Rédiger un PREreview

Microbiome phenotype analysis usually captures broad organism-level traits, yet clinically and ecologically important programmes for trace-element acquisition, storage, detoxification, and cofactor biosynthesis remain difficult to summarize because the underlying loci are often strain-variable and only partly reflected by taxonomy. We present TracePheno, a function-first framework for inferring microbial phenotypes related to eight common trace elements from gene- or KO-level evidence. The current phenotype panels span iron, zinc, manganese, copper, cobalt/vitamin B ₁₂ , nickel, molybdenum, and selenium. The implementation combines curated core/accessory/ambiguous marker tiers, cohort-invariant bounded support transforms, deterministic core-gated calling, presence/absence-oriented genome trait inference, and a publicationoriented visualization bundle. The bundled release covers ten phenotype panels and three complementary workflows: direct scoring of function matrices, genome-to-trait construction, and taxon-abundance scoring with a precomputed trait matrix. Using the current release, we analysed two local demonstrations that were regenerated for this manuscript. In 11 representative human-gut genomes from the MGnify catalogue, copper homeostasis/resistance and iron acquisition were the most prevalent high-scoring programmes, whereas Firmicutes in this small panel showed stronger cobalamin biosynthesis and selenium-utilization signals than Bacteroidota. In a PICRUSt2-style KO example, zinc acquisition was higher in the case group, whereas iron acquisition, corrinoid transport/cobalt uptake, and selenium utilization were relatively higher in controls. Together, these analyses show that TracePheno can convert genome annotations and predicted KO tables into interpretable, publication-ready trace-element phenotype landscapes while keeping the decision rules explicit, portable, and biologically constrained.