This document discusses statistical and visualization methods for analyzing metagenomic data. It introduces several R/Bioconductor packages for metagenomic analysis, including metagenomeSeq for differential abundance analysis of 16S data and metagenomicFeatures for annotating 16S features. It also describes msd16s example data. Additionally, it discusses the benefits of R/Bioconductor including infrastructure objects, documentation, and reproducibility. Finally, it introduces Metaviz, an interactive browser-based tool for exploring hierarchical metagenomic data through integration and visualization of multiple data sources.

1. Statistical and Visualization Methods for Metagenomic Analysis
Héctor Corrada Bravo
Center for Bioinformatics and Computational Biology

2. • metagenomeSeq
– 16S differential abundance
– R/Bioconductor infrastructure for
metagenomic assays
– Longitudinal data
• metagenomicFeatures
– Incipient attempt regularizing 16S feature
annotations in R/Bioconductor
– E.g., greengenes13.5MgDb
• msd16s
– Example data, as infrastructure object

3. R/Bioconductor Strengths
• Infrastructure objects
– Interoperability, speed up startup time for method development
• Strict development practices
– Documentation, use cases, vignettes
• Annotation infrastructure
– Again, interoperability across experiments and data types
• Exploratory analysis
• Reproducibility
– Vignettes, Rmarkdown, etc.
• Recently, exploratory and interactive visualization
– Shiny, epiviz

4. Integrative, visual and computational
exploratory analysis of genomic data
• Browser-based
• Interactive
• Integration of data
• Reproducible dissemination
• Communication with R/Bioconductor: epivizr package
software systems to support creative exploratory analysis of large genome-wide datasets...

5. • Computed Measurements: create new measurements from
integrated measurements and visualize

6. • Summarization: summarize integrated measurements
(computed on data subsets)

7. Dynamically extensible: Easily integrate new data sources, data
types and add new visualizations.
Data providers define coordinate
space

8. One interpretation of Big Data is many sources of relevant
contextual data
• Easily access/integrate contextual data
• Driven by exploratory analysis of immediate
data
• Iterative process
• Visual and computational exploration go
hand in hand

9. Visualization design goals
Context
• Integrate and align multiple data sources;
navigate; search
• Connect: brushing
• Encode: map visualization properties to
data on the fly
• Reconfigure: multiple views of the same
data

10. Visualization design goals
Data
• Select and filter: tight-knit integration with
R/Bioconductor
• (current work) filters on visualization
propagate to data environment
Model
• New 'measurements' the result of
modeling; suggested by data context

11. Metagenomic Visualization
• How to effectively navigate large datasets
where features are organized hierarchically?
• Metaviz: browser-based, interactive
exploratory analysis of metagenomic
data
• Connection to R/Bioconductor with
metavizr package
• Built on metagenomeSeq and
metagenomeFeatures infrastructure

15. Metaviz
• Exploration of hierarchically organized
features
• Geared towards 16S for now
– Hierarchical organization relevant to WGS
• Integration is a big part of design
– Framework designed for data integration

16. Acknowledgements
Brianna Lindsey, O. Colin Stine, Owen White, Anup Mahurkar: University of Maryland Baltimore
Jim Nataro: University of Virginia
NIGMS, Genentech
Florin Chelaru
(now @ MIT)
Joseph Paulson
(now @ Harvard)
Mihai Pop
(@ UMD)