Wiki Navigation

translation between NeuroML and FacetsML

The translation of Populations and Projections between those standards should not be a problem, because the way they are declared in FacetsML is inspired from NeuroML. The translations of Cells is much more complex.

Populations

A population declared in FacetsML like that :

<populations>
    <population name="exc_cells" label="Excitatory_Cells">
        <dim size="4000"/>
        <cell_type name="cellA"/>
        <randomInit rng="rng">
            <rand:uniform min="-60." max="-50."/>
        </randomInit>
    </population>
</populations>

will naturally become in NeuroML :

<net:populations>
    <net:population name="exc_cells">
      <net:cell_type>cellA</net:cell_type>
      <net:pop_location>
        <net:random_arrangement>
          <net:population_size>4000</net:population_size>
          <net:rectangular_location>
            <meta:corner x="0" y="0" z="0"/>
            <meta:size depth="4000" height="4000" width="4000"/>
          </net:rectangular_location>
        </net:random_arrangement>
      </net:pop_location>
    </net:population>
</net:populations>

The only thing which could be a problem is the spatial location of the population, and its size. That could be calculated in the XSLT transformation.

Projections

A projection declared in FacetsML like that :

<projections>
    <projection name="e2e"
        presynaptic_population_name="exc_cells"
        postsynaptic_population_name="exc_cells"
        method="fixedProbability"
        weight="0.004"
        target="excitatory"
        rng="rng">
            <methodParameter name="p_connect" value="0.02"/>
    </projection>
</projections>

(the "p_connect" methodParameter could become an attribute of a markup "fixedProbability" if the parameters of the projection methods can be stabilized)

That projection would become in NeuroML :

<net:projections units="Physiological Units">
    <net:projection name="e2e">
      <net:source>exc_cells</net:source>
      <net:target>exc_cells</net:target>
      <net:synapse_props>
        <net:synapse_type>ExcitatorySynapse_exc_cells</net:synapse_type>
        <net:default_values weight="0.004"/>
      </net:synapse_props>
      <net:connectivity_pattern>
        <net:fixed_probability probability="0.02"/>
      </net:connectivity_pattern>
    </net:projection>
</net:projections>

Cells

For now, the only PyNN standardized model which can be translated in NeuroML is the "IF_cond_exp" neuron, defined like that in common.py :

class IF_cond_exp(StandardCellType):

"""Leaky integrate and fire model with fixed threshold and decaying-exponential post-synaptic conductance."""

default_parameters = {

'v_rest' -65.0, Resting membrane potential in mV
'cm' 1.0, Capacity of the membrane in nF
'tau_m' 20.0, Membrane time constant in ms
'tau_refrac' 0.0, Duration of refractory period in ms
'tau_syn_E' 5.0, Decay time of the excitatory synaptic conductance in ms
'tau_syn_I' 5.0, Decay time of the inhibitory synaptic conductance in ms
'e_rev_E' 0.0, Reversal potential for excitatory input in mV
'e_rev_I' -70.0, Reversal potential for inhibitory input in mV
'v_thresh' -50.0, Spike threshold in mV
'v_reset' -65.0, Reset potential after a spike in mV
'i_offset' 0.0, Offset current in nA
'v_init' -65.0, Membrane potential in mV at t = 0

}

Here is its declaration in FacetsML :

<cells>
    <cell name="cellA">
        <class:IF_cond_exp v_rest="-60."
                           cm="1."
                           tau_m="20."
                           tau_refrac="5."
                           tau_syn_E="5."
                           tau_syn_I="10."
                           e_rev_E="0."
                           e_rev_I="-80."
                           v_thresh="-50."
                           v_reset="-60."
                           i_offset="0.1"
                           v_init="-65.0" />
    </cell>
</cells>

which becomes in NeuroML :

<cells>
    <cell name="cellA">
      <meta:notes>Instance of PyNN IF_cond_exp cell type</meta:notes>
      <mml:segments>
        <mml:segment id="0" cable="0" name="Soma">
          <mml:proximal y="0" x="0" z="0" diameter="318.309886184"/>
          <mml:distal y="0" x="0" z="100" diameter="318.309886184"/>
        </mml:segment>
      </mml:segments>
      <mml:cables>
        <mml:cable id="0" name="Soma">
          <meta:group>all</meta:group>
        </mml:cable>
      </mml:cables>
      <biophysics units="Physiological Units">
        <bio:mechanism type="Channel Mechanism" name="IandF_cellA"/>
        <bio:mechanism type="Channel Mechanism" name="pas_cellA">
          <bio:parameter name="gmax" value="5e-05">
            <bio:group>all</bio:group>
          </bio:parameter>
        </bio:mechanism>
        <bio:specificCapacitance>
          <bio:parameter value="1.">
            <bio:group>all</bio:group>
          </bio:parameter>
        </bio:specificCapacitance>
        <bio:specificAxialResistance>
          <bio:parameter value="0.1">
            <bio:group>all</bio:group>
          </bio:parameter>
        </bio:specificAxialResistance>
        <bio:initialMembPotential>
          <bio:parameter value="-65.0">
            <bio:group>all</bio:group>
          </bio:parameter>
        </bio:initialMembPotential>
        <net:potentialSynapticLocation>
          <net:synapse_type>ExcitatorySynapse_cellA</net:synapse_type>
          <net:synapse_direction>preAndOrPost</net:synapse_direction>
          <net:group>all</net:group>
        </net:potentialSynapticLocation>
        <net:potentialSynapticLocation>
          <net:synapse_type>InhibitorySynapse_cellA</net:synapse_type>
          <net:synapse_direction>preAndOrPost</net:synapse_direction>
          <net:group>all</net:group>
        </net:potentialSynapticLocation>
      </biophysics>
    </cell>
</cells>

and its synapses would be described like that :

