Differences in metagenome coverage may confound abundance-based and diversity conclusions

Although the importance of rarefying 16S rRNA gene data to the same coverage for reliable comparisons of diversity between samples has been well appreciated, the impact of (shotgun) metagenome coverage (i.e., what fraction of diversity was sequenced) on biological conclusions is commonly overlooked. We demonstrate that uneven coverage can lead to misleading conclusions about which features (e.g., genes, genomes) may differentiate two metagenomes or their diversity comparisons. We outline an approach to minimize this impact.