Venue

Pune, India. Dec 16-20, 2019

Learning goals

After this workshop the students will have a deeper understanding of the statistical aspects in modeling taxonomic variation in microbial communities. Commonalities between various learning methods are demonstrated by using latent variable models as a unifying framework. In addition to learning new modeling techniques, the students gain an overall understanding of the underlying statistical concepts.

Prerequisites

Targeted to researchers and PhD students actively involved in microbiome research (max. 30).

Flyer

• R / Rmarkdown
• Basic statistics
  
• For suggested reading, see the links below

Software

Install the following software before the course:

• R and Rmarkdown
• Optional: RStudio.
• Required R packages (installation script)

Rmarkdown:

• Rmarkdown tips
• Rmarkdown cheatsheet

R programming:

• R cheat sheets
• R programming basics: Base R
• R visualization: ggplot2 and R graphics cookbook

RStudio:

• RStudio IDE
• Using Github with R and RStudio

Microbiome research:

• List of R tools for microbiome analysis.

Online support

Welcome to join the course Slack channel!

• Request to join the course Slack channel
• What is Slack?

Further reading

• Molecular ecologist’s view on code sharing
• Data Analysis and Visualization in R for Ecologists by Data Carpentry
• Modern Statistics for Modern Biology. Holmes & Huber (2018)
• Microbiome Data Science. Shetty & Lahti, 2019
• Data Analysis and Visualization in R for Ecologists by Data Carpentry
• Happy Belly Bioinformatics, Mike Lee, 2019

Schedule (all 5 days follow this structure)

• Lectures (9-11:30 am). Including discussion & short break between lectures.
• Morning Tea (11:00-11:30).
• Practicals I (11:30-13:00)
• Lunch (13:00-14:00)
• Practicals II (14:00-16:00)
• Wrap-up (16-17; conclusions, time for questions & discussion)

Welcome & introductions

Lecture 1 (LL): Standard amplicon data analysis workflow

• Standard amplicon data analysis workflow PDF
• See also: Callahan et al. (2016) Bioconductor Workflow for Microbiome Data Analysis: from raw reads to community analyses. F1000Research 5:1492
• Microbiome Data Science Shetty & Lahti. 2018.

Lecture 2 (SH): Heterogeneous data and phyloseq, methods for amplicon data analysis: DADA2

• R/Bioconductor environment for high quality Statistics and Reproducible Research PDF
• DADA2 Method PDF
• DADA2 Workflow PDF

See also: - Why exact ASVs? : PDF from Ben Callahan at ASM

Practicals: Amplicon data analysis with R/phyloseq

• Optional: Reproducible research with Rmarkdown html | Rmd
• Phyloseq Lab html | Rmd
• DADA2 Workflow html | Rmd
• Extra: Tutorial on trimming primers

Lecture 1 (LL): Data properties and visualization

• The anatomy of taxonomic profiling data PDF

• Visualization Lab PDF

Lecture 2 (SH): Exploratory and Confirmatory Statistics

• Parameters, statistics, mixture models PDF

• Non standard data: trees, networks and distances PDF

• Robust testing: examples.

Practicals:

• Visualization with ggplot2: html Rmd

• Robust methods : html Rmd

• Using graphs and networks : html Rmd

References: - Book chapter on Statistical Models
- Very good video about Bayes Theorem
- Threshold Rank Transformations complete workflow
- Friedman Rafsky test with the Julia Fukuyama’s phyloseqGraphTest package

Lecture 1 (SH): Dimension reduction in the analysis and visualization of beta diversity

• Multivariate Analysis Lecture PDF

Lecture 2 (LL): Microbiome variation and individuality in population studies

• Landscape model PDF

See also:

• 10 tips for dimension reduction. Plos Comp Biol.
• Book chapter in Modern Statistics for Modern Biology
• Tipping elements of the human intestinal ecosystem Lahti et al. Nat. Comm. 5:4344, 2014.
• Intestinal microbiome landscaping: Insight in community assemblage and implications for microbial modulation strategies. Shetty et al. FEMS Microbiology Reviews, 41(2):fuw045, 2017.
• Multi-stability and the origin of microbial community types Faust et al. ISME Journal, 11:2159–2166, 2017.

Practicals: Multivariate dimension reduction visualization in R

• Microbiome data manipulation and analysis tools html | Rmd

• Lab on PCA : html and Rmd.

• Lab on rank PCA and PCoA/MDS: html and Rmd

• Lab on dPCoA: html and Rmd

• Aspects of beta diversity: feature selection, dissimilarity, ordinations

• Illustration Lab, the Enterotypes: html and Rmd

See more advanced modulated version of DPCoA by Julia Fukuyama: - agPCA package,

Lecture 1 (SH): Advanced statistical techniques in longitudinal modeling

Latent Dirichlet allocation, uncertainty quantification PDF

Testing in presence of dependent data (longitudinal and spatial), bootLong.

• Lecture about using the block bootstrap
  
• Latent Dirichlet and the need for more than simple multinomials Sankaran & Holmes (2017)

Lecture 2 (LL): Models for microbial time series

Standard dynamical models and the element of time in microbial ecology PDF

Further reading:

• Microbial communities as dynamical systems. Gonze et al. Current Opinion in Microbiology, 44:41–49, 2018.
• Signatures of ecological processes in microbial community time series Faust et al. Microbiome, 6(120) 2018.
• Metagenomics meets time series analysis: unraveling microbial community dynamics. Faust et al. Current Opinion in Microbiology, 25:56–66, 2015.

Practicals: Examples on microbiome time series analysis and visualization

• Simulating and visualizing microbial time series html | Rmd

• Testing in presence of dependent longitudinal data: bootLongdemo html
  bootLongdemo Rmd
  

• intermediary data needed, you can download directly at links:
  MBB_Sim.rds MSE_Sim.rds
  MSE_Sim_lC12.rds
  MSE_Sim_lC13.rds
  MSE_Sim_lC14.rds

• bootLong Vignette

• Testing in presence of dependent longitudinal data: bootLong

• Regime switching and dynamical models arXiv Kris Sankaran, 2017

• Treelapse and seeing the tree and the time series Article: Sankaran and Holmes, 2017
  treelapse Tutorial

Lecture 1 (SH): Multidomain data integration

• Multiway data and multiomics data integration Slides

• Review article: Sankaran and Holmes, Frontiers in Genetics, 2019
  Multitable Methods for Microbiome Data Integration

• Associated Code in github repository

Lecture 2 (LL): Overview of the week and additional topics

• Contemporary and future perspectives on statistical microbiome research PDF

Practicals: Examples on multi-omics techniques

• Statistical rethinking and Bayes pdf

• Mantel, sparse CCA and RV html
  
• Mantel, sparse CCA and RV Rmd - Link to data:microbe.rda
  

• Link to data:metabolites.csv

Feedback

• Feedback form

• Q & A session

Concluding remarks