Deeptracer's target, the coronavirus.

A research team at UW Bothell is using artificial intelligence in the battle against COVID-19. Also, the team is adding advanced mapping technologies to the list of tools used to fight the virus. ‘Deeptracer’ uses machine learning and 3D imaging technology to tease out the structure or the coronavirus.

According to UW Bothell, their AI driven web-based software leverages AI to recreate a three-dimensional model of a virus protein. Due to this setup, the software is able to ‘trace the connection’ between the atoms of a protein.

Dong Si, said, ‘To fuse these technologies together and make the prediction very fast and accurate and more importantly automated.’

Dong Si is the assistant professor of computer sciences at UW Bothell.

Making sense of a now-familiar image

You may have just thought that those red spikes seen on artist renderings of the coronavirus were just what the artist thought the pathogen looked like, but those little spikes seen in those renderings are actually what the researchers are trying model. Although there are some who may not feel the same way, the team is confident that the research they are conducting will be critical in the battle against the deadly virus. If accurately modeled, the data gathered will help the fight the coronavirus.

If we know all these detail structures those pharmaceutical people and biologists and other people will be able to design a vaccine or drug.

Dong Si

Something like this may sound like it’s expensive, but Scientists all over the world are able to use the DeepTracer web-app for free.

Si said, ‘That makes us feel that our work is an important benefit to the people around the world.’

The Deeptracer team is growing

Around 16 students recently joined the team. The new members are college students mostly from UW Bothell and a high school student. Website management, data tracking, algorithm refinement, and database engineering are some of the things that need to be done.

According to Si, independent researchers found the accuracy of their model to be about 85%. He does, however, insist that that’s not good enough. 100% is the goal, but the team will more than have to settle for a scientifically reasonable 99.999%.

Leave a Reply

Your email address will not be published. Required fields are marked *