Identification of Macrocyclic Peptide Families from Combinatorial Libraries Containing Noncanonical Amino Acids Using Cheminformatics and Bioinformatics Inspired Clustering

In the past decade,macrocyclic peptides gained increasinginterestas a new therapeutic modality to tackle intracellular and extracellulartherapeutic targets that had been previously classified as "undruggable".Several technological advances have made discovering macrocyclic peptidesagainst these targets possible: 1) the inclusion of noncanonical aminoacids (NCAAs) into mRNA display, 2) increased availability of nextgeneration sequencing (NGS), and 3) improvements in rapid peptidesynthesis platforms. This type of directed-evolution based screeningcan produce large numbers of potential hit sequences given that DNAsequencing is the functional output of this platform. The currentstandard for selecting hit peptides from these selections for downstreamfollow-up relies on the frequency counting and sorting of unique peptidesequences which can result in the generation of false negatives dueto technical reasons including low translation efficiency or otherexperimental factors. To overcome our inability to detect weakly enrichedpeptide sequences among our large data sets, we wanted to developa clustering method that would enable the identification of peptidefamilies. Unfortunately, utilizing traditional clustering algorithms,such as ClustalW, is not possible for this technology due to the incorporationof NCAAs in these libraries. Therefore, we developed a new atomisticclustering method with a Pairwise Aligned Peptide (PAP) chemical similaritymetric to perform sequence alignments and identify macrocyclic peptidefamilies. With this method, low enriched peptides, including isolatedsequences (singletons), can now be clustered into families providinga comprehensive analysis of NGS data resulting from macrocycle discoveryselections. Additionally, upon identification of a hit peptide withthe desired activity, this clustering algorithm can be used to identifyderivatives from the initial data set for structure-activityrelationship (SAR) analysis without requiring additional selectionexperiments.