Closed anilbey closed 2 months ago
I believe that for this change to take effect in SynapseFactory
, we need to change the following starting in synapse_factory.py:L117:
@@ -114,10 +114,10 @@ class SynapseFactory:
section: NeuronSection = cell.get_psection(section_id=isec).hsection
# old circuits don't have it, it needs to be computed via synlocation_to_segx
- if ("afferent_section_pos" in syn_description and
- not np.isnan(syn_description["afferent_section_pos"])):
+ if (SynapseProperty.AFFERENT_SECTION_POS in syn_description and
+ not np.isnan(syn_description[SynapseProperty.AFFERENT_SECTION_POS])):
# position is pre computed in SONATA
- location = syn_description["afferent_section_pos"]
+ location = syn_description[SynapseProperty.AFFERENT_SECTION_POS]
else:
ipt = syn_description[SynapseProperty.POST_SEGMENT_ID]
syn_offset = syn_description[SynapseProperty.POST_SEGMENT_OFFSET]
Yes you're right. That must have gone unnoticed. A patch will follow.
No worries, that happens. Thanks! 🤘
It's addressed here: https://github.com/BlueBrain/BlueCelluLab/pull/169
Spotted by @joni-herttuainen. When afferent_section_pos is not in SONATA, it is computed in BCL and this can cause small precision differences. This change avoids recomputing it when it is already available in SONATA.