Fig. 1

Percent of soil and human metagenomic reads mapped to the GEM and NCBI RefSeq databases as a function of nonpareil kmer diversity. Nonpareil kmer diversity is a measure of genetic diversity within a metagenome. Human microbiomes are less diverse than soil (agricultural, forest and desert) metagenomes. From the efforts of the Human Microbiome Project there is a large collection of bacterial genome sequences in NCBI’s RefSeq database and consequently a large proportion of reads from human metagenomes map to this database. Typically 50–90% of human metagenome reads map to the combined databases. In contrast very few soil metagenomic reads map to genomes in NCBI’s RefSeq, and less than 5% map to the GEM catalog