### Ion channel bugs in Santhakumar/Morgan dentate gyrus model

Posted:

**Tue Nov 05, 2013 7:45 pm**Hello, we have come across a few bugs in the ion channels of the Santhakumar/Morgan model on on ModelDB (http://senselab.med.yale.edu/modeldb/Sh ... del=124513). We are addressing these in a future model version. As the cell models used in this model were fit to experimental data and their overall behavior known and found acceptable to the original modelers, we stand by the results achieved from these network models. However, we recommend that the ion channel bugs not be propagated further across future network, cell, or ion channel models. Anyone who uses cells or ion channels from this version of the model should be aware of and correct for the typos. This will require the cell electrophysiology to be refit (possible using the Multiple Run Fitter) to achieve the desired cell behavior.

File: http://senselab.med.yale.edu/modeldb/Sh ... us\nca.mod

Effect: ion channel current is smaller than it should be

Details: In computing the "d" parameter describing channel inactivation kinetics, the equation for alpha is "0.00016/exp(-v/48.4)", but the equation on Page 6 of Aradi and Holmes (1999) (upon which this channel was based) reads that alpha_d is "0.00016*exp(-v/48.4)", so the two terms should be multiplied, not divided. When corrected, the inactivation kinetics are slower and the current is therefore greater.

Correction:

File: http://senselab.med.yale.edu/modeldb/Sh ... s\CaBK.mod

Effect: The Potassium channel depends only on the initial, not the current, Calcium concentration, so the current will be much lower than it should be if an action potential or other significant depolarization occurs

Details: This bug was discovered by Phillip Hendrickson. The potassium channel is supposed to be dependent on the current Calcium concentration, but it actually has no knowledge of the current Calcium concentration, only the initial Calcium concentration at the start of the simulation.

Correction: In the DERIVATIVE state definition, prior to calling the rate function, add the line "cai= ncai + lcai + tcai" so that the DERIVATIVE state block looks like:

**Issue 1: N-type Calcium channel equation is wrong**File: http://senselab.med.yale.edu/modeldb/Sh ... us\nca.mod

Effect: ion channel current is smaller than it should be

Details: In computing the "d" parameter describing channel inactivation kinetics, the equation for alpha is "0.00016/exp(-v/48.4)", but the equation on Page 6 of Aradi and Holmes (1999) (upon which this channel was based) reads that alpha_d is "0.00016*exp(-v/48.4)", so the two terms should be multiplied, not divided. When corrected, the inactivation kinetics are slower and the current is therefore greater.

Correction:

Code: Select all

```
PROCEDURE rates(v) {
... : other code lines ahead of this section
:"d" NCa inactivation system
alpha = 0.00016*exp(-v/48.4) : the two terms should multiplied, not divided (see Aradi & Holmes, 1999, page 6)
beta = 1/(exp((-v+39)/10)+1)
sum = alpha+beta
dtau = 1/sum dinf = alpha/sum
}
```

**Issue 2: Calcium dependent, voltage dependent Potassium channel does not update Calcium concentration**File: http://senselab.med.yale.edu/modeldb/Sh ... s\CaBK.mod

Effect: The Potassium channel depends only on the initial, not the current, Calcium concentration, so the current will be much lower than it should be if an action potential or other significant depolarization occurs

Details: This bug was discovered by Phillip Hendrickson. The potassium channel is supposed to be dependent on the current Calcium concentration, but it actually has no knowledge of the current Calcium concentration, only the initial Calcium concentration at the start of the simulation.

Correction: In the DERIVATIVE state definition, prior to calling the rate function, add the line "cai= ncai + lcai + tcai" so that the DERIVATIVE state block looks like:

Code: Select all

```
DERIVATIVE state { : exact when v held constant; integrates over dt step
cai= ncai + lcai + tcai
rate(v, cai)
o' = (oinf - o)/otau
}
```