Merge pull request #194 from Neurosim-lab/development

PR from development to master - VERSION 0.6.7

.gitignore

@@ -39,3 +39,5 @@ umac
 /examples/NeuroMLImport/LEMS_SimpleNet_pynn.py
 /examples/NeuroMLImport/LEMS_SimpleNet_pynn.py_main.json
 /nb-configuration.xml
+/examples/sandbox/Sim_sandbox.PYR1.v.dat
+/examples/sandbox/Sim_sandbox.PYR2.v.dat

.travis.yml

@@ -50,10 +50,13 @@ script:
     - cd ../M1
     - $NEURON_HOME/bin/nrnivmodl
     - python M1_run.py -nogui
-
     # - cd ../RL_arm
     # - $NEURON_HOME/bin/nrnivmodl
     # - python main.py
+
+    - cd ../..
+    # Run all OMV tests...
+    - omv all -V
 
 
 notifications: 

CHANGES.md

@@ -1,3 +1,18 @@
+# Version 0.6.7
+
+- Added gridSpacing option to create populations with cells placed in grid with fixed spacing
+
+- Added netParams functions: addCellParamsWeightNorm(), saveCellParamsRule() and loadCellParamsRule()
+
+- Added option to importCell based on instantiated cell (not just template or class)
+
+- Modified importCellParams to store mod mechanism global variables in cellRule['globs']
+
+- Added support to export gap junctions to NeuroML
+
+- Fixed bug during import that set section name to 'soma' if only 1 section
+
+
 # Version 0.6.6
 
 - Batch parameters can now include list or dict items

doc/source/code/mod/ElectSyn.mod

@@ -54,14 +54,14 @@ NEURON {
     RANGE g, i
     RANGE weight
 
-    RANGE vgap     : Using a RANGE variable as opposed to POINTER for parallel mode
+    RANGE vpeer     : Using a RANGE variable as opposed to POINTER for parallel mode
 
 
 }
 
 PARAMETER {
     v (millivolt)
-    vgap (millivolt)
+    vpeer (millivolt)
     g = 0.000049999999999999996 (microsiemens)
     weight = 1
 
@@ -73,5 +73,5 @@ ASSIGNED {
 }
 
 BREAKPOINT {
-    i = weight * g * (v - vgap)
+    i = weight * g * (v - vpeer)
 } 

doc/source/reference.rst

@@ -112,11 +112,13 @@ The value consists in turn of a dictionary with the parameters of the population
 * **cellType** - Arbitrary cell type attribute/tag assigned to all cells in this population; can be used as condition to apply specific cell properties. 
 	e.g. 'Pyr' (for pyramidal neurons) or 'FS' (for fast-spiking interneurons)
 
-* **numCells** or **density** - The total number of cells in this population or the density in neurons/mm3 (one or the other is required). 
+* **numCells**, **density** or **gridSpacing** - The total number of cells in this population, the density in neurons/mm3, or the fixed grid spacing (only one of the three is required). 
 	The volume occupied by each population can be customized (see ``xRange``, ``yRange`` and ``zRange``); otherwise the full network volume will be used (defined in ``netParams``: ``sizeX``, ``sizeY``, ``sizeZ``).
 
 	``density`` can be expressed as a function of normalized location (``xnorm``, ``ynorm`` or ``znorm``), by providing a string with the variable and any common Python mathematical operators/functions. e.g. ``'1e5 * exp(-ynorm/2)'``.
 
+	``gridSpacing`` is the spacing between cells (in um). The total number of cells will be determined based on spacing and ``sizeX``, ``sizeY``, ``sizeZ``. e.g. ``10``.
+
 * **cellModel** - Arbitrary cell model attribute/tag assigned to all cells in this population; can be used as condition to apply specific cell properties. 
 	e.g. 'HH' (standard Hodkgin-Huxley type cell model) or 'Izhi2007' (Izhikevich 2007 point neuron model). 
 
@@ -1026,7 +1028,7 @@ Population class methods
 * **pop.createCellsList()**
 
 
-Population class methods 
+Cell class methods 
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
 * **cell.create()**

examples/HHTut/HHTut_export.py

@@ -3,7 +3,7 @@
 
 np = HHTut.netParams
 print("********************\n*\n*  Note: setting noise to 1, since noise can only be 0 or 1 in NeuroML export currently!\n*\n********************")
-np.popParams['background']['noise'] = 1
+np.stimSourceParams['bkg']['noise'] = 1
 
 sim.createExportNeuroML2(netParams = np, 
                        simConfig = HHTut.simConfig,

examples/HybridTut/HybridTut_export.py

@@ -3,7 +3,7 @@
 
 np = HybridTut.netParams
 print("********************\n*\n*  Note: setting noise to 1, since noise can only be 0 or 1 in NeuroML export currently!\n*\n********************")
-np.popParams['background']['noise'] = 1
+np.stimSourceParams['bkg']['noise'] = 1
 
