CA1 pyr cell: Inhibitory modulation of spatial selectivity+phase precession (Grienberger et al 2017)

 Download zip file 
Help downloading and running models
Accession:225080
Spatially uniform synaptic inhibition enhances spatial selectivity and temporal coding in CA1 place cells by suppressing broad out-of-field excitation.
Reference:
1 . Grienberger C, Milstein AD, Bittner KC, Romani S, Magee JC (2017) Inhibitory suppression of heterogeneously tuned excitation enhances spatial coding in CA1 place cells. Nat Neurosci [PubMed]
Citations  Citation Browser
Model Information (Click on a link to find other models with that property)
Model Type: Neuron or other electrically excitable cell; Realistic Network;
Brain Region(s)/Organism: Hippocampus;
Cell Type(s): Hippocampus CA1 pyramidal cell;
Channel(s):
Gap Junctions:
Receptor(s): NMDA;
Gene(s):
Transmitter(s):
Simulation Environment: NEURON; Python;
Model Concept(s): Active Dendrites; Detailed Neuronal Models; Place cell/field; Synaptic Integration; Short-term Synaptic Plasticity; Spatial Navigation; Feature selectivity;
Implementer(s): Milstein, Aaron D. [aaronmil at stanford.edu];
Search NeuronDB for information about:  Hippocampus CA1 pyramidal cell; NMDA;
/
GrienbergerEtAl2017
morphologies
readme.txt
ampa_kin.mod *
exp2EPSC.mod
exp2EPSG.mod
exp2EPSG_NMDA.mod
gaba_a_kin.mod *
h.mod
kad.mod *
kap.mod *
kdr.mod *
km2.mod
nas.mod
nax.mod
nmda_kin2.mod
nmda_kin3.mod
nmda_kin5.mod *
pr.mod *
vecevent.mod *
batch_EPSP_attenuation.sh
batch_place_cell_r_inp.sh
batch_place_cell_record_i_syn.sh
batch_place_cell_single_compartment.sh
batch_place_cell_subtr_inh.sh
batch_place_cell_subtr_inh_shifted.sh
batch_place_cell_subtr_inh_vclamp.sh
batch_process_i_syn_files.sh
batch_rinp.sh
batch_spine_attenuation_ratio.sh
build_expected_EPSP_reference.sh
build_expected_EPSP_reference_controller.py
build_expected_EPSP_reference_engine.py
consolidate_i_syn_files.py
consolidate_tracked_spine_data.py
fit_parameter_exponential_distribution.py
function_lib.py
optimize_AMPA_KIN.py
optimize_dendritic_excitability_020416.py
optimize_GABA_A_KIN.py
optimize_NMDA_KIN2.py
parallel_branch_cooperativity.sh
parallel_branch_cooperativity_no_nmda.sh
parallel_clustered_branch_cooperativity_nmda_controller_110315.py
parallel_clustered_branch_cooperativity_nmda_engine_110315.py
parallel_EPSP_attenuation_controller.py
parallel_EPSP_attenuation_engine.py
parallel_EPSP_i_attenuation_controller.py
parallel_EPSP_i_attenuation_engine.py
parallel_expected_EPSP_controller.py
parallel_expected_EPSP_engine.py
parallel_optimize_branch_cooperativity.sh
parallel_optimize_branch_cooperativity_nmda_kin3_controller.py
parallel_optimize_branch_cooperativity_nmda_kin3_engine.py
parallel_optimize_EPSP_amp_controller.py
parallel_optimize_EPSP_amp_engine.py
parallel_optimize_pr.sh
parallel_optimize_pr_controller_020116.py
parallel_optimize_pr_engine_020116.py
parallel_rinp_controller.py
parallel_rinp_engine.py
parallel_spine_attenuation_ratio_controller.py
parallel_spine_attenuation_ratio_engine.py
plot_channel_distributions.py
plot_NMDAR_kinetics.py
plot_results.py
plot_spine_traces.py
plot_synaptic_conductance_facilitation.py
process_i_syn_files.py
record_bAP_attenuation.py
simulate_place_cell_no_precession.py
simulate_place_cell_single_compartment.py
simulate_place_cell_single_compartment_no_nmda.py
simulate_place_cell_subtr_inh.py
simulate_place_cell_subtr_inh_add_noise.py
simulate_place_cell_subtr_inh_add_noise_no_na.py
simulate_place_cell_subtr_inh_no_na.py
simulate_place_cell_subtr_inh_no_nmda_no_na.py
simulate_place_cell_subtr_inh_r_inp.py
simulate_place_cell_subtr_inh_rec_i_syn.py
simulate_place_cell_subtr_inh_shifted.py
simulate_place_cell_subtr_inh_silent.py
simulate_place_cell_subtr_inh_vclamp.py
specify_cells.py
                            
