Getting poster data...
Thomas SB Schmidt, João F Matias Rodrigues, Christian von Mering (Institute of Molecular Life Sciences, University of Zurich)The demarcation of Operational Taxonomic Units (OTUs) from complex sequences datasets is a key step in large-scale characterizations of microbial communities. However, it remains unclear how well the various proposed OTU clustering algorithms approximate 'true' microbial taxa, and many conceptually distinct approaches continue to be used. We explored different aspects of OTU clustering, in particular the robustness, reproducibility and ecological consistency of OTU demarcation (available online: https://doi.org/10.1371/journal.pcbi.1003594; https://doi.org/10.1111/1462-2920.12610). In a global and comprehensive survey of roughly one million microbial SSU rRNA sequences, we objectively quantified biases introduced by several widely employed algorithms. We observed surprising trends in the robustness of sequence clustering to changing parameters, and with regard to the reproducibility of OTU-based findings. Moreover, we explored the ecological consistency of OTUs – based on the assumption that, like true microbial taxa, they should show measurable habitat preferences (niche conservatism). We systematically parsed sequence annotations to obtain broad ecological descriptions of s