AI-assisted Discovery (AID) PLATFORM

Lead discovery is a long and expensive endeavor with multiple unforeseen difficulties creating setbacks for discovery teams. The preclinical candidate is typically a compromise among multiple parameters and candidate selection at the end is often influenced by management pressure due to the escalating discovery timelines.

ChemPass’ AI-assisted lead discovery platform promises to revolutionize the process. AI-assisted design methods coupled with innovative solutions in evaluation, scoring and selection make major differences: significant expansion of patentable IP, reduction of # compounds synthesized, reduction of lead optimization time, and higher success-rate. Each component in the platform is carefully optimized to fulfill its key objectives in the process while together as a whole they deliver a unified and powerful AI-platform.

Powerful design components include scaffold hopping by SynSpace API, side-chain optimization design by SynSpace, and library enumeration by SynSpace. In addition, our proprietary derivatization design methodology creates lead analogs that efficiency explore relevant chemical space around the lead while guaranteeing full synthesizability of the entire designed set (see our virtual POC study here). Property and med-chem filters automatically remove undesirable molecules.

The project relevant analog cloud is analyzed and filtered for synthesis time, reagent cost, reagent availability, and patentability. Docking or ligand-based evaluations of on-target and off-target activities are further enhanced by machine learning and deep learning models if sufficient datasets are available. Finally, ADMET models with the aid of active learning help generate an MPO score for the selection of best candidates in one or more virtual cycles.

ChemPass’ AI-assisted lead discovery platform promises to revolutionize the process. AI-assisted design methods coupled with innovative solutions in evaluation, scoring and selection make major differences: significant expansion of patentable IP, reduction of # compounds synthesized, reduction of lead optimization time, and higher success-rate. Each component in the platform is carefully optimized to fulfill its key objectives in the process while together as a whole they deliver a unified and powerful AI-platform.

Powerful design components include scaffold hopping by SynSpace API, side-chain optimization design by SynSpace, and library enumeration by SynSpace. In addition, our proprietary derivatization design methodology creates lead analogs that efficiency explore relevant chemical space around the lead while guaranteeing full synthesizability of the entire designed set (see our virtual POC study here). Property and med-chem filters automatically remove undesirable molecules.

The project relevant analog cloud is analyzed and filtered for synthesis time, reagent cost, reagent availability, and patentability. Docking or ligand-based evaluations of on-target and off-target activities are further enhanced by machine learning and deep learning models if sufficient datasets are available. Finally, ADMET models with the aid of active learning help generate an MPO score for the selection of best candidates in one or more virtual cycles.