search

AlexsLemonade / OpenPBTA-analysis

The analysis repository for the Open Pediatric Brain Tumor Atlas Project
Other
99 stars 66 forks source link

Install MutSigCV on the project Docker image #546

Open jaclyn-taroni opened 4 years ago

jaclyn-taroni commented 4 years ago

Splitting up #529 into more focused tickets. This is related to #517.

We are interested in trying out MutSigCV for this project. I am recording what I've found so far. Here's where we can download MutSigCV from: https://software.broadinstitute.org/cancer/cga/sites/default/files/data/tools/mutsig/MutSig2CV.tar.gz

MutSigCV requires MATLAB R2013a Runtime to be installed, which we can probably install with:

RUN mkdir /mcr-install-v81 && \
    mkdir /opt/mcr && \
    cd /mcr-install-v81 && \
    wget https://ssd.mathworks.com/supportfiles/MCR_Runtime/R2013a/MCR_R2013a_glnxa64_installer.zip && \
    unzip -q MCR_R2013a_glnxa64_installer.zip && \
    rm -f MCR_R2013a_glnxa64_installer.zip && \
    ./install -destinationFolder /opt/mcr -agreeToLicense yes -mode silent && \
    cd / && \
    rm -rf mcr-install-v81

Note I'm extrapolating from the other things we've done so far, this is untested!

The instructions from the MutSigCV README.txt for setting up LD_LIBRARY_PATH are:

Once the runtime is successfully installed, you must add it to your LD_LIBRARY_PATH. You will likely want to add the following lines to your .bashrc/.cshrc, so that the MATLAB runtime is always on your path.

For a bash shell:

mcr_root=<path to runtime you specified when installing>
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$mcr_root/bin/glnxa64/
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$mcr_root/sys/java/jre/glnxa64/jre/lib/amd64
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$mcr_root/sys/java/jre/glnxa64/jre/lib/amd64/server
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$mcr_root/sys/java/jre/glnxa64/jre/lib/amd64/native_threads
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$mcr_root/sys/os/glnxa64
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$mcr_root/bin/glnxa64
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$mcr_root/runtime/glnxa64
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$mcr_root/lib

where mcr_root should be /opt/mcr/v81. We're going to have multiple versions of MCR installed, so I'm inclined to deal with the setting and unsetting of LD_LIBRARY_PATH in the bash scripts that run a particular tool – here's the GISTIC example:

https://github.com/AlexsLemonade/OpenPBTA-analysis/blob/56f82b9f23d369a5e208c7564ce8fc22fa0b64dd/analyses/run-gistic/scripts/run-gistic-openpbta.sh#L5

and

https://github.com/AlexsLemonade/OpenPBTA-analysis/blob/56f82b9f23d369a5e208c7564ce8fc22fa0b64dd/analyses/run-gistic/scripts/run-gistic-openpbta.sh#L68

jaclyn-taroni commented 4 years ago

This is beyond the scope of this particular ticket but MutSigCV related things that I also found that could be helpful (with minimal commentary): I happened upon this Biostars thread and one of the answers mentioned CovGen and the CovGen docs state:

MutSig provides a "territory" table (exome_full192.coverage.txt) for times when detailed coverage information is not available for each sample in your cohort. This coverage file may not properly represent the target space utilized by your capture kit and can adversely affect the results of your MutSig analysis. CovGen bridges the gap between detailed sample level coverage information and the exome_full192.coverage.txt table that MutSig provides with a target specific full coverage table.