From functional annotation to functional meaning in microbiome research.

Microbiome research has moved from cataloging community composition to asking what these communities do, but "function" is often used to describe fundamentally different levels of evidence. Functional claims may refer to functional capacity (what is encoded), functional realization (what molecular functions are actively engaged under specific conditions), or functional impact (the resulting consequences for hosts, microbial communities, or ecosystems). In this Perspective, we examine how functional annotations in microbiome research generate biological meaning and why current approaches support different kinds of inference. We propose a framework that distinguishes three levels of functional inference: capacity, realization, and impact. Homology-based, domain-centric, pathway-based, machine learning-driven, and multi-omics approaches each contribute differently to these levels, but none alone captures microbiome function in full. We argue that microbiome function should be interpreted as a hierarchy of inferences rather than a single property. Recognizing this distinction should improve how functional findings are interpreted, reported, and compared across microbiome studies. Accordingly, functional studies should explicitly state whether their conclusions concern capacity, realization, or impact, thereby clarifying the evidential basis of functional claims and improving their interpretation and comparison across microbiome studies.