// systematic_explorations.hoc
// Explore what ranges of parameters leads to which changes in inhibition.
// How robust is the model to perturbations in parameters.
/*
Parameters to check:
membrane capacitance, cm
axial resistance, Ra
resting leak conductance, gbar_pas
Where "parameter" appears below it refers to one at a time of "cm",
"Ra", gbar_pas", etc.
What ranges of the parameters gives a 10, 15, 20% change in
dCa_inhib/dCa_ctrl?
Output Folder results:
systematic_explorations/cm_statistics.txt
systematic_explorations/Ra_statistics.txt
systematic_explorations/gbar_pas_statistics.txt
where ctrl_or_inhib is bAP or bAPspineinhib
See where the 10, 15, 20% levels fall for the selected parameters by
using the multiple run fitter and praxis.
Just calculate the dCa_inhib/dCa_ctrl and record that in a results_vec
that is paired with a param_vec.
*/
/* Initially set up manually for cm where the error function has ratio
involved where ratio is defined:
ratio = dCa_inhib/dCa_ctrl
And ratio_0 is defined as ratio for the default model parameters. The
error function as a function of a fixed percent and the model
parameter to be optimized over is defined to be
error(percent, parameter) = abs( ratio - (1+percent/100) * ratio_0) *
abs( ratio - (1 - percent/100) * ratio_0)
When the current model ratio is a percent above the original ratio or
below the original the error drops to zero. Check latter that the
ratio is inbetween the original ratio and the boundary ratio for
values of the parameter between the default and the optimized value.
create a function ratio_calc(cai_inhib_vec, cai_ctrl_vec, t_vec,
t_baseline, t_window) where the inside calcium concentration recorded
vectors in inhibited and controlled cases are passed along with the
time points they are recorded at and the baseline time t (which
presumably occurs right before stimulation of the model). t_windows
refers to the length of time after the baseline that the max Ca2+
value is expected to occur. This may eliminate the comparison of
subsequent APs Ca2+ values.
Sample values:
t_baseline = 99.9
t_window = 45 // (approximately 145-99.9)
*/
/* every time a parameter ratio is calculated graph the (parameter,
ratio) pair. This may help in gaining insight into the behaviour of
the model with respect to the parameter.
Start with some recent code that calculates the ratio. The last code
that I made that does this was:
dend_shaft_bAP_functions.hoc
Base the fitness function, fitness_ratio() on the negative of the above error function which is calculated by:
Run the model twice, once with bAP, and secondly bAP plus inhib. For
the compartment receiving inhibition calculate the ratio by passing
the recorded vec values and parameters to ratio_calc() (see above).
*/
// code borrowed from dend_shaft_bAP_functions.hoc:
strdef proto_dir
strdef filename
objref f // temporary file object
// f = new File()
objref dend_v_vec, spine0_v_vec, spine1_v_vec, spine2_v_vec
objref dend_cai_vec, spine0_cai_vec, spine1_cai_vec, spine2_cai_vec
// tstop= 1 // test condition
// below are graphs (the 5 conditions turned into 6 with the addition
// of protocol bAP + 1x inhibition in dendrite
// num_of_conditions = 10 // ctrl and inhib for 1,1/2,1/4,1/8,1/16 // 18
// objref graphs_v[num_of_conditions], graphs_ca[num_of_conditions], graphs_gates[num_of_conditions]
// objref graph_bAP_envelopes
// add the activation and inactivation curves for the ca chan in spines and dend
objref m_ca_dend_vec, h_ca_dend_vec, m_ca_spine1_vec, h_ca_spine1_vec
objref m_ca_spine0_vec, h_ca_spine0_vec
objref bAP_peak_v_vec, bAP_peak_v_x_vec, bAP_profile_vec
objref t_vec
/*
for i=0, num_of_conditions-1 {
graphs_v[i] = new Graph(0)
graphs_ca[i] = new Graph(0)
graphs_gates[i] = new Graph(0)
}
graph_bAP_envelopes = new Graph()
proto_num=0 // index into above graphs of protocol results
*/
proc setup_vectors_to_record() {
// record voltages in the dendrite and the Spine[X].head X=0,1,2 (corresponds to paper spines 1 2 3)
dend_v_vec = new Vector()
dend_cai_vec= new Vector()
spine0_v_vec= new Vector()
spine1_v_vec= new Vector()
spine2_v_vec= new Vector()
spine0_cai_vec = new Vector()
spine1_cai_vec = new Vector()
spine2_cai_vec = new Vector()
m_ca_dend_vec = new Vector()
h_ca_dend_vec = new Vector()
m_ca_spine1_vec = new Vector()
h_ca_spine1_vec = new Vector()
m_ca_spine0_vec = new Vector()
h_ca_spine0_vec = new Vector()
bAP_peak_v_vec = new Vector()
bAP_peak_v_x_vec = new Vector()
t_vec = new Vector()
t_vec.indgen(0, tstop, dt)
spinelocation=adjacent_shaft_x_loc // record the dendrite where Spine[1] is attached
sprint(tmpstr, "dendrite.v(%f)", adjacent_shaft_x_loc) // used to be 180/dendrite.L
dend_v_vec.label(tmpstr)
