Transfer impedance gives infinity?

[Sherif]

Hi all,
I have a model whose topology is a ball (soma) and several sticks (dendrites). I use the impedance class in order to calculate the electrotonic distance of each dendritic segment from the soma using the following code:

Code: Select all

   finitialize(-65)                            
   zz = new Impedance()                        
   soma { zz.loc(0.5) }                   
   zz.compute($1, 1)                   
   zn = 0
   soma { zn = zz.input(0.5) }       
   forsec "den"  {                            
      for (x,0) {
         ElectroDist_info(x) = -log(zz.transfer(x)/zn)    
      }
   }

I use this code to iterate over each dendritic section and compute the electrotonic distance. The code works pretty well; however, I found that some dendritic segments have infinity electrotonic distance because the zz.transfer(x) gives infinity. Any idea what would cause that?

[ted]

I use this code to iterate over each dendritic section and compute the electrotonic distance.

For the benefit of readers who may not have thought much about this stuff, it's like injecting a current at the soma, observing the resulting voltage at the soma and at some point x on a dendrite, then computing v_at_dend/v_at_soma, and finally taking the negative of the natural log of that ratio. And that indeed is the electrotonic distance for voltage spreading from the soma to the dendrite.

The code works pretty well; however, I found that some dendritic segments have infinity electrotonic distance because the zz.transfer(x) gives infinity. Any idea what would cause that?

Wow, I've never seen that. Sounds like a bug somewhere. Could you send me the hoc, mod, and ses files needed to reproduce this?

[ted]

Hi, Sherif. Sorry it took so long to get back to you on this.

The problem is that the cell isn't completely connected. The "infinite" results are for the sections called dend120..dend133. The Model View tool (NEURON Main Menu / Tools / Model View) tells me there are two model cells--one with root section soma, the other with root section dend120.

Clicking on
Tools / Distributed Mechanisms / Viewers / Shape Name
brings up a tool that shows a shape plot in its left panel and a list of section names in its right panel. Clicking on dend120 shows that it is the proximal section of a long chain of neurites that stretch out to the right, and that these are sections dend121..dend133. Clicking on the shape plot's menu box and selecting 3D Rotate from its menu allows me to rotate the cell "in 3 dimensional space" by clicking on the shape plot's canvas and dragging. This reveals that dend120..dend133 lie in a different plane than all of the other sections, which wouldn't happen if the proximal end of dend120 were connected to soma.

Connecting dend120 to soma during model setup should eliminate the strange impedance results.