Improving protein-ligand binding models using multidimensional, mixed-confidence training data
Artificial Intelligence is powered by data. In general, the predictive power of AI improves with higher data abundance and higher data quality. However, real-world datasets vary greatly in quality and quantity. In drug design, like many other tasks, there is an abundance of low quality data and scarcity of high quality data. This project will be developing new computational approaches to balance data quality and quantity to make the most out of all available information. These computational innovations will help boost the predictive power of Cyclica’s computational drug profiling and drug design software. Cyclica is a Canadian startup named by Deep Knowledge Analytics as one of the top 20 AI in Pharma companies globally.