spine0_v_vec.label("Spine[0].head.v(0.5)")
spine1_v_vec.label("Spine[1].head.v(0.5)")
spine2_v_vec.label("Spine[2].head.v(0.5)")
sprint(tmpstr,"dendrite.cai(%g)", spinelocation) // dend. comp. adj. to spine
dend_cai_vec.label(tmpstr)
spine0_cai_vec.label("Spine[0].head.cai(0.5)")
spine1_cai_vec.label("Spine[1].head.cai(0.5)")
spine2_cai_vec.label("Spine[2].head.cai(0.5)")
sprint(tmpstr,"dendrite.m_ca(%g)", spinelocation) // dend. comp. adj. to spine
m_ca_dend_vec.label(tmpstr)
sprint(tmpstr,"dendrite.h_ca(%g)", spinelocation) // dend. comp. adj. to spine
h_ca_dend_vec.label(tmpstr)
m_ca_spine1_vec.label("Spine[1].head.m_ca(0.5)")
h_ca_spine1_vec.label("Spine[1].head.h_ca(0.5)")
m_ca_spine0_vec.label("Spine[0].head.m_ca(0.5)")
h_ca_spine0_vec.label("Spine[0].head.h_ca(0.5)")
dend_v_vec.record(&dendrite.v(spinelocation), t_vec)
dend_cai_vec.record(&dendrite.cai(spinelocation), t_vec)
spine0_v_vec.record(&Spine[0].head.v(0.5), t_vec)
spine1_v_vec.record(&Spine[1].head.v(0.5), t_vec)
spine2_v_vec.record(&Spine[2].head.v(0.5), t_vec)
spine0_cai_vec.record(&Spine[0].head.cai(0.5), t_vec)
spine1_cai_vec.record(&Spine[1].head.cai(0.5), t_vec)
spine2_cai_vec.record(&Spine[2].head.cai(0.5), t_vec)
m_ca_dend_vec.record(&dendrite.m_ca(spinelocation), t_vec)
h_ca_dend_vec.record(&dendrite.h_ca(spinelocation), t_vec)
m_ca_spine1_vec.record(&Spine[1].head.m_ca(0.5), t_vec)
h_ca_spine1_vec.record(&Spine[1].head.h_ca(0.5), t_vec)
m_ca_spine0_vec.record(&Spine[0].head.m_ca(0.5), t_vec)
h_ca_spine0_vec.record(&Spine[0].head.h_ca(0.5), t_vec)
}
// code not from dend_shaft_inhib_functions.hoc:
objref cai_inhib_vec, cai_ctrl_vec, t_vec
cai_inhib_vec = new Vector()
cai_ctrl_vec = new Vector()
t_vec = new Vector()
ratio_0 = 1 // declare. Get's replaced with function call after below defintion.
percent = 10 // declare. Run over values latter.
// ratio_calc(cai_inhib_vec $o1, cai_ctrl_vec $o2, t_vec $o3, t_baseline $4, t_window $5)
func ratio_calc() { local t_baseline, t_window
cai_inhib_vec = $o1
cai_ctrl_vec = $o2
t_vec = $o3
t_baseline=$4
t_window=$5
prestimulus_time=t_baseline
prestimulus_index=prestimulus_time/dt // 1998, the 1999th element for dt=.05
index_end = (t_baseline +t_window)/dt // for 150/dt might be 3000 (dt=0.05) or 6000 (dt=0.025) to find
// an index after the first AP but hopefully before
// the second if another AP should occur
spine1_B_ctrl=cai_ctrl_vec.x[prestimulus_index]
spine1_B_inhib=cai_inhib_vec.x[prestimulus_index]
spine1_Ca_peak_ctrl = cai_ctrl_vec.c(prestimulus_index,index_end).max()
spine1_Ca_peak_inhib = cai_inhib_vec.c(prestimulus_index,index_end).max()
spine1_dCa_ratio = (spine1_Ca_peak_inhib-spine1_B_inhib)/(spine1_Ca_peak_ctrl-spine1_B_ctrl)
return spine1_dCa_ratio
}
// global_param or parameter to study is set to an initial (default) value here
global_cm = .75
global_Ra = 200
global_g_pas = 2.5e-05
func ratio_fitness() {
// First make sure that the global_param value is assigned to
// its appropriate places
print "ratio_fitness start:"
print "global_cm=", global_cm,", global_Ra=",global_Ra,", global_g_pas=",global_g_pas,", dendrite.diam(.5)=",dendrite.diam(.5),", Spine[1].neck.diam(.5)=",Spine[1].neck.diam(0.5)
forall {cm = global_cm}
forall {Ra = global_Ra}
forall {g_pas = global_g_pas}
// Now run model to calculate and return fitness
// turn on bAP only and have no inhibition
// turn on bAP
MultIClamp[0].number=1 // turn on current injection that creates bAP
for nc_index=0,3 {
NC[nc_index].weight = 0.0 // 0.0004 is 0.4 nS (0 dend and 1,2,3 spine synapses off)
}
// run_model
tstop = 151
setup_vectors_to_record()
init()
run()
// save appropriate ctrl cai vector
cai_ctrl_vec = spine1_cai_vec.c
// leave bAP on and inhibit the spine
NC[2].weight = 0.0004 // 0.0004 (uS) is 0.4 nS (0 dend and 1,2,3 spine synapses off)
// run_model
setup_vectors_to_record()
init()
run()
// save appropriate inhib cai vector
cai_inhib_vec = spine1_cai_vec.c
// calculate ratio
t_baseline = 99.9
t_window = 50 // search for cai maxs between t_baseline to t_baseline + t_window
ratio = ratio_calc(cai_inhib_vec, cai_ctrl_vec, t_vec, t_baseline, t_window)
// calculate error function
error = abs( ratio - (1+percent/100) * ratio_0) * abs( ratio - (1 - percent/100) * ratio_0)
// return fitness value as error - (accidentally called error func a fitness func.)
print "for cm=", global_cm,", the dCa_ratio = ",ratio," and the error function is ", error
return error
}
// the first fitness value is garbage because ratio_0 is not set
// properly until ratio_fitness called and ratio_0 is set properly
// below: calculate the default ratio called ratio_0
dummy_fitness_value = ratio_fitness() // calculated with default parameters
ratio_0 = ratio