SARS-CoV-2 (like all RNA viruses) mutates rapidly and creates potentially more infectious and lethal variants. In the age of increasingly powerful computers, algorithms can be developed that can predict virus variants well in advance. What are the advantages and potential?

We discuss this with Professor Sabrina Pricl, Associate Professor of Principles of Chemical Engineering at the University of Trieste (Italy) and she is the Scientific Principal of the Molecular Biology and Nanotechnology Laboratory (MolBNL@UniTs) at the Department of Engineering and Architecture of the University of Trieste.

Professor Pricl, in the recent paper published by your group, computational analysis was used. Before talking about the results obtained, please explain what computational analysis is and where does your interest come from?

Computational analysis uses the computational power of the latest computers (High Performance Computing) in order to analyze complex data in a short time. I started using computational analysis about 20 years ago, when the available computers were much less powerful than those currently available. The first studies I did were aimed at predicting molecular structures, particularly in the biological field. I had the good fortune to be a researcher at AIRC (Italian Association for Cancer Research) and collaborating with the Tumor Institute of Milan, we worked on the possibility of predicting the effects of certain mutations on oncogenes, and in particular, in response to targeted therapies.

With the entry of SARS-CoV-2 now more than a year ago, we thought of employing this experience and knowledge to be able to predict the impact of new mutations on virus infectivity. In addition, we had already applied computational analysis to the study of the Hepatitis C virus, respiratory syncytial virus and HIV, so we started with some solid foundations.”

Please tell us about the computational study you recently published.

In the study published in ACS Nano, we focused on a specific region of the virus, more specifically the Spike protein that recognizes and interacts with ACE2, the receptor found on human cells through which SARS-CoV-2 enters cells and then replicates. We wondered which parts of the Spike protein and ACE2, or amino acids, conferred the highest affinity to enable binding.

The most interesting thing about this study is that we didn’t limit ourselves to mutations in the virus but also investigated which key mutations in the human ACE2 protein could affect the efficacy of the infection.

However, there are many mutations that have been recognized and not only in the gene region that produces the Spike protein, so we started to produce mutations in the regions that we scientifically knew to be fundamental, both in the virus and in the ACE2 receptor.

What do you mean by “key mutations”?

I mean those mutations that confer better binding between the Spike protein and the human ACE2 receptor and therefore gave us reasonable confidence that they were important in influencing the degree of infectivity of the virus. All this was done at the supercomputer level and therefore with a very low financial commitment and with much faster timelines than the traditional laboratory investigation. In addition, I’d like to say that all our recent research on SARS-CoV-2 concretized in two papers appearing in high-level scientific journals has been completely self-financed by our group, without any financial aid.

We conducted our studies during the summer of 2020 and they were published in late 2020. Since then, two international groups of researchers did experimental work in the laboratory, and our approach, completely based on computer simulations, revealed a 92% agreement with such experimental data. This additional result lends further value for and credibility to our approach and predictions.”

What mutations did you therefore predict before they were even identified in the general population?

We’ve predicted eight, including the famous N501Y that’s found in the UK, South African and Brazilian variants. This mutation is the most common and widespread, as it’s located in the contact zone between the Spike protein and the ACE2 receptor, and as we know, it confers a higher degree of infectivity to the mutated viral strain.

What benefit can the computational study of variants provide in relation to vaccine efficacy?

In this regard, we’re analyzing antibody data from both post-disease immunized and vaccinated individuals. Unfortunately, the first data suggest that the E484K mutation that’s present in the South African and Brazilian variant demonstrates a very low binding between the Spike protein and the antibody (i.e. the antibodies do not protect against infection).

Therefore, it’s very important that we mention the ability to identify and monitor variants and to regulate the infected individuals, in order to spread as little as possible the variants that reduce the effectiveness of antibodies produced by vaccines.

Moreover, thanks to the new technologies that are at the very core of the production of DNA and RNA vaccines, it would be possible to change even a very small part of the vaccine that will provide a good effectiveness, against the new variants, to the antibodies induced by vaccination. The possibility to predict (with a computational study) the onset of new variants could provide a time advantage to pharmaceutical companies and surveillance bodies“.

What future developments do you foresee for computational studies?

A possible future development of the predictive computational approach may be its use as a first approach to “skim” or “screen” variants and then focus laboratory experimental studies on the most likely scenarios. This would lead to a significant acceleration of experiments as well as a reduction in cost.

Another very interesting approach is the use of computers to identify existing drugs, antivirals first, that may be effective on the virus. This is an avenue that would greatly accelerate treatment because drugs already approved have passed safety criteria and could be validated for a different purpose by the regulatory agencies (e.g. EMA and AIFA) rather quickly.

This post is also available in: Italiano


Please enter your comment!
Please enter your name here