Computational generation of an annotated gigalibrary of synthesizable, composite peptidic macrocycles

Significance We describe computations to anticipate products of multistep reaction sequences. The work is based on experimental methods developed earlier to amalgamate synthetic scaffolding reagents with small linear peptides. Hybrid products retain molecular recognition elements in the peptide, but display that functionality as part of amphipathic macrocycles having defined conformations and improved pharmacological properties. The hypothetical scope of the chemistry is large and far outpaces the experimental format. To explore the structure space more extensively, we devised algorithms to predict outcomes of more than 2 billion processing sequences. Software was also developed to generate accurate three-dimensional structures for each product. The resultant virtual library is a resource that can be deployed broadly in search of novel ligands for protein receptors.