A catalog of reference genomes from the human microbiome

Abstract

The human microbiome refers to the community of microorganisms, including prokaryotes, viruses, and microbial eukaryotes, that populate the human body. The National Institutes of Health launched an initiative that focuses on describing the diversity of microbial species that are associated with health and disease. The first phase of this initiative includes the sequencing of hundreds of microbial reference genomes, coupled to metagenomic sequencing from multiple body sites. Here we present results from an initial reference genome sequencing of 178 microbial genomes. From 547,968 predicted polypeptides that correspond to the gene complement of these strains, previously unidentified ("novel") polypeptides that had both unmasked sequence length greater than 100 amino acids and no BLASTP match to any nonreference entry in the nonredundant subset were defined. This analysis resulted in a set of 30,867 polypeptides, of which 29,987 (approximately 97%) were unique. In addition, this set of microbial genomes allows for approximately 40% of random sequences from the microbiome of the gastrointestinal tract to be associated with organisms based on the match criteria used. Insights into pan-genome analysis suggest that we are still far from saturating microbial species genetic data sets. In addition, the associated metrics and standards used by our group for quality assurance are presented.

Figure 1. Phylogenetic tree of 16S rDNA sequences

The tree was created using ~1500 16S rDNA representing single species. Organisms sequenced as part of the HMP project are highlighted in blue. Additional coloring indicates separation by phylum: yellow, Actinobacteria; dark green, Bacteroidetes; light green, Cyanobacteria; red, Firmicutes; cyan, Fusobacteria; dark red Planctomycetes; grey, Proteobacteria; magenta, Spirochaetes; light pink, TM7; tan, Tenericutes. The purpose of this analysis is not the details of the branching structure (which include minor known artifacts), but the overall distribution of the HMP strains (in blue) around the tree of life.

Figure 2. Contig N50 comparison for twenty-six draft and improved genomes

High quality draft contig N50 bases are shown in magenta and improved high quality draft sequences are shown in green. These data represent the variety of approaches from the four data generation centers. The majority of shotgun data were produced on the Roche-454 platform, though some assemblies include paired Sanger reads to improve contiguity. All draft assemblies are based on the Roche-Newbler assembler, though some of the improved assemblies are based on PCAP (11) and the Celera assembler due to existing integration with finishing and improvement pipelines. Additional variation comes from the improvement approach. Directed Sanger reads from gap spanning PCR amplicons serves as the primary approach while some assemblies have been subjected only to manipulation of the shotgun data, making unrealized joins, removing poor quality data and placing unincorporated shotgun reads.

Figure 3. Inter-strain diversity among Lactobacillus genomes

Each point represents a whole-genome comparison between two Lactobacillus genomes and shows the percentage average nucleotide identity (ANI) on the x-axis as a measure of evolutionary distance, plotted against the percentage of gene content similarity on the y-axis. Only comparisons with ANI values above 85% are shown. The vertical line at 95% corresponds to a recommended cut-off of 70% DNA–DNA reassociation for species delineation. Different intra- and inter-species comparisons are color-coded, with full or open circles respectively, and labeled with given taxonomical name in corresponding color. Colored ovals assist in identifying related data points belonging to a single named species.