maikia / neuroneap Goto Github PK

nosetests test_cell
gives assertion failure:
FAIL: test_cell.test_current_balance_synapse_at_section_end
assert (np.abs(total_current(coord, I))<1e-6).all()

this was temporally patched, but proper correction should be done

in full morphological models the current balance is not conserved. This is due to the difference in the way area is calculated in NEURON and NeuronEAP:

1. in neuronEAP diameter (diam) is fixed for whole section, whereas in neuron it can change between segemetns - this is a bug and should be changed using the diam3d function

2. taking the corrected diam and segment length still does not give the same value as the "area" function of NEURON. I don't know where the difference stems from, but from this explanation I understand that the area is interpolated from the 3d specification of the neuron (see below, [1,2])

[1] https://www.neuron.yale.edu/phpBB/viewtopic.php?f=13&t=2332&p=9207&hilit=area+calculation#p9207

Surface area calculation depends on how section geometry was specified. For sections specified with the stylized method (L & diam) all segments are treated as cylinders. For sections specified with the pt3d method (i.e. a series of detailed x,y,z,diam values, read about Geometry, pt3dadd, and related topics in the Programmer's Reference), the section is treated as a sequence of frusta (truncated cones) whose circular ends are at the xyz coordinates and have the associated diameter value.

The way the geometry works in NEURON is explained in the documentation:

[2] https://neuron.yale.edu/neuron/static/docs/help/neuron/neuron/geometry.html#Geometry