omicverse-microbiome-meta-analysis

Cross-cohort meta-analysis merges per-study microbiome AnnData objects at a shared taxonomic rank like genus, aligning features across cohorts to produce one statistically coherent table of shared biomarkers instead of isolated findings.

Differential abundance pooling runs per-study analysis and combines effect sizes via inverse-variance fixed-effects or random-effects models, outputting a ranked table with combined log fold-change, standard error, FDR, and study counts.

Heterogeneity assessment inspects Cochran I2 and per-study counts to determine whether a differential abundance signal truly replicates across cohorts or is driven by a single study.

Yes, you can compare wilcoxon, deseq2, and ancombc meta workflows to evaluate which method best recovers consistent cross-cohort signals during the differential abundance pooling process.

Yes, cross-cohort harmonization merges per-study microbiome AnnData objects at a shared taxonomic rank such as genus, specifically applying to 16S studies requiring feature alignment before pooling.

The meta-analysis returns a ranked table containing combined log fold-change, combined standard error, p-value, FDR, number of studies, and I2 heterogeneity diagnostics for the evaluated biomarkers.