A human gut bacterial genome and culture collection for improved metagenomic analyses

Abstract

Understanding gut microbiome functions requires cultivated bacteria for experimental validation and reference bacterial genome sequences to interpret metagenome datasets and guide functional analyses. We present the Human Gastrointestinal Bacteria Culture Collection (HBC), a comprehensive set of 737 whole-genome-sequenced bacterial isolates, representing 273 species (105 novel species) from 31 families found in the human gastrointestinal microbiota. The HBC increases the number of bacterial genomes derived from human gastrointestinal microbiota by 37%. The resulting global Human Gastrointestinal Bacteria Genome Collection (HGG) classifies 83% of genera by abundance across 13,490 shotgun-sequenced metagenomic samples, improves taxonomic classification by 61% compared to the Human Microbiome Project (HMP) genome collection and achieves subspecies-level classification for almost 50% of sequences. The improved resource of gastrointestinal bacterial reference sequences circumvents dependence on de novo assembly of metagenomes and enables accurate and cost-effective shotgun metagenomic analyses of human gastrointestinal microbiota.

Fig. 1. Phylogenetic diversity of the human gastrointestinal microbiota genome collection.

Maximum-likelihood tree generated using the 40 universal core genes from the 737 HBC genomes (green outer circle) and the 617 high-quality public genomes derived from human gastrointestinal tract samples, which together make up the HGG. Branch color distinguishes bacterial phyla belonging to Actinobacteria (gold; n = 129 genomes), Bacteroidetes (green; n = 231 genomes), Firmicutes (blue; n = 772 genomes), Fusobacteria (black; n = 26 genomes), Synergistetes (pink; n = 2 genomes) and Proteobacteria (orange; n = 194 genomes) shown.

Fig. 2. Comparison of high-quality reference genomes from de novo assembly and HGG.

a, Read base usage as a percentage of total read bases present within the metagenomics samples (n = 13,490) that could be mapped to their respective de novo assembled contigs (min., 22.23; Q1: 62.87; median, 76.89; Q3, 89.99; max., 99.98) and metagenome-assembled genomes (MAGs; min., 0.16; Q1, 8.17; median, 16.09; Q3, 31.16; max., 65.64). b, Total number of classified bins using HGG (blue), genomes derived from the HBC collection alone (HBC; orange), the HMP (purple) and gastrointestinal derived isolates from the HMP (HMP-GI; green). c, Total number of 39,913 MAGs classified using subsampled sets of genomes from the HGG (blue), HBC (orange), HMP (purple) and HMP-GI (green). Error bars show mean and s.d. (n = 100 bootstraps).

Fig. 3. Classification efficiency using the HGG.

a, Contig assignment from 13,490 metagenomic samples at genus (90%; min., 31.35; Q1, 62.92; median, 74.48; Q3, 84.10; max., 100.0), species (95%; min., 18.94; Q1, 54.80; median, 67.35; Q3, 78.73; max., 100.0) and strain (99%; min., 0.0; Q1, 30.03; median, 40.82; Q3, 54.35; max., 90.77) identity compared to the HGG. b, Classification of metagenomic sequenced samples from North America (n = 2,064; min., 1.31; Q1, 79.07; median, 88.16; Q3, 98.42; max., 99.97), Europe (n = 1,431; min., 52.07; Q1, 76.28; median, 80.66; Q3, 84.47; max., 99.52), Asia (n = 191; min., 72.37; Q1, 86.56; median, 90.84; Q3, 94.13; max., 98.93) and the other undefined locations (n = 9,804; min., 1.45; Q1, 76.28; median, 82.14; Q3, 88.25; max., 99.94).

Fig. 4. Dominant bacterial species within the human gastrointestinal microbiota.

Dominant species, ordered by prevalence, found within the 13,490 human gastrointestinal metagenomic samples and their relative abundance within each sample. Color denotes Bacteroidetes (green), Firmicutes (blue), Proteobacteria (orange), Actinobacteria (gold).

Fig. 5. Bacterial functions in the human gastrointestinal tract.

DAPC analysis of functional categories shows a clear separation of functions associated with each dominant phylum (Bacteroidetes (green; n = 231 genomes), Firmicutes (blue; 772 genomes), Proteobacteria (orange; n = 194 genomes), Actinobacteria (gold; n = 129 genomes)) within the HGG collection.