Simple bash scripts that are useful
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
cluster-bin
.gitignore
LICENSE
README.md
filetypes-pics-models-list.txt
gr
grsed
poorquery
sync-from-scc-conflux
sync-pics-models-only-to-local
sync-to-scc-conflux

README.md

bin

Simple bash scripts that are useful, especially on the SCC cluster

Install

  • Note that you will have to personally fill in anything in angle-brackets (i.e. < and >) in these scripts to personalize them.
  • The scripts in "cluster-bin" are intended for use from inside the cluster, whereas everything else is intended for local usage.
  • Both the "cluster-bin" and other scripts assume you install them into a folder in $HOME/bin, but this is only important for locating the filetypes-pics-models-list.txt files for use in copy-pics-models and sync-pics-models-only-to-local. This is also easily changed if you want to put them somewhere else.

What they do

  • cluster-bin/backup-project-to-nb: Backs up your personal "project" folder into your personal "projectnb" folder.

  • cluster-bin/copy-pics-models: Copies ONLY the filetypes in filetypes-pics-models-list.txt from your personal "projectnb" data to a pics-models-only folder in your personal "project" folder.

  • cluster-bin/grsed: Replaces all instances of the first string argument with the second string argument in ALL files in ALL subfolders from the current working directory. Also gives you a warning listing the number of changes before it runs!

  • cluster-bin/lastsim: Prints the last 10 lines of the most recent sim_job<N>.out file in all subfolders of your batchdirs folder to terminal, so you can see what the latest status is. This also indicates whether the associated sim_job<N>.err standard error file is empty or not. Note that this has not been tested when there are very many sim_job<N>.out files.

  • gr: A trivial alias for using grep to find all instances of a string in all files in all subfolders from the current working directory.

  • grsed: Same as the version in cluster-bin/grsed.

  • poorquery: Matches regex-identified png-files (or other extensions if you want) anywhere on a filesystem and outputs an HTML file displaying them.

  • sync-from-scc-conflux: Syncs everything FROM inside a folder on the SCC called scc-conflux TO a local version of that folder, but does not delete any files (can be made to do so).

  • sync-to-scc-conflux: Syncs everything FROM a local version of the folder scc-conflux TO a folder on the SCC of the same name, but does not delete any files (can be made to do so).

  • sync-pics-models-only-to-local: Copies ONLY the filetypes in filetypes-pics-models-list.txt FROM the pics-models-only folder in your personal "project" folder on the SCC TO a copy of the folder on your local machine.

My (former) workflow

What I used to do was:

  1. Run all my simulations, saving everything (including flat PNGs saved as a result of passing dsPlot to dsSimulate) into a folder on my personal "projectnb" directory. I would only save the results/plots, NOT the data, since I didn't need to run secondary analyses.
  2. Even if I did save the data, I would then use cluster-bin/copy-pics-models to save only the plots/metadata of my recent simulations to my personal "project" directory.
  3. I would then use sync-pics-models-only-to-local to grab an offline copy of all the plots.
  4. Finally, I would use poorquery to quickly "slice" the plots I wanted to see. However, this requires that all of the varied parameters and their values be embedded in the filenames for ALL plots, which DynaSim does not do by default anymore.