16S Microbiome Analysis

This repository is a collection of notes and scripts for 16S microbiome analysis in the Gloor lab written by Jean Macklaim

Before getting started

You will need some familiarity and comfort using command line and R

You need to know the details of your experiment. You should have a metadata table of all the information you've collected about your samples including clinical (age, weight, conditions, etc.) and procedural (extraction dates, amplification quantification, barcodes used, etc.)

Getting started

You should use the miseq workflow to prepare your OTU table from your Illumina reads https://github.com/ggloor/miseq_bin (scroll down for the README)

This workflow will take your V4 (or other variable region) sequencing output and generate an OTU table with taxonomic annotations

Take note of the log file for information on program versions and parameters used

What to do with your OTU table

At the end of the miseq workflow you will have a counts table with extimated taxonomy: usually called

td_OTU_tag_mapped_lineage.txt

You will want to explore your data. This will be some combination of multivariate toolsets like QIIME (UniFrac) analysis and CoDa (Compositional Data) Analysis.

• UniFrac is an analysis of abundance
• CoDa is an anlaysis of variance

Other things you'll probably want to do:

• Plotting: use R. Including barplots, boxplots. Some of my example R scripts - HERE
• Data manipulation: you'll likely want to add/drop samples, filter OTUs, and change you data in some way for final analysis. This repository has some example R scripts to do some of these common tasks - e.g. HERE
• Differential analysis: you'll want to use ALDEx2 - example HERE

Exploratory data analysis

Budget more time than you think you need for your data analysis. As the protocols are constantly changing, and each dataset has it's own unique issues, you will spend a lot of time adjusting and re-running parameters. No workflow is set in stone. What you did last week will likely not be the correct thing to do this week!

It's highly recommended you spend time in the Gloor lab to examine your data. Every dataset is different and will have its own unique problems. We're here to help before you dig yourself into a pit you can't climb back out of....

---

Some of Jean's notes and examples to get started

My branch of the miseq workflow

may be different (i.e. more up to date) from Greg's master branch
https://github.com/ggloor/miseq_bin/

Example scripts in R, Perl, Bash, QIIME, etc.

https://github.com/mmacklai/example-scripts

Compositional Data Analysis (CoDa):

https://github.com/mmacklai/example-scripts/tree/master/CoDa

ALDEx2:

https://bioconductor.org/packages/release/bioc/html/ALDEx2.html
https://github.com/mmacklai/example-scripts/tree/master/aldex2

Some of Greg's resources

https://github.com/ggloor/compositions