Despite many decades of work by biomedical researchers and therapeutic developers, human targets that can bind drug-like molecules are only a small proportion of the human genome, and even fewer are actual drug targets. To explore new targets and find new leads or tool compounds, phenotypic screens are increasingly utilized in drug discovery. Potent and selective chemical tool compounds against well-defined targets can be used as probes for these targets in disease-relevant phenotypes. Unfortunately, identifying such tools from large heterogeneous (and potentially noisy) bioactivity databases is non-trivial. We present here an evidence-based quantitative confidence metric, the compound-target tool score, which can be used to systematically rank tool compounds for targets. In Novartis we have collected over 4000 tool compounds, also known as the mechanism of action box (MOA Box), and we are using this collection to connect targets with novel biology. Furthermore, we are also exploring machine learning algorithms that can predict compound-target activities, especially for previously un-drugged targets.<br/>

Computational methods help find chemical matter to uncover novel biology

256th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.