A single CSV — layers_banc_626.csv — recording the forward / backward
graph-traversal layer count for every BANC v626 neuron. Two columns:
Root ID, FWD layer, BWD layer
The "FWD layer" is the smallest hop count from any sensory neuron;
"BWD layer" is the smallest hop count to any effector neuron. Both
derived using the same graph-traversal model as Schlegel et al., 2024
(see ../schlegel_2024_fafb_neuron_ranks/).
This was the original cross-check for the early version of the influence
metric and appears as the layer_mean join column in
R/startup/banc-distances.R. For v888 we use query_influence() on
the live graph instead, which collapses the same information into the
adjusted-influence formulation (Methods §"Influence", Eq. 9). The v626
layers are kept for back-compatibility with the preprint-era panels
(legacy validation tail of R/figures/panels_influence_validation.R).
External — produced by a collaborator-supplied graph-traversal run on BANC v626. Not regenerated by anything in this repo.
R/figures/panels_influence_validation.R only (and only when the
legacy comparison branch is enabled).