Rarity of microbial species: In search of reliable associations.

摘要: The role of microbial interactions in defining the properties microbiota is a topic key interest ecology. Microbiota contain hundreds to thousands operational taxonomic units (OTUs), most them rare. This feature community structure can lead methodological difficulties: simulations have shown that methods for detecting pairwise associations between OTUs, which presumably reflect interactions, yield problematic results. performance association detection tools impaired when there high proportion zeros OTU tables. Our goal was understand impact rarity on associations. We explored utility common statistics testing associations; sensitivity alternative measures; and network inference tools. found large associations, especially negative cannot be reliably tested. constraint could hamper identification candidate biological agents used control rare pathogens. Identifying testable serve as an objective method filtering datasets lieu current empirical approaches. trimming strategy significantly reduce computational time needed infer networks quality. Different possibilities improving analysis within are discussed.