Dr. Fanwang Meng (Heidar-Zadeh’s Group) has been awarded the prestigious Banting Postdoctoral Fellowship to pursue “Drug Discovery with Machine Learning: Model Development and Applications”. Read more at the Queen’s Gazette website.
“Spurred by advances in computer hardware and numerical algorithms, machine learning (ML)— the core of artificial intelligence—is revolutionizing the way drugs are developed. ML has been widely adopted in all the stages of drug discovery and development because of its ability to reveal hidden patterns in data. However, good ML models require many high-quality data points, which is uncommon in medicinal chemistry because experimental measurements are costly (ergo, data is scarce, and publicly available data tends to be biased in terms of successful/active molecules and/or systems where experiments were easier to perform) and corrupted with low-quality data (datasets with missing and/or unreliable values). Merely discarding all the data associated with noisy/missing values is undesirable, as an enormous amount of potentially useful information is discarded in the process. Inspired by the success of recommender systems in e-commerce, I will create a suite of machine learning models and computational tools tailored for medicinal chemistry. This provides a novel framework for drug design: the data quality problem is solved because imbalanced/biased/unreliable/missing data are imputed using ML; the data scarcity problem is ameliorated because more data is accessible, and information about pharmacophores can be transferred between different drug targets. This potentially impacts the whole pipeline of early-stage drug discovery and allows us to target malaria and rare diseases. All the models and datasets will be shared as free and open-source research software/data in accord with FAIR principles, aiming to provide computational medicinal chemists with tools to accelerate drug discovery.”