<channels units="Physiological Units">
    <cml:ion charge="1" name="non_specific" default_erev="-65.0"/>
    <cml:channel_type name="pas_cellA" density="yes">
      <meta:notes>Simple example of a leak/passive conductance</meta:notes>
      <cml:current_voltage_relation>
        <cml:ohmic ion="non_specific">
          <cml:conductance default_gmax="5e-05"/>
        </cml:ohmic>
      </cml:current_voltage_relation>
    </cml:channel_type>
    <cml:channel_type name="IandF_cellA">
      <meta:notes>Spike and reset mechanism</meta:notes>
      <cml:current_voltage_relation>
        <cml:integrate_and_fire threshold="-50." g_refrac="0.1" t_refrac="5." v_reset="-60."/>
      </cml:current_voltage_relation>
    </cml:channel_type>
    <cml:synapse_type name="ExcitatorySynapse_cellA">
      <cml:doub_exp_syn rise_time="5." max_conductance="1.0e-5" reversal_potential="0." decay_time="5.0"/>
    </cml:synapse_type>
    <cml:synapse_type name="InhibitorySynapse_cellA">
      <cml:doub_exp_syn rise_time="10." max_conductance="1.0e-5" reversal_potential="-80." decay_time="5.0"/>
    </cml:synapse_type>
</channels>

So the FacetsML/PyNN parameters are translated that way :

'v_rest' -> default_erev in <channels> <cml:ion default_erev="">
'cm' -> gmax in <cell> <biophysics> <bio:mechanism> <bio:parameter name="gmax" value="">
'tau_m' -> g_refrac in <channels> <cml:channel_type> <cml:current_voltage_relation> <cml:integrate_and_fire g_refrac="">
'tau_refrac' -> t_refrac in <channels> <cml:channel_type> <cml:current_voltage_relation> <cml:integrate_and_fire t_refrac="">
'tau_syn_E' -> decay_time in <channels> <cml:synapse_type name="ExcitatorySynapse_..."> <cml:doub_exp_syn decay_time="">
'tau_syn_I' -> decay_time in <channels> <cml:synapse_type name="InhibitorySynapse_..."> <cml:doub_exp_syn decay_time="">
'e_rev_E' -> reversal_potential in <channels> <cml:synapse_type name="ExcitatorySynapse_..."> <cml:doub_exp_syn reversal_potential="">
'e_rev_I' -> reversal_potential in <channels> <cml:synapse_type name="InhibitorySynapse_..."> <cml:doub_exp_syn reversal_potential="">
'v_thresh' -> threshold in <channels> <cml:channel_type> <cml:current_voltage_relation> <cml:integrate_and_fire threshold="">
'v_reset' -> v_reset in <channels> <cml:channel_type> <cml:current_voltage_relation> <cml:integrate_and_fire v_reset="">
'i_offset' -> something to do with "pas" channel mechanism ?
'v_init' -> value of initialMembPotential in <cell> <biophysics> <bio:initialMembPotential> <bio:parameter value="">

Download in other formats:

Plain Text

'v_rest'	-65.0,	Resting membrane potential in mV
'cm'	1.0,	Capacity of the membrane in nF
'tau_m'	20.0,	Membrane time constant in ms
'tau_refrac'	0.0,	Duration of refractory period in ms
'tau_syn_E'	5.0,	Decay time of the excitatory synaptic conductance in ms
'tau_syn_I'	5.0,	Decay time of the inhibitory synaptic conductance in ms
'e_rev_E'	0.0,	Reversal potential for excitatory input in mV
'e_rev_I'	-70.0,	Reversal potential for inhibitory input in mV
'v_thresh'	-50.0,	Spike threshold in mV
'v_reset'	-65.0,	Reset potential after a spike in mV
'i_offset'	0.0,	Offset current in nA
'v_init'	-65.0,	Membrane potential in mV at t = 0

'v_rest'	-> default_erev in <channels> <cml:ion default_erev="">
'cm'	-> gmax in <cell> <biophysics> <bio:mechanism> <bio:parameter name="gmax" value="">
'tau_m'	-> g_refrac in <channels> <cml:channel_type> <cml:current_voltage_relation> <cml:integrate_and_fire g_refrac="">
'tau_refrac'	-> t_refrac in <channels> <cml:channel_type> <cml:current_voltage_relation> <cml:integrate_and_fire t_refrac="">
'tau_syn_E'	-> decay_time in <channels> <cml:synapse_type name="ExcitatorySynapse_..."> <cml:doub_exp_syn decay_time="">
'tau_syn_I'	-> decay_time in <channels> <cml:synapse_type name="InhibitorySynapse_..."> <cml:doub_exp_syn decay_time="">
'e_rev_E'	-> reversal_potential in <channels> <cml:synapse_type name="ExcitatorySynapse_..."> <cml:doub_exp_syn reversal_potential="">
'e_rev_I'	-> reversal_potential in <channels> <cml:synapse_type name="InhibitorySynapse_..."> <cml:doub_exp_syn reversal_potential="">
'v_thresh'	-> threshold in <channels> <cml:channel_type> <cml:current_voltage_relation> <cml:integrate_and_fire threshold="">
'v_reset'	-> v_reset in <channels> <cml:channel_type> <cml:current_voltage_relation> <cml:integrate_and_fire v_reset="">
'i_offset'	-> something to do with "pas" channel mechanism ?
'v_init'	-> value of initialMembPotential in <cell> <biophysics> <bio:initialMembPotential> <bio:parameter value="">