Data analysis & differential expression

We’ll be using edgeR to do the basic differential expression analysis of our counts.

To run edgeR, you need to write a data loading and manipulation script in R. In this case, I’ve provided one – lung_saliva.R. This script will load in two samples with two replicates, execute an MA plot, and provide a spreadsheet with differential expression information in it. To download it, click here.


Running edgeR on a data subset

To run the script on the HPC, download the script and data files for the 100k read subsets. At the command line, do:

cd ~/rnaseq
curl -O
curl -O
curl -O
curl -O
curl -O

Note: the saliva_repl1_counts.txt and lung_repl1_counts.txt are the files produced by Mapping reads to the transcriptome with TopHat and Processing another sample with TopHat and HTSeq, respectively. I’ve also run them on the replicate data sets, which produced the *repl2_counts.txt files.

Next, to run the R script, do:

module load R
Rscript lung_saliva.R

This will produce two files, edgeR-MA-plot.pdf and edgeR-lung-vs-salivary.csv; they will be in your rnaseq folder in your home directory on the HPC. The CSV file can be opened directly in Excel; you can also look at it here. It consists of five columns: gene name, log fold change, P-value, and FDR-adjusted P-value.


Functional and network analysis on differentially expressed genes

There are a number of sites that let you analyze gene lists, including DAVID. I have gotten a number of recommendations for PANTHER. PANTHER lets you upload gene lists and explore their functional categories interactively or in a gene list/annotation format.

More specifically, you can explore your RNAseq data with

  • functional classifications, in pie chart or in list;
  • tests for statistical overrepresentation;
  • tests for statistical enrichment based on associated fold change.

You need to crop and transform the data a little bit before using the functional classification. The steps are:

  1. Download the CSV file. If you’ve produced your own, copy it over from the HPC.
  2. Open it in Excel.
  3. Choose an FDR cutoff (suggest FDR < 0.05 or lower) and delete all the rows after that (or, copy the rows into a new spreadsheet – might be quicker).
  4. Save as a “Tab-delimited text.” (Note, on Mac OS X, you may need to save this as “Windows formatted text” instead.)

This is now a file you can upload to PANTHER.

(You can download my copy of this here)

To do the analysis, go to In box 1, select “Choose file” and find the CSV file you want to upload to PANTHER. Nothing else in box one should be changed.

In box 2, select “Homo sapiens.”

In box 3, select either “Functional analysis classification viewed in gene list” or “Functional analysis classification viewed in pie chart.”

Click submit.

Now you can explore the results!


Next: Submitting jobs to the MSU HPC queue

LICENSE: This documentation and all textual/graphic site content is licensed under the Creative Commons - 0 License (CC0) -- fork @ github. Presentations (PPT/PDF) and PDFs are the property of their respective owners and are under the terms indicated within the presentation.