Skip to content

miaTime

Microbiome time series analysis — a Bioconductor package providing tools for longitudinal microbiome studies on top of the (Tree)SummarizedExperiment infrastructure. Sister to the broader mia framework.

Talk at EuroBioC2025

  • Speaker: Geraldson T. Muluh
  • Day / session: Day 1 (Wed Sep 17, 2025) — morning short-talk session, after the Crowell keynote
  • Talk title: "miaTime framework for longitudinal microbiome studies"
  • Slides (Zenodo DOI): to verify after publish
  • Video: not recorded
  • Abstract / schedule: eurobioc2025 schedule

What it does

miaTime layers time-aware analyses (e.g. trajectories, time-resolved dissimilarity, change-point detection) on top of mia's TreeSummarizedExperiment containers. The split keeps the core mia package focused on cross-sectional analyses while letting longitudinal users opt in to the time-series surface separately.

How it works

Core idea. Time-aware operations on a TreeSummarizedExperiment: per-subject samples are ordered by a time column, then beta-diversity (Bray-Curtis by default) is computed either against a reference baseline sample (addBaselineDivergence) or between consecutive timepoints (addStepwiseDivergence). The output is a per-sample divergence score appended to colData so it can be modeled, plotted, or used to flag change-points.

Inputs / outputs. Input: a TreeSummarizedExperiment whose colData includes a time-of-sampling column and a subject/group column. Output: the same object with new colData columns for divergence, time-difference to reference, and reference-sample IDs.

Key innovation. Keeping longitudinal microbiome methods inside the same (Tree)SummarizedExperiment substrate as the rest of mia, rather than spinning out a parallel data class — so a user can move from cross-sectional mia analysis to time-resolved analysis without reshaping data.

Parameters worth knowing. time.col — colData column with sampling time. group — subject/group identifier (divergence is computed within group). name — prefix for the new colData columns. The dissimilarity metric is inherited from mia::getDissimilarity (Bray-Curtis default).

Canonical example. Vignette uses the hitchip1006 human-gut dataset: addStepwiseDivergence(tse, time.col = "time") appends a Bray-Curtis dissimilarity between each sample and its predecessor in the same subject's series. The resulting colData column can be plotted against time to surface periods of rapid community change versus stability.

Where it fits in the corpus

  • AACR 2026: no current dossier
  • mia (entry) — base package; miaTime extends it for time-series studies
  • GBCC 2025: queued

Notes

Borman maintains both mia and miaTime (and presented the mia workshop on Day 2). The two-package split mirrors the Bioconductor convention of keeping core data classes separate from method extensions.