What is a "Feature" exactly?

Hi

I am doing the “Moving Pictures” tutorial, but I don’t understand the “feature” meaning, and it seems to be a core concept in QIIME 2. Can you help me with that? Sorry if the question is too basic. Thank you

Hi @David_Leonardo_Lopez,
Thanks for asking! A feature is essentially any unit of observation, e.g., an OTU, a sequence variant, a gene, a metabolite, etc, and a feature table is a matrix of sample X feature abundances (the number of times each feature was observed in each sample).

We use the generic term “feature” because these features can be many different data types — but if you are familiar with OTUs (operational taxonomic units) and/or qiime1, you will probably be more comfortable with thinking of features as OTUs and feature tables as OTU tables. In qiime1 the term “OTU table” was used instead because the only observational unit really supported in qiime1 was OTUs. In QIIME2 (currently) most features will be OTUs or sequence variants, and we already support multiple observational units (OTUs via clustering methods in q2-vsearch, sequence variants via q2-dada2 and q2-deblur, metagenome features via MetaPhlAn2). Many other types of feature data can be imported into QIIME2 for analysis, and with the proliferation of different technologies and data types, stronger support for these may come in the near future. Thus, we adopt the generic term “feature”, which is much more inclusive than “OTU”.

We need to put in more work on the QIIME2 documentation, and have a number of open issues on our radar to bolster these docs in early 2018, including expanding the glossary, tutorials, and other information. Thanks for bringing this up, and for posting to the forum! :sun_with_face:

24 Likes

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.