Google DeepMind designs lung cancer drug candidates with fewer molecular 'bends'

Health

Google DeepMind used matched molecular pair analysis to design rigidified macrocycles targeting lung cancer treatments. The approach reduced rotatable bonds in crizotinib analogues from five to zero while increasing predicted potency. Thousands of compounds were screened against ChEMBL and PubChem databases using RDKit tools, generating synthetic routes for lorlatinib-like molecules. This method could lower costs for targeted lung cancer therapies if scaled, though results are specific to this cancer type and require integration with chemical databases. Experimental validation of the compounds was not reported in the source. The innovation advances drug discovery speed for lung cancer but remains limited to this therapeutic area and depends on existing chemical infrastructure.

Source: Google DeepMind