Estimated reading time: 4 minutes

There are many things the global coronavirus pandemic crisis has shown us, but there are a few things that stand out. One of them is that the current vaccine and drug development process takes a lot of time. Some would even say too long for having the desired impact on containing the spread.

Source: Own representation based on IFPMA 2019

There are for sure elements in the vaccine development process shown above that can’t be done much faster in order to develop a safe and effective vaccine. But for other steps, modern technology can contribute significantly to shorten the time needed for creating new vaccines and drugs.

While obtaining his doctorate degree at the Institute for Research in Immunology and Cancer at the Université de Montréal, Tariq Daouda, PhD, developed an AI algorithm that can predict which parts of a virus are more likely to be exposed at the surface of infected cells, which are called epitopes.

Source: Image Created by Maude Dumont-Lagacé

Over the past weeks, a research team consisting of several PhDs in immunology and bioinformatics volunteered to build a platform open-sourcing their research. Together with the pro-bono help of several software developers, they launched today, so that everyone around the globe can access the results the neural net found and utilize it for further research. The research on which is based on has also been a team effort and can be accessed here (PDF).

So how does it work? is using a neural net to analyze the genomic sequences of humans and different mutations of viruses to predict the best epitope candidates for further vaccine and drug research. This process previously took weeks or even months of manual lab work –  assuming the labs could get their hands on enough living viruses to conduct their research. With the new approach it only takes hours and the genetic sequence of a particular virus to let the neural net do its work.

  1. Cells present at their surface tiny fragments of the proteins that they contain. These bits are called epitopes and serve to indicate to the immune system what’s happening inside the cells. When a cell gets infected by a virus, the genetic material from the virus is injected into the cell.
  2. Then, the production of new viruses from this genetic material begins. However, some (but not all) fragments from the viral proteins produced also get presented at the cell surface.
  3. By analyzing the genetic sequence of the virus, our artificial intelligence algorithm (CAMAP) can predict which fragments of this virus are more likely to be presented at the cell surface.
  4. We can then select a combination of fragments that have a high chance of being presented at the surface, we can design a vaccine that will efficiently train our immune system.
  5. If immune cells then encounter infected cells, they can rapidly recognize them and kill the virus.

“COVID-19 stresses the need to accelerate the design of vaccines and therapies to reduce the human and economic impact of global pandemics,” said Tariq Daouda, PhD, research fellow at Harvard Medical School and team lead of “People infected with COVID-19 tend to have less immune cells, making it difficult to get enough infected cells to study them appropriately in a lab — and, because they are so rare, labs are in competition with each other to obtain them. makes code the petri dish — utilizing open source technologies to connect machine learning to biomedicine to help accelerate learnings and findings.”
The team around Tariq decided to completely go the open-source way to follow their ultimate vision of open-sourcing vaccine research, and is their first step.

How can I contribute?

The team of is working tirelessly to improve and expand the epitope.worlds frontend as well as the backend, which are both open-source under Apache 2. Also all bioinformaticians, data scientists and machine learning specialists are invited to join the team.

epitope.worlds highly appreciates contributions to expand the platform’s functionality and allowing researchers to analyze data in new, even better ways.

Kudos to the whole team!

It is absolutely remarkable what the team achieved in such a short time! We feel lucky to have the chance to support:

  • Tariq Daouda (PhD): Team lead, Research (AI) & Web Development
  • Maude Dumont-Lagacé (PhD): Research (Immunology)
  • Albert Feghaly (Msc): Bioinfomatician
  • Oliver Caron-Lizotte (Msc): Web Infrastructure and web development
  • Antoine Zieger (Ing. Msc): Web Infrastructure
  • Logan Schwartz (Msc): Web Development
  • Walter Sorbal (Bsc): Web Development
  • Jörg Schad (PhD): ArangoDB Oasis Cluster

The whole ArangoDB team feels blessed to have the opportunity to sponsor, which is running on ArangoDB Oasis, and to have people on our team investing their free time to support this cause.

Continue Reading

RC1 ArangoDB 3.4 – What’s new?

ArangoDB Among Highest Rated Operational Databases Management Systems solutions in Gartner Report with 4.7/5 Rating

Time traveling with graph databases