 sim.createExportNeuroML2(netParams = np, 
                        simConfig = HybridTut.simConfig,

examples/M1/M1_export.py

@@ -3,8 +3,8 @@
 
 np = M1.netParams
 print("********************\n*\n*  Note: setting noise to 1, since noise can only be 0 or 1 in NeuroML export currently!\n*\n********************")
-np.popParams['background_E']['noise'] = 1
-np.popParams['background_I']['noise'] = 1
+np.stimSourceParams['background_E']['noise'] = 1
+np.stimSourceParams['background_I']['noise'] = 1
 
 sim.createExportNeuroML2(netParams = np, 
                        simConfig = M1.simConfig,

examples/batchCell/README.md

@@ -0,0 +1,57 @@
+# Running batch simulations using NetPyNE
+
+Includes two toy examples using the 6-compartment M1 Corticospinal cell.
+1) f-I curve as a function of the amplitude of an IClamp and dendritic Na conductance.
+2) EPSP amplitude as a function of the weight of aNetStim and the rise time of an NMDA synapse.
+
+The code runs the required batch simulations in parallel using MPI and plots the corresponding figures.
+
+## Requirements
+- NEURON with MPI support
+- NetPyNE 
+
+## Quick steps to run batch sims and plot results 
+
+1) Compile mod files: 
+	cd mod
+	mkmod
+
+2) Run batch simulations: 
+	./runbatch [num_cores]
+
+3) Plot results:
+	ipython -i analysis.py
+
+
+## Code structure:
+
+* netParams.py: Defines the network model. Includes "fixed" parameter values of cells, synapses, connections, stimulation, etc. Changes to this file should only be made for relatively stable improvements of the model. Parameter values that will be varied systematically to explore or tune the model should be included here by referencing the appropiate variable in the simulation configuration (cfg) module. Only a single netParams file is required for each batch of simulations.
+
+* cfg.py: Simulation configuration module. Includes parameter values for each simulation run such as duration, dt, recording parameters etc. Also includes the model parameters that are being varied to explore or tune the network. When running a batch, NetPyNE will create one cfg file for each parameter configuration.
+
+* init.py: Sequence of commands to run a single simulation. Can be executed via 'ipython init.py'. When running a batch, NetPyNE will call init.py multiple times, pass a different cfg file for each of the parameter configurations explored. 
+
+* batch.py: Defines the parameters and parameter values explored in the batch, as well as the run configuration (e.g. using MPI in a multicore machine or PBS Torque for HPCs). Includes examples of how to create 2 different batch simulations: 
+	1) batchNa(): explore the effect of changing the amplitude of an IClamp and the dendritic Na conductance.
+	2) batchNMDA(): explore the effect of changing the weight of NetStim and the rise time of an NMDA synapse.
+
+To run, add a call to either of these functions in __main__(), or import the batch module and run interactively. 
+
+To add new batch explorations, follow the format of the above example, making sure that the parameters being explored are included in cfg.py and referenced appropiately in netParams.py
+
+* runbatch: Shell script to run the batch simulation using mpi. Alternatively, can run "manually" using: 
+mpiexec -np [num_cores] nrniv -python -mpi batch.py 
+
+* analysis.py: Functions to read and plot figures from the batch simulation results. Can call readPlotNa() or readPlotNMDA() from __main__() or import analysis and call interactively.  
+
+* utils.py: Support functions to read output data from batch simulations, and others. Eventually, some of these functions will be moved to NetPyNE. 
+
+* /cells: Folder containing cell model. In this case only includes SPI6.py. NetPyNE will import the cell specifications from this file. 
+
+* /data: Folder to store output of simulations.
+
+* /mod: Folder containing mod files (need to clean up to keep just mod files required for this model)
+
+
+For further information please contact: [email protected] 
+