__author__ = 'Aaron D. Milstein'
from specify_cells import *
import random
import os
"""
Builds a cell locally so each engine is ready to receive jobs one at a time, specified by an index corresponding to
which synapse to optimize (coarse sampling of the full set of spines).
"""

morph_filename = 'EB2-late-bifurcation.swc'

mech_filename = '043016 Type A - km2_NMDA_KIN5_Pr'

def epsp_amp_error(x, syn):
    """
    Function called by optimize.minimize. Sets specified synaptic point_process parameters, runs a simulation
    stimulating one spine synapse, and calculates error based on distance from target amplitude of resulting somatic
    EPSP.
    :param x: list of parameters
    :param syn: :class:'Synapse'
    :return: float: error
    """
    for i in range(len(x)):
        setattr(syn.target(syn_type), param_names[i], x[i])
    start_time = time.time()
    sim.run()
    t = np.array(sim.tvec)
    vm = np.array(sim.rec_list[0]['vec'])
    interp_t = np.arange(0, duration, 0.001)
    interp_vm = np.interp(interp_t, t, vm)
    left, right = time2index(interp_t, equilibrate-3.0, equilibrate-1.0)
    baseline = np.average(interp_vm[left:right])
    start, end = time2index(interp_t, equilibrate, duration)
    amp = np.max(interp_vm[start:end]) - baseline
    result = {'EPSP_amp': amp}
    Err = 0.
    for target in result:
        Err += ((target_val[target] - result[target])/target_range[target])**2.
    print 'Process:', os.getpid(), 'Spine:', syn.node.index, 'Node:', syn.node.parent.parent.name, 'Time: %.3f s, x: ' \
                                                            '%.2E, Amp: %.3f, Error: %.2E' % (time.time() - start_time,
                                                            x[0], amp, Err)
    return Err


def optimize_single_synapse(syn_index):
    """
    Called by controller, mapped to each engine. Runs optimization procedure for a single spine, returns the optimized
    parameters, distance of the spine from the soma, and the sec_type of the associated dendritic branch.
    :param syn_index: str
    :return: dict
    """
    start_time = time.time()
    syn = syn_list[syn_index]
    syn.source.play(spike_times)
    #result = optimize.minimize(epsp_amp_error, x0, method='L-BFGS-B', args=(syn,), options={'ftol': 1e-3},
    #                           bounds=xbounds)
    # options={'maxfun': 25}
    result = optimize.minimize(epsp_amp_error, x0, method='Nelder-Mead', args=(syn,), options={'xtol': 1e-7,
                                                                                    'ftol': 1e-3, 'maxiter': 20})
    syn.source.play(h.Vector())  # playing an empty vector turns this synapse off for future runs while keeping the
                                 # VecStim source object in existence so it can be activated again
    print 'Process:', os.getpid(), 'optimized Spine:', syn.node.index, 'on Node:', syn.node.parent.parent.name, ' in ' \
                                                '%.3f s, x: %.2E, after %i iterations with Err: %.2E' % \
                                                (time.time() - start_time, result.x[0], result.nfev, result.fun)
    distance = cell.get_distance_to_node(cell.tree.root, syn.node.parent.parent, syn.loc)
    param_vals = [p for p in result.x]
    return {'distance': distance, 'result': param_vals, 'sec_type': syn.node.parent.parent.type}


equilibrate = 250.  # time to steady-state
duration = 300.
v_init = -67.
syn_type = 'AMPA_KIN'
param_names = ['gmax']
param_ylabels = ['Peak Conductance (uS)']
local_random = random.Random()

syn_list = []
cell = CA1_Pyr(morph_filename, mech_filename, full_spines=True)
cell.zero_na()
local_random.seed(0)

for branch in cell.trunk:  # cell.basal+cell.trunk+cell.apical+cell.tuft:
    if len(branch.spines) > 1:
        if branch.sec.L <= 10.:
            node = branch.spines[local_random.sample(range(0, len(branch.spines)), 1)[0]]
            syn = Synapse(cell, node, [syn_type], stochastic=0)
            syn_list.append(syn)
        else:
            num_syns = min(len(branch.spines), int(branch.sec.L//10.))  # a random synapse every 10 um
            for i in local_random.sample(range(0, len(branch.spines)), num_syns):
                node = branch.spines[i]
                syn = Synapse(cell, node, [syn_type], stochastic=0)
                syn_list.append(syn)
    elif branch.spines:
        node = branch.spines[0]
        syn = Synapse(cell, node, [syn_type], stochastic=0)
        syn_list.append(syn)
cell.init_synaptic_mechanisms()
sim = QuickSim(duration, verbose=0)
sim.append_rec(cell, cell.tree.root, loc=0., description='soma')
spike_times = h.Vector([equilibrate])

#the target values and acceptable ranges
target_val = {'EPSP_amp': 0.2}
target_range = {'EPSP_amp': 0.01}

#the initial guess
# x = [gmax]
x0 = [0.0005]

# the bounds
xmin = [0.000001]
xmax = [1.]

# rewrite the bounds in the way required by L-BFGS-B
xbounds = [(low, high) for low, high in zip(xmin, xmax)]