Public application
A Shiny application for exploring microbiome data through controls and visualisations. My contribution is to organise the analytical workflow and make R tools accessible to people who primarily use a graphical interface.
The problem
Microbiome analysis combines abundance tables, taxonomy and sample metadata. Connecting these elements, choosing diversity measures and exploring group differences requires both biological judgement and statistical tools.
Andera is intended for biologists and biotechnologists who need to explore these data without writing every operation in an R console.
The solution
The application works with phyloseq objects, which bring together the data needed for analysis. It supports exploration of composition, alpha and beta diversity, ordination, networks and differential abundance.
The workflow starts with loading data and offers different perspectives on the same samples. Visualisations help formulate questions about composition and variability before interpreting group comparisons.
Development decisions
Shiny provides the interactive interface; phyloseq and vegan support part of the analytical work. The modular R structure separates exploration tasks and makes it easier to review individual components without concentrating all the logic in one screen.
The design aims to maintain a clear connection between the input data, selected options and displayed output. The interface supports analytical decisions, which continue to depend on the research question and study design.
Outcome and limits
Andera is publicly available on Posit Connect Cloud and can be used in a browser. The source code is maintained in a private repository.
Interpretation depends on metadata quality, filtering and normalisation. Visual separation in an ordination does not by itself demonstrate a group difference, and network associations do not establish causality. The number of samples and variables also affects computation time and memory requirements.