The study of cells from convalescent patients of CoViD-19 would pave the way for the creation of a vaccine. Illustration: Gerd Altmann, Pixabay, CC
Around the world, several teams involved in the race for a Vaccine against CoViD-19 believe that artificial intelligence could contribute to a breakthrough.
Tariq Daouda, a postdoctoral researcher at Harvard University’s School of Medicine, has developed with his Quebec team an interactive platform that is supposed to be able to predict promising vaccine targets.
Like a telescope
“Artificial intelligence plays the role of the telescope in astronomy, an augmented tool that does not replace humans, but allows us to deal with things that are impossible with the naked eye. In this case, tens of thousands of epitopes, small sequences of proteins that can alert the immune system,” explains Daouda.
This new, yet preliminary work, which was pre-published on 4 June, is based on an algorithm developed in the quest for a cancer vaccine and adapted to the new threat.
This trail is the result of the researcher’s previous doctoral work, under the aegis of Claude Perreault, professor of the Institute for Research in Immunology and Cancer at the University of Montreal.
When a cell is infected with the virus, small pieces of degraded viruses (epitopes) on the surface of the cell activate the immune system’s response and are attacked by antibodies – T-cells or “soldiers” in that system.
The algorithm will compile a list of epitopes (small coding sequences) that could be targets for a future vaccine.
In theory, through an artificial neural network called Codon Arrangement MAP Predictor (CAMAP), AI will predict probable candidates it has learned to recognize, following the analysis of thousands of coding sequences from the human genome.
Epitopes vary depending on mutations, the researcher explains. With SARS-Cov-1 (the virus responsible for the SARS outbreak in 2003-2004), our results show that CAMAP gives particularly high marks to experimentally validated epitopes.”
Given the close kinship between SARS-CoV-1 and SARS-CoV-2 (the CoViD-19 viruis), “these results are encouraging for sarS-CoV-2 application.”
Having a list of all SARS-CoV-2 epitopes would therefore provide the ideal targets for vaccine creation. This type of epitope vaccine is already being developed to control infection of the parasite responsible for a deadly disease (visceral leishmaniasis) and against certain types of cancers.
Healed patient cells
This work has yet to be validated with cured patients — or rather their cells. “There are different ways to develop a vaccine: the one with epitopes remains among the least toxic, because it is designed with very small sequences. A few words in a 400-page book,” says Daouda.
‘This study presents a fascinating and revolutionary new way of conceptualizing this antigen through the immune system,’ commented Réjean Lapointe, Scientific Director of the Montreal Cancer Institute.
‘The conventional approach is based on the search for pieces of peptides in the large chain of cellular proteins.’
Finding the best targets
A great admirer of the immunology work of Claude Perreault and his team, Mr. Lapointe assures that he will integrate this algorithm into this work to counter-validate them.
“We’re going to add that to our toolbox. For our part, we are working on the biological side, to design a bank of human cells from convalescent patients of CoViD-19.”
‘When we have the funding, we can then test the different peptides expressed by the three main coronavirus proteins and see which are the best targets for a vaccine,’ added
Journalist at Agence Science-Presse, an independent, non-profit media outlet based in Montreal. The only scientific news agency in Canada and the only one in the french-speaking world that caters to the mainstream media rather than to business.