You should now use https://github.com/drpowell/degust
Access a public web service running Degust.
View a short video of the interface in use.
Read a summary on the Degust home page.
This can happen when using the backend of Degust. Degust uses voom to perform the expression analysis. Voom adds a small constant (0.5), to each count, normalizes for library size, then takes the log. This means when you have a count of 0 across all samples, but different library sizes, it is possible to compute a non-zero fold-change.
We recommend setting Min read count on the configuration page to a small value, say 10.
This is the minimum number of reads required in at least one sample to keep the gene in the analysis. That is, a given gene is omitted if the number of reads across all samples is below this setting.
This appears to be a problem with Chrome when there are many (thousands) of genes in the table. We suggest using Firefox if this happens.
The MDS plot is only available when you have included "count" columns
If you do not want to use the public Degust installation, you may install your own.
You first need to grab a copy of Degust.
git clone git@github.com:Victorian-Bioinformatics-Consortium/degust
Degust can be installed in two ways:
To use the frontend visualisation, you will need to have done your own DGE analysis with a tool like edgeR or voom. You will need CSV file contain a line per gene, and the following columns:
The simplest approach is to download degust.py then run it with your csv file as a parameter. This will create a single HTML page that you view or share. Run `degust.py --help
to find the parameters to specify the column names for your CSV.
There are some javascript tests which can be run locally (or with travis). Build the js:
./build-tests.sh
Then you can either run the tests in your browser (navigate to http://localhost:8000/)
(cd tests/js/ ; python -mSimpleHTTPServer)
Or, if you have phantomjs installed you can run the tests from the command line: ./test-js.sh
Feel free to contribute with pull requests, bug reports or enhancement suggestions.
For building from sources, you will need nodejs and the following modules.
npm install -g browserify
npm install -g clean-css
npm install hbsfy@1.3
npm install handlebars-runtime
npm install uglify-js
npm install coffeeify
# Builds files into build/ for deployment
./build.sh prod
This will watch the js & coffeescript files and rebuild CoffeeScript as needed.
npm install -g watchify
./build.sh dev
./build-watchify.sh &
(cd build ; ../server.py)
To build the degust.py script for embedding csv into an html file for local
./build-embed.sh local
It is also useful to access the pages with "debug=1" (eg. http://vicbioinformatics.com/degust/compare.html?code=example&debug=1) which enables extra debug logging to the console.
The above production build only includes the front-end. To also build the back-end you can use the following. (The haskell library requirements are not well documented yet.)
./build.sh prod-server
Requirements:
The resulting build/ directory can then be installed as a CGI site.
These specific steps are known to work on an ubuntu 14.04 install.
Degust is released under the GPL v3 (or later) license, see COPYING.txt