LCMS-based hierarchal clustering as a method to assess biosynthetic relationships among bacteria

Abstract

With the recent outpouring in whole genome sequencing technologies, natural product chemists now have greater access than ever before to the genomic information surrounding biosynthetic potential. The continued decrease in cost for whole genome sequencing holds great promise as a tool for prioritizing bacterial strains for drug discovery. Nonetheless, rapid methods to prioritize strains for discovery programs are still needed. We sought to investigate hierarchal clustering analysis (HCA) based on LCMS profiles as a method to evaluate metabolomic relationships among strains and examine how these relationships were reflected in genomes using whole genome analysis. To test this method, we analyzed a group of marine Streptomyces spp. using LCMS, and developed open-source workflows in R with XCMS to generate hierarchal clusters. We ensured clusters produced using R were comparable to clusters made using proprietary software. Trees from LCMS data were directly compared to 16S rDNA phylogenetic trees, and relationships between these clusters were investigated further using principal component analysis (PCA). A second group of Streptomyces spp. with whole genomes assembled were also analyzed by LCMS-HCA. The relationships identified based on metabolites were assessed in tandem with whole genome analysis. Overall, our LCMS-based HCA methods aim to assist the strain selection process for natural product drug discovery platforms, and in addition can offer an alternative “pre-selection” tool to guide future genomic analyses.