A new approach to vaccines with a machine learning twist could put an end to boosters and seasonal variant shots, according to MIT researchers. This “panvariant” vaccine ignores the virus itself, but rapidly controls the infection by policing the infected cells.
To be clear, this is still in animal testing and is far from deployment. We sought.
The problem is, as amazing as mRNA vaccines are, they are reactive and not aggressive. Find a mutant, sample its spike protein or other signature feature, and slip it into your immune system to know it’s on. Observatory. It’s like letting a rescue dog sniff out a lost hiker’s belongings.
Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory wanted to find another, more permanent way to keep the body safe from COVID attacks. A paper describing their findings was published today in Frontiers in Immunology.
Since the virus’s most distinctive feature, the spike protein, is constantly changing, the team decided to bet on the idea of attacking the virus itself.Instead, they focused on specific molecular signals that reliably appear on the cell’s surface. infected by a virus. If these are caught early and her T cells of the immune system are deployed quickly, the infection could stall before it reaches dangerous or even potentially contagious levels.
These surface signals, called human leukocyte antigens, present various peptides to T cells. It’s like raising a semaphore flag. If all is well, it’s the usual combination of familiar peptides, and the T cells are on the move. If something went wrong, fragments of the virus could wind up on the flagpole and set T cells on fire.
So what does machine learning have to do with all this? We have a lot of data cataloging the different proteins and amino acid chains found in COVID, and how they change when they infiltrate cells. , how to use HLA to indicate that a cell is infected.
Machine learning algorithms are well suited to solving such optimization problems. I need to sort a lot of noisy data for a certain quality combination. In this case, they cataloged relevant peptides, present or “conserved” in all versions of the virus, but also associated with HLA, which could be used as flags for T cells to see. We designed an algorithm that selects about 30 high.
Transgenic mice given our version of HLA and this new vaccine developed a much larger immune response in a short time after infection, and no mice died from the virus.
“This study demonstrates in a living system, in real mice, that a vaccine devised using machine learning can protect against the COVID virus,” said Dr. MIT, one of the paper’s lead authors. Course student Brandon Carter said in an MIT news article. .
An interesting potential benefit is that immunocompromised people who are immune to mRNA vaccines may benefit from this approach with important protection. Some relief can be obtained in the form of a comprehensive immune attack.
The study abstract states:
Undetectable specific antibody responses in mice immunized with MIT-T-COVID indicate that specific T-cell responses alone can effectively attenuate the pathogenesis of SARS-CoV-2 infection. Our results suggest that panvariant T-cell vaccines are worthy of further research, including for individuals unable to produce neutralizing antibodies and for their ability to mitigate long-term COVID-19.
This is a promising area of research and an excellent way to harness computational advances for global health. However, it is also important to recognize that the “pan-variant” option is still in its early stages. For one thing, it could work alongside or against existing vaccines. What if the best peptides for immune response vaccines are those that target destruction by mRNA priming? An overly strong immune response also carries risks such as collateral damage and misselection of ambiguous signaling cells.
But these are good kinds of questions. It’s a relevant question, as the basic features of the new vaccine appear to be working.
