AppliedBioinformatics / runBNG

An easy way to run BioNano genomic analysis
MIT License
27 stars 7 forks source link

Does runBNG now requires docker? #34

Open ryao-mdanderson opened 2 years ago

ryao-mdanderson commented 2 years ago

Hello @yuxuanyuan and the support team:

I am trying to install runBNG 2.0.1 on our institution research HPC, a RHEL 7.7 cluster. Per README:

"BioNano Sovle (containing Bionano Pipeline and RefAligner/Assembler) needed by runBNG version 2 and above can be downloaded ..... "

After I downloaded BioNano Solve on the cluster, I refer to the installation instruction https://bionanogenomics.com/wp-content/uploads/2017/10/30182-Bionano-Tools-Installation-Guide.pdf (This is suggested in Bionano website)

One installation step is "python bng-install all", I reviewed the script bng-install and found out the script invokes to bring docker deamon and use docker images... Since our HPC does not support docker at this point, my understanding is we can't install runBNG version 2.0 and above on our HPC cluster, right?

Thank you! Rong

ryao-mdanderson commented 2 years ago

Although docker is not one of the prerequisites ...

yuxuanyuan commented 2 years ago

Hi @ryao-mdanderson,

You can unzip the dowloaded access.tools.tgz file and then find tools --> pipeline --> 1.0.

The bionano Solve folder is the 1.0 mentioned above. I guess you don't need to install BionanoSolve for runBNG. What you need are the python packages, perl and R packages (asssume you have installed python2 and R). After the steps i mentioend, you can try runBNG.

Cheers, Andy

ryao-mdanderson commented 2 years ago

Hi @yuxuanyuan 👍 Thank you for your reply.

After I download and unpack access.tools.tgz, The tools/pipeline/1.0 folder has the following contents $ ls bionano_packages FSHD Process_Control_Datasets SMSV VCFConverter cohortQC HybridScaffold RefAligner SVviewer EnFocus_Repeats Pipeline RefGenome VariantAnnotation

Does the above contents all I need? Do I not need to refer to the installation instruction https://bionanogenomics.com/wp-content/uploads/2017/10/30182-Bionano-Tools-Installation-Guide.pdf to install Solve, right? i.e. Do not run bng-install script --> so, Docker is not needed?

I was confused, the latest runBNG version is 2.0.1, however, per https://github.com/AppliedBioinformatics/runBNG README.md "BioNano Sovle (containing Bionano Pipeline and RefAligner/Assembler) needed by runBNG version 2 and above can be downloaded at https://bionanogenomics.com/support/software-downloads " As the download page refers to review pdf file to install BioNano Solve (see my original post question).

The dependency python (3.7.3) and R (4.0.0) packages are ready on our system. If BioNano Solve installation can be ignored, I certainly can try to test runBNG.

Best Reagrds, Rong

yuxuanyuan commented 2 years ago

Hi @ryao-mdanderson,

runBNG only needs the folder (tools/pipeline/1.0) and the content inside. runBNG is not from Bionano but it relies on its 'BionanoSolve' package (package only, no need to follow Bionano's instruction to install).

You can execute runBNG directly if the original BionaoSolver folder is provided and the relative python packagedateutil, lxml and R packages data.table, igraph, intervals and argparser are installed.

Please also check if perl is properly installed in your system.

Cheers, Andy

ryao-mdanderson commented 2 years ago

@yuxuanyuan

Thank you again for the information. This helps!

Happy New Year! Rong

ryao-mdanderson commented 2 years ago

@yuxuanyuan

I hit one issue during the testing.

I download the Solve package as service account ris_hpc_apps, as you see the following, Solve package is in pipeline, $ ls -l /risapps/rhel7/runBNG/2.0.1/tools/pipeline total 0 lrwxrwxrwx 1 ris_hpc_apps rists 22 Oct 28 17:12 1.0 -> Solve3.7_10192021_74_1 drwxr-xr-x 15 ris_hpc_apps rists 315 Oct 28 17:12 Solve3.7_10192021_74_1

I am able to successfully tested one function fa2cmap using this account by running the following command /risapps/rhel7/runBNG/2.0.1/runBNG fa2cmap -f /risapps/rhel7/runBNG/2.0.1/Examples/Test.fa -o /risapps/rhel7/runBNG/2.0.1/ -e BspQI -s /risapps/rhel7/runBNG/2.0.1/tools/pipeline/1.0 -- it generated 4 output files Test_BSPQI_20kb_5labels_key.txt Test_BSPQI_20kb_5labels.cmap Test_BSPQI_20kb_5labels_summary.txt status.txt

However, if I wish to run the test not as service account, i.e. as a regular user, $ /risapps/rhel7/runBNG/2.0.1/runBNG fa2cmap -f /risapps/rhel7/runBNG/2.0.1/Examples/Test.fa -o $HOME -e BspQI -s /risapps/rhel7/runBNG/2.0.1/tools/pipeline/1.0 where $HOME is a regular user's home directory

I hit the error message: "Oops! It seems the directory '/risapps/rhel7/runBNG/2.0.1/tools/pipeline/Solve3.7_10192021_74_1' is not existent or writable, please check!"

I tried to set up symbolic link at user's home directory ln -s /risapps/rhel7/runBNG/2.0.1/tools/pipeline/1.0 Solve or ln -s /risapps/rhel7/runBNG/2.0.1/tools/pipeline/Solve3.7_10192021_74_1 Solve

then run a modified command: /risapps/rhel7/runBNG/2.0.1/runBNG fa2cmap -f /risapps/rhel7/runBNG/2.0.1/Examples/Test.fa -o $HOME -e BspQI -s $HOME/Solve

Still no luck , The message is Oops! It seems the directory '$HOME/Solve' is not existent or writable, please check! -- where $HOME is the user's home directory.

What is the best approach to allow users to share using the command installation/download directory? Any suggestion?

Thank you very much! Rong

yuxuanyuan commented 2 years ago

Hi @ryao-mdanderson,

You may try chmod -R 777 or chmod -R 766 on Solve3.7_10192021_74_1 under /risapps/rhel7/runBNG/2.0.1/tools/pipeline using your service account.

If the error is still there, you may try chmod -R 777 /risapps/rhel7/runBNG/2.0.1/tools/pipeline/Solve3.7_10192021_74_1

Hope this helps.

Cheers, Andy

ryao-mdanderson commented 2 years ago

Hi @yuxuanyuan 👍 Thank you again!

I am afraid of changing mode to user writable in a a cluster shared directory, so I copied Solve3.7_10192021_74_1 (6GB only) to user directory. Now I can successfully to run the above testing command. I appreciate your suggestions!

Regards, Rong