease involving different clusters of symptoms. The coronavirus can infect different organs of the body leading to a variety of symptoms. While pharmaceutical companies are working on a vaccine, a team of scientists led by Dan Jacobson at the Oak Ridge National Laboratory (ORNL) has been working to understand the systems biology of the virus using data analytic, and explainable AI tools on ORNL's Summit supercomputer. Recently, theypublisheda paper on a mechanistic model for Covid-19 that can lead to more targeted therapeutic interventions for patients.
Severely ill Covid-19 patients often end up on a ventilator as their lungs are unable to take in enough oxygen. Analyzing gene expression data, these researchers took a holistic approach to the study using Systems Biology frameworks. By understanding the body's underlying mechanisms and how they respond to the coronavirus, severe symptoms of Covid-19 can be explained. If the team's mechanistic model is proven to be accurate, then time and money can be saved by repurposing existing FDA approved drugs to treat severe cases of Covid-19.
Jacobson says, "We are systems biologists, and so this is how we view the world, and we're trying to understand holistically all the molecular interactions that are happening in cells that lead to phenotypic outcomes, whether those outcomes are diseases or other traits. Our understanding of complex processes focuses on looking at all these omics layers, from genome to gene, protein or metabolite expression from a population as well as the microbiome and how that's all conditional on environment. And, overall, that is what we are doing for Covid-19 as really a holistic systems-based approach."
Analyzing the gene expressions of infected individuals against a control group as well as population-scale data, researchers used the Summit and Rhea supercomputers, housed at the Oak Ridge Leadership Computing Facility at ORNL, to discover that the bradykinin system may be responsible for much of the viral pathogenesis. Bradykinin is a peptide that helps to manage blood pressure and can promote inflammation. When more of it is present, it can dilate blood vessels and makes them permeable. If produced excessively, it causes blood vessels to leak and thus leads to a fluid buildup in the surrounding tissues.
Jacobson says, "What we've found is that the imbalance in the renin-angiotensin system (RAS) pathway that appeared to be present in Covid-19 patients could be responsible for constantly resensitizing bradykinin receptors. So, this imbalance in the RAS pathways will take the brakes off the bottom of the bradykinin pathway at the receptor level. In addition, the downregulation of the ACE gene in Covid-19 patients, which usually degrades bradykinin, is another key imbalance in the regulation of bradykinin levels. We have also observed that the key negative regulator at the top of the bradykinin pathway is dramatically down-regulated. Thus, you likely have an increase in bradykin production as well, stopping many of the braking mechanisms usually in place, so the bradykinin signal spirals out of control. "
Using the Summit to run 2.5 billion correlation calculations, the team found gene expression changes that would likely trigger the production of bradykinin. It decreased the expression of enzymes that can break down bradykinin or change how it perceived by cell-surface receptors. Such an escalating buildup of bradykinin would cause blood vessels to leak.
Jacobson says, "It could affect other organs in this way as well. There is a broad range of symptoms being observed across the patient population. For example, if you have a lot of fluid leaking out of the blood vessels in your brain, this could tend to lead to many of the neurological symptoms
A normal blood vessel, shown at top, is compared with a blood vessel affected by excess bradykinin. ... [+] A hyperactive bradykinin system permits fluid, shown in yellow, to leak out and allows immune cells, shown in purple, to squeeze their way out of blood vessels.
The research team also examined the relationship between vitamin D binding sites and the genes in the RAS-bradykinin pathways. Vitamin D helps to regulate the RAS pathway. Vitamin D deficiency has been associated with severe cases of COVID-19. Clinical, pharmaceutical, and research partners are needed to understand Vitamin D's role in treatment.
Jacobson says, "The vitamin D link was an interesting one that affects the very early steps of the RAS pathway. It is simply one component involved in a complex system, and we're probably going to have to target multiple treatments across the entire system to break the cascade. One single intervention alone is probably not going to solve it. But if we can understand all the different components and target those collectively, I think we have a better shot at it."
Another potential therapeutic development path is to repurpose existing FDA approved drugs such as Danazol, Stanasolol, Icatibant, Ecallantide, Berinert, Cynryze, Haegarda, etc.. to reduce the amount of bradykinin signaling to prevent the escalation of the bradykinin storm. Partnerships with pharmaceutical companies and clinical research are needed to design and implement the right clinical trials to see how these types of treatments can be applied.
Jacobson says, "In other work, we are also looking at the SARS-CoV-2 virus itself from a systems biology perspective and think that attempts to inhibit the virus itself will also probably require a combinatorial strategy. It's probably unlikely that there will be a single solution but instead, there will need to be a collection of therapies, similar to what's been done with HIV. We will probably need to have a cocktail of different drugs to help contain the virus. So, it's possible that we will need a combinatorial approach to therapies both on the human side and on the viral side."
Dan's team at ORNL has been consciously building explainable-AI tools for applications across many research areas. Coupled with the Summit supercomputer, Dan's team can examine gene expression data at a much larger scale in a fraction of the time it would take on a desktop computer. One of the difficulties of large-scale gene expression research is that often associations must be generated across a large population of people and tissues. It takes significant computing power and the integration of results from other existing research to make sense of the data. They examined 17,000 different samples of people and their organ tissues to understand the normal gene expression patterns involved in uninfected individuals.
IBM's Summit Supercomputer at ORNL
Jacobson says, "There was a Sunday afternoon eureka moment just staring at the data in the context of different pathways. We've been very interested in the RAS pathways because coronaviruses so often target them. When we looked at the Covid-19 expression data in the context of the RAS pathways, these patterns jumped out at me by simply looking at the data in a different way."
Using system biology, the underlying environmental and biological considerations can be examined using explainable-AI and supercomputing. The group works on other projects that involve a broad range of biology, including bioenergy, microbiomes, cardiovascular disease, autism, opioid addiction, and suicide to name a few. Building tools that can apply to a variety of projects not only allows researchers to save time; it can also add a level of additional transparency into the process to ensure accuracy. The necessary creative aspects of the research process can be taken to the next level with the productive use of various explainable-AI tools.
Dan Jacobsons work at ORNL for the Department of Energy along with his colleagues at ORNL, the Veterans Administration, Yale University, Cincinnati Childrens Hospital and the Versiti Blood Research Institute will likely usher in a new era of using supercomputing and explainable AI to help researchers take a more holistic view to basic scientific research emphasizing the need to understand the bodys mechanisms to find cheaper and better ways to develop clinical treatments.
Original article: Garvin, M.R., Alvarez, C., Miller, J.I., Prates, E.T., Walker, A.M., Amos, B.K., Mast, A.E., Justice, A., Aronow, B. and Jacobson, D., 2020. A mechanistic model and therapeutic interventions for COVID-19 involving a RAS-mediated bradykinin storm. Elife, 9, p.e59177.
The rest is here:
Researchers Use Supercomputers To Discover New Pathway For Covid-19 Inflammation - Forbes