These are chat archives for openworm/ChannelWorm

7th
Jun 2015
Stephen Larson
@slarson
Jun 07 2015 07:10
@VahidGh i believe #50 is a start on SciUnit Capabilities you are looking for
@VahidGh Nice fits!
Stephen Larson
@slarson
Jun 07 2015 07:16
@VahidGh hmm we need to think more about the role inactivation plays. Shouldnt the presence of inactivation be visible in the shape of the i/v curve? If so, then i'd expect we would be optimizing for it... I may need to outline this argument more clearly
Vahid Ghayoomie
@VahidGh
Jun 07 2015 09:05
@slarson, As for the Capabilities, I commented under #50.
And regarding the inactivation, if I consider inactivation expressions for this plot, we will have no current value greater than zero, so we can not compare these two plots.
This is also true for the implementation codes for Boyle & Cohen. And this is due to the problem that it's not being considered in the I/V formulation by the original paper. So as the cost function is based on comparing the simulated plot with the original one, we have to ignore inactivation parameters.
One solution would be to consider multiple (weighted and normalized) cost functions, e.g. one with this plot, and one with the plots from I- or V- clamp plots.
Vahid Ghayoomie
@VahidGh
Jun 07 2015 11:47
Better saying this is due to the fact that this I/V plot is for the steady state current (see this). So if we include inactivation, at some voltage threshold the inactivation gate will be in a completely closed state (as we are considering the steady state), and we will have no more current (although the activation gates are open).
maybe we can improve this by also including the I/V curve for peak values!