Incorporating AI technologies that highlight the most confident, accurate compound predictions to enable effective project decision-making
CAMBRIDGE, UK, 10 October, 2019 – Optibrium™, a developer of software for drug discovery, today announced the introduction of its Augmented Chemistry™ services, which provide collaborators with novel artificial intelligence technologies to supplement their skills and experience, enabling them to make more effective decisions and advance their drug discovery projects.
In discovery projects, it is important to base decisions on reliable data to avoid wasted effort pursuing incorrectly selected compounds or missing opportunities by inappropriately discarding potentially valuable compounds. Augmented Chemistry™ services are built on a unique deep learning capability for data imputation, Alchemite™, which has been developed in collaboration with Intellegens Limited. Alchemite™ has been demonstrated to outperform traditional predictive models, both in benchmarking studies  and in partnerships with global pharma and biotech research organisations. Unlike conventional machine learning approaches, Alchemite™ learns simultaneously across all experimental endpoints in a project or corporate database, even based on limited data. The resulting model can automatically highlight the most confident, and therefore accurate, predictions on which to base experimental decisions, including the identification of new or previously overlooked high-quality compounds.
Dr Matthew Segall, Optibrium’s CEO, said: “Optibrium has an extensive track record of successfully introducing and delivering cutting-edge technologies to drug discovery, and our intimate knowledge of the unique challenges of drug discovery and the underlying data enable us to go beyond the hype and deliver results that make a difference. After successfully demonstrating its capabilities in collaboration with our partners, we’re excited to now be able to bring Augmented Chemistry™ services to our global customer base.”
 Whitehead et al. J. Chem. Inf. Comput. Model. (2019) 59(3) pp. 1197-1204