AI Unlocks Protein Secrets for Fighting Diseases

Picture proteins that do not stand still; they twist, turn, and change forms like dancers at a busy bar. Proteins that have significant fundamental use within the cell and chicken coops: these proteins are also known as intrinsically disordered proteins (IDPs).

While traditional methods are incapable of assisting researchers with the dynamic nature of their proteins (IDPs) based upon their environment, Disobind, an open-access, Artificial Intelligence (AI) induced analytical tool, has the ability to predict IDP interactions accurately.

Why Disobind?

Historically, protein researchers have had difficulty predicting IDP binding interactions due to the fact that binding interactions vary depending on the chemical environment in which they form. Disobind allows researchers to identify binding sites for their protein of interest consistently. This is accomplished through a protein language model built from millions of unique protein sequences, as well as by using data from a variety of different types of proteins.

Core Innovation: Disobind analyzes both proteins in a given pair to predict how the binding sites of the pairing proteins change based on the surrounding context within the cell. In cases where multipleIDPs interact together, the elastic 'floppy' behavior is accurately represented.

Disobind has tested as superior to AlphaFold-multimer on previously untested protein pairs. In addition, by using the predicted results from both of these models, researchers can expect an additional boost in accuracy by over 20%.

Open-source model of Disobind (available through Cell Systems) will allow researchers to download the tool and begin working with it today.