readme.NSG

This is the readme for configuring and running simple circuit
simulations on the NSG for the manuscript under preparation

Short et al. 2015

To configure the simulation to run edit these files in 1) and 2) and
then follow step 3) etc.:

1) create_arrays.py: first edit create_arrays.py to assign values of
the breathing (B) and light-stimulated (S) event peak rates and also
the number of additional columns (add_columns python list). The
"additional" refers to that there is always one glomerular column
containing the mitral cell which has its voltage recorded analagous to
the wetlab experiments.

2) pre_init.py: The basic idea with pre_init.py is that it creates two
python lists "parameters", and "columns", each of length equal to the
number of simulations.  By copying verbatim corresponding elements of
these lists, a parameter.hoc and a num_of_columns.hoc file is written
in each simulation folder run_X, where X=1,2,3, ...,
number_of_simulations.  For example: the run_0 folder gets a
parameter.hoc file assigned from parameters[0] and num_of_columns.hoc
file from columns[0]. The number of simulations is figured out
automatically by taking the create_arrays data into consideration with
a net_type list: Set the net_type list (found near the top of this
program) to something like

net_type=['full_net','pg_net','gc_net']

These variables will trigger network simulations with these types
being created, with each of these network types having the commands to
make them present in the parameters.hoc file.  (See the code that
creates "parameters" elements with an append and then modifies these with
indexed assignments).

Also edit the lines in the parameters.append(""" section which will
look something like:

breath_peak_rate = %d
light1_peak_rate = 0
light2_peak_rate = %d

so that the second %d is either on light1 or light2 depending on
whether you want the light stimulation (S) to fall either on the
primary column being recorded from (light1) or adjacent column(s)
(light2).

3) Run pre_init.py by typing ./pre_init.py on the command line.  run_X
simulation folders will be created.  See what the highest number of
the X is. This is one less than the number of simulations since X
starts at 0.  Check that this number of simulations is present in the
line in init.py that checks id's, for example:

if id<2888:

(This checking is present because there is no gaurentee that you can
ask for the same number of processors (nodes*cores_per_node) as there
number of simulations.)

4) zip the folder containing all these files into an input file for
the NSG.

5) Sign onto the NSG and follow the directions in its help to create a
data folder containing the input file created in step 4) and a task
folder where you select NEURON with Python on a machine that contains
at least as many processors as there are number of simulations as
determined in step 3) (probably stampede which note has 16 cores per
node).  Set the number of nodes to be such that the number of nodes
times the number of cores per node is the smallest number greater than
the number of simulations (there will be left over processors with nothing to do).

6) save and run the task!