examples/batchCell/analysis.py

@@ -0,0 +1,102 @@
+"""
+analysis.py 
+
+Functions to read and plot figures from the batch simulation results. 
+Can call readPlotNa() or readPlotNMDA() from __main__() or import analysis and call interactively.  
+
+Contributors: [email protected]
+"""
+
+import utils
+import json
+import matplotlib.pyplot as plt
+
+
+def plotfINa(dataFolder, batchLabel, params, data):
+    utils.setPlotFormat(numColors = 8)
+
+    Pvals = params[0]['values']
+    Ivals = params[1]['values']
+    Pvalsdic = {val: i for i,val in enumerate(Pvals)}
+    Ivalsdic = {val: i for i,val in enumerate(Ivals)}
+
+    rates = [[0 for x in range(len(Pvals))] for y in range(len(Ivals))] 
+    for key, d in data.iteritems():
+        rate = len(d['simData']['spkt'])
+        Pindex = Pvalsdic[d['paramValues'][0]]
+        Iindex = Ivalsdic[d['paramValues'][1]]
+        rates[Iindex][Pindex] = rate
+        print d['paramValues']
+        print rate
+
+    filename = '%s/%s/%s_fIcurve.json' % (dataFolder, batchLabel, batchLabel)
+    with open(filename, 'w') as fileObj:
+        json.dump(rates, fileObj)
+
+    plt.figure()
+
+    handles = plt.plot(rates, marker='o', markersize=10)
+    plt.xlabel('Somatic current injection (nA)')
+    plt.xticks(range(len(Ivals))[::2], Ivals[::2])
+    plt.ylabel('Frequency (Hz)')
+    plt.legend(handles, params[0]['values'], title = 'dend Na', loc=2)
+    plt.savefig('%s/%s/%s_fIcurve.png' % (dataFolder, batchLabel, batchLabel))
+    plt.show()
+
+
+def plotNMDA(dataFolder, batchLabel, params, data, somaLabel='soma', stimRange=[5000,10000]):
+    utils.setPlotFormat(numColors = 8)
+
+    Pvals = params[0]['values']
+    Wvals = params[1]['values']
+    Pvalsdic = {val: i for i,val in enumerate(Pvals)}
+    Wvalsdic = {val: i for i,val in enumerate(Wvals)}
+
+    epsps = [[0 for x in range(len(Pvals))] for y in range(len(Wvals))] 
+    for key, d in data.iteritems():
+        cellLabel = d['simData']['V_soma'].keys()[0]
+        vsoma = d['simData']['V_'+somaLabel][cellLabel]
+        epsp = max(vsoma[stimRange[0]:stimRange[1]]) - vsoma[stimRange[0]-1]
+
+        Pindex = Pvalsdic[d['paramValues'][0]]
+        Windex = Wvalsdic[d['paramValues'][1]]
+        epsps[Windex][Pindex] = epsp
+        print d['paramValues']
+        print epsp
+
+    filename = '%s/%s/%s_epsp.json' % (dataFolder, batchLabel, batchLabel)
+    with open(filename, 'w') as fileObj:
+        json.dump(epsps, fileObj)
+
+    plt.figure()
+
+    handles = plt.plot(epsps, marker='o', markersize=10)
+    plt.xlabel('Weight (of NetStim connection)')
+    plt.ylabel('Somatic EPSP amplitude (mV) in response to 1 NetStim spike')
+    plt.xticks(range(len(Wvals))[::2], Wvals[::2])
+    plt.legend(handles, params[0]['values'], title = 'NMDA tau1 (ms)', loc=2)
+    plt.savefig('%s/%s/%s_epsp.png' % (dataFolder, batchLabel, batchLabel))
+    plt.show()
+
+
+def readPlotNa():
+    dataFolder = 'data/'
+    batchLabel = 'batchNa'
+
+    params, data = utils.readBatchData(dataFolder, batchLabel, loadAll=0, saveAll=1, vars=None, maxCombs=None) 
+    plotfINa(dataFolder, batchLabel, params, data)
+
+
+def readPlotNMDA():
+    dataFolder = 'data/'
+    batchLabel = 'batchNMDA'
+
+    params, data = utils.readBatchData(dataFolder, batchLabel, loadAll=0, saveAll=1, vars=None, maxCombs=None) 
+    plotNMDA(dataFolder, batchLabel, params, data)
+
+
+# Main code
+if __name__ == '__main__':
+    readPlotNa()
+    # readPlotNMDA()
+

examples/batchCell/batch.py

@@ -0,0 +1,50 @@
+"""
+batch.py 
+
+Batch simulation for M1 model using NetPyNE
+
+Contributors: [email protected]
+"""
+
+from netpyne.batch import Batch
+import numpy as np
+
+
+def createBatch(params):
+	# Create Batch object
+	b = Batch()
+	for k,v in params.iteritems():
+		b.params.append({'label': k, 'values': v})
+	return b
+
+
+def runBatch(b, label):
+	b.batchLabel = label
+	b.saveFolder = 'data/'+b.batchLabel
+	b.method = 'grid'
+	b.runCfg = {'type': 'mpi', 
+				'script': 'init.py', 
+				'skip': True}
+
+	b.run() # run batch
+
+
+def batchNa():
+	b = createBatch({'dendNa': [0.025, 0.03, 0.035, 0.4],  
+					('IClamp1', 'amp'): list(np.arange(-2.0, 8.0, 0.5)/10.0)})
+	runBatch(b, 'batchNa')
+
+
+def batchNMDA():
+	b = createBatch({'tau1NMDA': [10, 15, 20, 25],  
+	  				('NetStim1', 'weight'): list(np.arange(1.0, 10.0, 1.0)/1e4)})
+	runBatch(b, 'batchNMDA')
+
+
+# Main code
+if __name__ == '__main__':
+	batchNa() 
+	# batchNMDA()
+
+
+

examples/batchCell/cells/SPI6.py

@@ -0,0 +1 @@
+/usr/site/nrniv/local/python/SPI6.py