Data-driven hypothesis is a resource for identifying novel functions for human genes developed by the Hirschey Lab. A typical use case starts by querying a gene, identifying genes that share similar patterns or behaviors across various measures, in order to discover new novel genes in established processes or new processes for well-studied genes.
Visit www.datadrivenhypothesis.org to query genes, generate data, and download reports.
Visit www.datadrivenhypothesis.org to view detailed methods for this project. Alternatively, after generating data files, knit code/methods.Rmd for the complete methods summary.
The .R scripts in the /code directory are used to generate files in the /data directory.
If you have R, the required libraries installed you can generate the data by running:
mkdir data
Rscript code/generate_data.R
If you have an NCBI API key this can speed up portions of the data generation process. Before running Rscript code/generate_data.R export your NCBI API key to an environment variable named ENTREZ_KEY like so:
export ENTREZ_KEY="your-key-here"
On systems that support the singularity container runtime, we recommend using our container image to run the data-generating R scripts with the required dependencies.
Generating data under singularity is also demonstrated in the submit-slurm-jobs.sh. It requires a config.sh script. This config script has three optional environment variables: ENTREZ_KEY, DDH_EMAIL, DDH_UPLOAD_RESULTS.
An example config.sh script that will email you about job progress and upload the results looks like so:
export ENTREZ_KEY="your-key-here"
export DDH_EMAIL="[email protected]"
export DDH_UPLOAD_RESULTS="TRUE"
This DDH_UPLOAD_RESULTS requires you to create a .ddsclient configuration file in your home directory on the cluster. See instructions to setup a DukeDS config file.
to set the ENTREZ_KEY environment variable. You can also speicfy a DDH_EMAIL email address to be emailed with job progress.
This script can be run like so from a slurm cluster:
./submit-slurm-jobs.sh
This script will submit multiple jobs that will run one after another with differing requirements. When this script finishes it will upload the results to DukeDS service.