Under-detection of endospore-forming <i>Firmicutes</i> in metagenomic data

Microbial diversity studies based on metagenomic sequencing have greatly enhanced our knowledge of the microbial world. However, one caveat is the fact that not all microorganisms are equally well detected, questioning the universality of this approach. <i>Firmicutes</i> are known to be a dominant bacterial group. Several <i></i>Firmicutes species are endospore formers and this property makes them hardy in potentially harsh conditions, and thus likely to be present in a wide variety of environments, even as residents and not functional players. While metagenomic libraries can be expected to contain endospore formers, endospores are known to be resilient to many traditional methods of DNA isolation and thus potentially undetectable. In this study we evaluated the representation of endospore-forming <i>Firmicutes</i> in 73 published metagenomic datasets using two molecular markers unique to this bacterial group (<i>spo0A</i> and <i>gpr</i>). Both markers were notably absent in well-known habitats of <i>Firmicutes</i> such as soil, with <i>spo0A</i> found only in three mammalian gut microbiomes. A tailored DNA extraction method resulted in the detection of a large diversity of endospore-formers in amplicon sequencing of the 16S rRNA and <i>spo0A</i> genes. However, shotgun classification was still poor with only a minor fraction of the community assigned to <i>Firmicutes</i>. Thus, removing a specific bias in a molecular workflow improves detection in amplicon sequencing, but it was insufficient to overcome the limitations for detecting endospore-forming <i>Firmicutes</i> in whole-genome metagenomics. In conclusion, this study highlights the importance of understanding the specific methodological biases that can contribute to improve the universality of metagenomic approaches.