Recorded conductance exceeding gbar value

NMODL and the Channel Builder.
Post Reply
slerts

Recorded conductance exceeding gbar value

Post by slerts »

I am playing with a potassium channel model I found on ModelDB (Gurkiewicz & Korngreen, 2007) with the end goal of building a calcium channel from scratch. Right now I'm trying to understand how to properly implement markov models by playing with the values of this model. I have coded a voltage step experiment (-80mv to +60mV in 20mV steps) and have 3 recording vectors:
reci.record(&soma.ik(0.5))
recg.record(&soma.g_k_3st(0.5))
rect.record(&t)

Where k_3st is the mechanism name that has been inserted into the soma. The model kinetics are as follows where I'm providing values for gbar as well as all the a and z variables in the hoc file.

Code: Select all

STATE { c2 c1 o }

BREAKPOINT {
      SOLVE states METHOD sparse
      g = gbar*o
      ik = (1e-4)*g*(v - ek)
}

INITIAL { SOLVE states STEADYSTATE sparse}

KINETIC states {
        rates(v)
        ~c1 <-> o (k12,k21)
      ~c2 <-> c1 (k23,k32)
        CONSERVE c2+c1+o=1
}

PROCEDURE rates(v(millivolt)) {

      k12 = a12*exp(z12*v)
      k21 = a21*exp(-z21*v)
      k23 = a23*exp(z23*v)
      k32 = a32*exp(-z32*v)
}
I've noticed that at the highest voltage step of +60mV the max conductance value that occurs in the recording is 19.9258 however I've set the value for gbar in the hoc file as 19.8. I don't understand how this can occur when g = gbar*o and the max value of o is 1.
ted
Site Admin
Posts: 6289
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Recorded conductance exceeding gbar value

Post by ted »

You're concerned about < 1% error in a numerical solution generated by a method that has first order accuracy? Try making dt smaller, or change to adaptive integration, and see what happens.
slerts

Re: Recorded conductance exceeding gbar value

Post by slerts »

Thanks Ted! Not so much concerned about such a small error for this as I was trying to figure out what the cause of this was to prevent it from occurring in future models. I'll have to go back and reread the chapter in the book on the integration methods.
ted
Site Admin
Posts: 6289
Joined: Wed May 18, 2005 4:50 pm
Location: Yale University School of Medicine
Contact:

Re: Recorded conductance exceeding gbar value

Post by ted »

Error is inherent in solutions generated by numerical integration. To reduce the error with fixed dt methods use a smaller time step, or change to an adaptive integration method and use a tight error tolerance.
Post Reply