Using Data Science to Accelerate Early Drug Discovery with Bayezian and Dataiku

Join us on this 3 part series where we take you through our production journey. Drug development is a lengthy, complex and costly process, entrenched with a high degree of uncertainty that a drug will actually succeed. To add to this complex process, scientists are required to manually perform many experimental techniques in the lab to determine the bioactivity of the compound of interest. Using Data Science we supplement the process of drug design and drug discovery by predicting the bioactivity to expedite the experimentation period in pre-clinical research, potentially taking us straight into pre-clinical trials. In this research, Bayezian has used ChEMBL, a database containing bioactive molecules with drug-like properties. Following on from this, a machine learning model has been built, specifically a random forest regression model, to predict the pIC50 value, these predicted values were then lined up against the experimental values to assess its accuracy. This implementation is further developed through deep learning techniques to predict and forecast the bioactivity space of novel drug candidate molecules that have limited data available. Chemical compound mining for bioactivity requires unique skills of biology, data science and machine learning. By developing this use case within Dataiku, we can scale early insights off of individual desktops and notebooks and into robust data pipelines and operational efficiencies. Bringing together data scientists, chemists, bench scientists and biostatisticians on a collaborative platform effectively releases the bottleneck on both cost and time in the drug development process. Speakers: -Ossama Shafiq, Clinical Data Scientist @ Bayezian -Kelci Miclaus, Director of AI Industry Solutions @ Dataiku Please be aware that by registering for this webinar, you agree to have your personal information shared with Dataiku's partner Bayezian. They may contact you with information that could be of interest to you.