Included the 240710_C0 model, as an alternative to 210710_C1.
Imported the model and modified the relevant simulation/analysis scripts. Note that setting sodium and potassium reversal potential over the whole cell crashes neuroConstruct for some reason.
Revert "Begun importing other version of detailed morphology GoC."
This reverts commit be055a8cbbf8955c1fc67608f7be31bddc890804.
Begun importing other version of detailed morphology GoC.
Varying the number of stimulation points on the detailed dendrites.
Still fairly hard-coded, but at least I now have more than 3 points for the exponential fit on the detailed morphology.
Removed recording of superfluous data during simulations.
Scripts seem to be working fine.
Attenuation seems to match between the two models (at least with the 4-5-6 dendrite on the reduced morphology and a specific choice of compartments on the detailed one), but they both seem to give different results than those in the paper.
Tidied up the code.
Working on the simulation/analysis scripts.
Added plotting of dependance of maximum somatic response amplitude on distance of stimulation along the dendritic tree.
Leak conductance on both models has been set to 6.67e-10mSum-2, as per supplementary information (page 3) in Vervaeke2012, where 15kOhmcm2 was reported from unpublished observations.
Also, fixed the (wrong) number of internal divisions of a segment.
Still playing around with reduced morphology. Added an axon.
Playing around with morphology to see effects on FI and attenuation.
Adjusted spike detection threshold to avoid picking up doublets in the Vervaeke cell when calculating the F-I curve.
Added scripts for simulating and plotting an F-I curve comparison.
Integrating fast segment-to-soma distance calculation in nC allows to meaningfully compare dendritic attenuation effects on different (and complex) morphologies.
Fixed a remaining issue with the detailed morphology cell.
Note that I turned off again all Na channels.
Now pure NeuroML morphology read/write seems to be working fine, except that it removes property elements tagged 'comment'. This will need to be fixed in neuroConstruct, but doesn't have...
Reverted morphologically-detailed cell model back to its original form to fix some issues (namely, loss of 'number of internal divisions' properties) generated when doing NeuroML write/read with older versions of neuroConstruct.
Modified leak conductance to match what Koen did.
Comparing attenuation in the original (Koen's) model and in my reduced one.
Important: to get to a meaningful comparison, I'm tweaking the spatial discretisation of the original model.
Simulating attenuation for original Golgi model (Na channels switched off as per TTX)
Added a new simulation configuration where GJGolgi_Reduced is simply left alone.
Actual reference morphologies.
Committing reference morphologies, now that neuroConstruct has been fixed to not shuffle channels and synaptic types in NeuroML serialisations and to keep track of the number of internal divisions in each segment. See r1686 and r1689 in nC's svn repo.
Now the location of the synaptic input can be changed, and can be iterated over. Also added a simple plotting script.
Trying to script an attenuation-length analysis that could be compared to Koen's 2012 paper.
The script itself is not yet usable. I have imported a cell from Vervaeke2012 (mainly for the synaptic model, but also to keep it there as a comparison) and a spike relay that I need to simulate a single...
Initial commit: nC project.
Initial commit: ignore file.