Accurate compound bioactivity data are the foundations of decisions on the selection of hits and the progression of compounds in discovery projects. However, in practice, the experimental data available on potential compounds of interest are sparse; for example, the ChEMBL dataset is just 0.05% complete and the sparsity of data in proprietary pharma databases is similar. Can new ‘AI’ methods learn from the available data and ‘fill in the blanks’ to find new, active compounds for our drug discovery projects?

In this webinar, presented by Matt Segall (Optibrium) and Tom Whitehead (Intellegens), you can learn more about Alchemite™, a novel deep learning algorithm. Unlike many deep learning methods, this approach is capable of being trained using sparse and variable input data, typical of those available in drug discovery. This enables Alchemite to learn from correlations between experimental endpoints, as well as between molecular descriptors and protein activities, to more accurately impute the missing activities.

We present a case study that demonstrates that Alchemite out performs traditional quantitative structure-activity relationship models and discuss how these results can be used to fill in missing data, predict compound activity profiles and identify new active compounds.