Automated Culturomics Enables Deep Strain Isolation and Analysis of Personalized Gut Microbiomes

Sequencing efforts over the past two decades have led to a structured metagenomic profile of the human gut microbiome. While the collective research community has generated a tremendous amount of sequencing data for the human gut microbiome, the repertoire of individual gut strains available for experimental studies remain limited and the amount of phenotypic data of individual gut strains lag even further behind. A typical gut microbiome may contain hundreds of unique taxa, with a long-tailed abundance distribution. As such, extracting unique strains become increasingly difficult using conventional tedious and labor-intensive “brute force” strategies. Here, we describe the development of an automated anaerobic strain isolation system and a high-throughput genotyping pipeline that together enables rapid and efficient generation of data-rich gut strain collections on-demand. This system leverages distinguishing visual and phenotypic differences between colonies using machine vision methods to assess strain uniqueness for automated colony isolation and high-throughput genotyping. We demonstrated the utility of this approach to isolate personalized strain collections from a broad range of human gut microbiome samples, yielding nearly ~13,000 isolates and ~700 draft genomes spanning ~350 unique OTUs. Machine learning of colony morphology and growth characteristics resulted in algorithmically-driven colony selection strategies that significantly improved targeted colony isolation. Furthermore, integration of all phenotypic data on plate allows systematic analysis of species-specific pattern and whole-genome sequencing of isolates revealed different levels of intra- and inter-individual variation. Collectively, these results exhibited the capacity and scalability of our automated high-throughput pipelines to generate personalized strain collections and integrated phenotype-to-genotype datasets which will greatly accelerate microbiome studies towards deeper mechanistic insights.