Microsoft works with ImmunityBio to model how key proteins cause COVID-19 infections

After the outbreak of the new coronavirus, many international research teams have been working on the development of potential drugs, which obviously depends on a great deal of computing and modeling. Immunotherapy start-up ImmunityBio, for example, is working with Microsoft’s Azure cloud computing division. With the GPU computing power of 24 PetaFlops, it was able to model the critical protein structure of COVID-19 in considerable detail to study how the new coronavirus enters human cells.

Microsoft works with ImmunityBio to model how key proteins cause COVID-19 infections

The green part of the image above is the receptor binding structure region of the protrusion protein, and the red part is the ACE-2 receptor, a combination of which is the first step in leading to the infection of the new coronavirus. (Pictured: ImmunityBio)

With the help of the powerful computing power of the Auzre cloud platform, the new collaboration means they can model the protrusion protein in a matter of days, rather than just months.

The time saved means that the model can be put into the hands of scientists and researchers working on potential vaccines and treatments more quickly to find ways to inhibit virus replication and prevent it from attaching to human ACE-2 protein receptors.

One of the simplest scenarios is to make sure that the protoprotein sits unable to connect to the receptor’s aiming. In patients with neo-coronavirus self-healing, their natural antibodies have been able to do just that.

The vaccine, which is currently being developed, is focused on the same thing. In addition, many treatments are seeking to reduce the likelihood of virus infection infecting new cells and locking them to block replication in the body.

Microsoft’s partnership with ImmunityBio is understood to involve an array of 1250 Nvidia V100 Tensor Core GPU components.

The software giant uses these GPUs for machine learning applications in Azure clusters, combined with ImmunityBio’s existing 320 GPU clusters to specifically adjust molecular modeling efforts.

The results of the collaboration will be provided to researchers studying COVID-19 remission and prevention therapy to facilitate the arrival of faster and more effective solutions.