Docker files and descriptions on how to run interactive and batch analysis. Useful as a reference for developing and testing new analysis.
-
Clone the test-data into a newly created data folder within this repository
git clone https://github.com/FASTGenomics/test-data data
Alternatively, supply your own test data - but remember to use the proper FASTGenomics directory structure (e.g.
./data/dataset_0001/my_dataset.loom
) and provide adataset_info.json
. Update the docker-compose file with the path to the data set (should point to a directory containing thedataset_0001
folder in the example above). -
Clone your favorite analysis (
analysis.ipynb
) into theanalysis
folder, e.g.,git clone https://github.com/FASTGenomics/analysis_empty_scanpy analysis
You can use any other path on your local machine, just remember to update the corresponding
docker-compose.yml
path. All the data in the analysis folder will be available in your analysis. Our available analyses can be found here -
If you use an existing image from dockerhub specify it in the
docker-compose.yml
.Alternatively, if you want to build and test an image - clone it to an
image
folder and uncomment thebuild: ...
option indocker-compose.yml
.The latest versions of our standard images (
fastgenomics/jupyter-scanpy
andfastgenomics/jupyter-seurat
) can be found on dockerhub.If you want to develop your own images make sure to always use the latest
fastgenomics/jupyter-base
version as a strating point.
If you are developing an image you will need to re-build it every time you make some changes to the images source code. To do this run
docker-compose build
To run an analysis in a batch-mode (non-interactive) simply run
docker-compose up batch
This will generate an output file analysis/analysis.html
.
To run an analysis in an interactive mode use
docker-compose up interactive
This will start an interactive jupyter under port 8886.