Shortening the discovery cycle is key to improving productivity and enhancing competitiveness in today’s Biopharmaceutical industry. BIOVIA Generative Therapeutics Design (GTD) is an artificial intelligence (AI) solution that automates the virtual creation, testing and selection of novel small molecules with a view to reducing expensive real-world testing. Research organizations can achieve true business transformation in molecular discovery by simultaneously improving lead quality and shortening discovery timelines, resulting in potential savings of millions of research dollars per program.
The BIOVIA GTD solution combines elements from five longterm and rapidly accelerating trends in drug discovery:
- Quantitative Structure-Activity Relationship Modeling (QSAR) – QSAR modeling of empirical data on assays of interest using established and validated machine learning methods yields highly predictive estimations of biological activity for proposed molecules.
- Target Product Profile (TPP) – Early attention to balancing all attributes required for a molecule to proceed from preclinical testing into human trials requires multi-parametric optimization at all stages of discovery (e.g., potency, safety, manufacturability, novelty, etc.).
- Active Learning (AL) – This special case of Machine Learning tightly couples real and virtual activities, allowing a learning algorithm to query a user (or another information source) to label new data points with the desired outputs. In GTD, this means intentionally synthesizing molecules that will expand the domains of applicability of Machine Learning models.
- Generative Chemistry – While enumeration of all molecules that could be drugs is not possible with reasonable resources, chemists can explore a large area of chemical space through iterative alteration of a starting molecule guided by a fitness function.