Installation

P3 uses a combination of Python and R.

Pipeline automation is handled using the Snakemake build system, and can be easily parallelized to run across multiple CPUs.

Requirements

The deploy/ directory contains requirements files for Python, R, and Ubuntu. These requirements have been included in the Docker container. To run the example in an isolated environment:

  • Install docker.

  • Pull the docker container:

    docker pull daler/p3

  • Run the docker container:

    docker run -i -t daler/p3 /bin/bash

  • You should now be in the running docker container, in the /data directory, From inside the container, clone the git repository:

    git clone https://github.com/NCBI-Hackathons/Pharmacogenomics_Prediction_Pipeline_P3.git

  • Change to the source code directory:

    cd Pharmacogenomics_Prediction_Pipeline_P3

  • Run snakemake using the prepare_example_data rule to unzip the example data into the aptly-named example_data directory:

    snakemake prepare_example_data

Do a dry run to see what needs to be run:

snakemake -n

Run the pipeline, using -j to specify the number of processors to use:

snakemake -j8

Output will be in example_data/runs/run_1/output/. The final output consists of one .RData file for each configured sample. Each file can be loaded into R, and contains a single object called outSL that contains the output of the SuperLearner run for that sample.

You can copy the results from the docker container to your machine. Easiest way to do this is to leave the docker container running, then open another terminal. Then:

# In another terminal . . .

# Get the ID of the running container. This is also part of the prompt in the running
# container, so you can just copy that if you want.
CONTAINER=$(docker ps | grep daler/p3 | awk '{print $1}')

# copy just the final output from the container
docker cp \
$CONTAINER:/data/Pharmacogenomics_Prediction_Pipeline_P3/example_data/runs/run_1/output \
final_output

tree final_output
#final_output/
#├── CX0030.log
#├── CX0030.RData
#├── CX0050.log
#├── CX0050.RData
#├── CX0051.log
#├── CX0051.RData
#├── CX0058.log
#├── CX0058.RData
#├── CX0065.log
#└── CX0065.RData
#
#0 directories, 10 files