//genesis //SynParams.g str AMPAname = "AMPA" float EkAMPA = 0.0 float AMPAtau1 = 1.1e-3 float AMPAtau2 = 5.75e-3 float AMPAgmax = 0.342e-9 //593e-12 I changed this to make the NMDA/AMPA ratio more like 2.75/1 which is what Ding 2008 finds int corticalstriatal synapses //more like 2/1 for thalamus so should be 0.47e-9 for thalamo-striatal syanpnse if NMDA is 0.94e-9 Rebekah Evans 6/25/10 str GABAname = "GABA" float GABAtau1 = 0.25e-3 // From Galarreta and Hestrin 1997 float GABAtau2 = 3.75e-3 //(used in Wolfs model) float EkGABA = -0.060 float GABAgmax = 750e-12 //Modified Koos 2004 (Wolf uses 435e-12) int GABA2Spine = 0 // = 0, No GABA; // 1, add GABA to spine head // 2, add GABA to spine neck int addCa2Spine = 1 // 0, no ca channels in spine, //1, yes ca channels in spine (non-synaptic) int NMDABufferMode = 0 // 1, connect both NMDA and AMPA calcium to NMDA_buffer // 0, connect only NMDA currents to NMDA_buffer float useAMPANMDAGHKchannels= 0 // we do not use GHK project for NMDA/AMPA setclock 0 5e-6 //was 5e-6 // Simulation time step (Second) setclock 1 2e-5 // time step for ascii output //setclock 1 1e-4 // time step for graphic output // parameters for NMDA subunits // cortex str subunit = "Cortex" float EkNMDA = 0 float Kmg = 3.57 float NMDAtau2 = (112.5e-3)/2 //ctx avg for .25 NR2B and .75 NR2A. (300e-3)/2 (NR2B) (50e-3)/2 (NR2A) 75+37.5 = 112.5 float NMDAgmax = 0.94e-9 //NR2A and B from (Moyner et al., 1994 figure 7) int ghk_yesno=0 str NMDAname = {subunit} //for saving info on distal or proximal dendrites or massed and spaced. formula typeof dend, # of spines.