Introduction- Scott's script to consolidate all of the OTU calling scripts and submit them to the cluster with one command
(smile_train: Get's you where you want to go and your happy)
Works on coyote and maybe the broad, but maybe not other clusters.
You should install in lib (bin get crowded)- each person should install their own.
Make a lib directory unless this already exists:
mkdir lib
Lone this into the lib folder you just make using git:
git clone https://github.com/swo/smile_train
To get some information, you have to have this documentation for each any time this code is modified (i.e. below, this is ok for 5/12/14), you need to have sphinx-build, which isn't in on coyote. If you can't generate this, don't give up, just email Scott (swo@mit.edu).
(open the html file within the _build folder on the Desktop to get really good documentation)
Alter train.cfg (you either have to link to usearch etc) to have your new names for the following (for example I changed these):
[User]
username=spacocha
tmp_directory=/net/radiodurans/alm/spacocha/tmp
library=/net/radiodurans/alm/spacocha/lib/smile_train
bashrc=/net/radiodurans/alm/spacocha/lib/smile_train/bashrcs/coyote.sh
Then, make the tmp_directory:
mkdir /net/radiodurans/alm/spacocha/tmp
You can have a detailed description of anything by using --help.
This will source the bashrc that script that Scott made.
Submit from the head node (when you log on), and if it's long, you;re going to have to use this:
ssh coyote
screen
Now you are inside screen (screen -ls will tell you which ones you are running):
You can name the screens
screen -S SPtest
and screen reattach will give you
screen -R SPtest
you can detatch and keep running:
RUN COMMAND
contol^A D
(or type man screen to get information)