Kantify announces new upgrades on ZeptoNet
Kantify announces new upgrades on ZeptoNet - its AI technology for drug discovery
We are excited to announce the recent improvement of Kantify’s AI technology for advancing drug discovery, called ZeptoNet. ZeptoNet is a deep neural network that has been trained to find the most promising drug candidates that are likely to have both therapeutic effect and be safe for use.
Our latest version of ZeptoNet, v12, has been trained on more, better curated data, using improved algorithms and a better optimization infrastructure. Through this update, we continue to significantly increase the performance gap between our results and the state of the art. Furthermore, through developing more extensive benchmarks, we can now prove for the first time that ZeptoNet has developed robust capabilities to predict whether compounds and proteins that are fully unknown to ZeptoNet will meaningfully interact with each other. This indicates that ZeptoNet is learning the underlying biophysical properties that make small molecules bind to proteins.
With the latest update, we further improve our ability to find novel drugs on new disease targets, and our prediction of the safety of these drugs before testing them on animal models. Thus we help accelerate the whole drug discovery pipeline and help find promising drug candidates with very limited needs for a physical screening of the compounds. Because our models run on computers rather than labs, we further decrease the time and cost required to research and find novel drugs.
We are excited to run ZeptoNet v12 in the following weeks and months with our partners on various drug discovery pipelines.