Adult T1 Leg (VFB_00120000) painted-domain `ie.index` missing on 750 of 754 channels

[Robbie1977]

Adult T1 Leg (VFB_00120000) painted-domain ie.index missing on 750 of 754 channels

Summary

VFBquery.get_template_roi_tree and the v2 ROI browser filter painted domains
with WHERE exists(ie.index) on the
(:Template)<-[:depicts]-(:Template)<-[ie:in_register_with]-(:Individual)
relationships. For VFB_00120000 (Adult T1 Leg, Kuan2020) only 4 of 754
in_register_with channels carry the index property, dropping 99.5% of the
painted-domain coverage on this template. The user-visible result is an ROI
tree that shows 4 painted domains (femur, tibia, L+R leg neuropil) where the
volume data on the CDN supports 48 distinct anatomy classes.

Counts (live against pdb, public read-only neo4j/vfb)

status	count
total in_register_with channels on VFB_00120000	754
with ie.index set	4
awaiting index assignment	750
distinct painted anatomy classes (FBbt)	48
distinct painted Individuals	678
distinct WLZ folders	678

What the 4 indexed entries are

ie.index	painted_class	label
3	FBbt_00004667	prothoracic femur
4	FBbt_00004665	prothoracic tibia
5	FBbt_00047140	adult T1 leg neuropil (L)
6	FBbt_00047140	adult T1 leg neuropil (R)

The 750 missing channels cover the sensory neurons / motor neurons / muscle
cells / fasciculating nerves / tendons / bristle scolopidia / chordotonal
organ painted regions on the same template. A representative slice of the
missing FBbt classes:

adult dorsal prothoracic nerve, arculum, fast tibial depressor motor
neuron, femoral chordotonal organ scolopidium, mechanosensory neuron of
femoral chordotonal organ, muscle cell of adult prothoracic leg muscle,
prothoracic coxal bristle sensory neuron, prothoracic long tendon, …

Volume data is on disk

Each painted-region Individual has its own folder at
http://www.virtualflybrain.org/data/VFB/i/<sf>/VFB_00120000/, and each of
those folders carries a volume.wlz mask. Spot checks (3 random unindexed
channels): all returned HTTP 200, sizes ~4–14 KB consistent with per-region
binary masks.

The template-level master WLZ at
https://www.virtualflybrain.org/data/VFB/i/0012/0000/VFB_00120000/volume.wlz
is the leg geometry only, not a labelled multi-region volume — meaning index
integers are free for the curator to assign per region; they don't need to
align with a pre-existing stamp scheme.

(For reference, the master labelled NRRD on the same path —
Leg40-poisson_simple_expanded.nrrd, 21 MB, uint8, 594×763×599 — does carry
~256 distinct stamp labels, suggesting at least one painted-region geometry
table exists somewhere; but the mapping from stamp → FBbt class isn't reflected
in pdb.)

Reproducer

MATCH (t:Template {short_form: "VFB_00120000"})
      <-[:depicts]-(:Template)
      <-[ie:in_register_with]-(:Individual)
      -[:depicts]->(:Individual)
      -[:INSTANCEOF]->(:Class)
RETURN COUNT(ie) AS n_total,
       sum(CASE WHEN exists(ie.index) THEN 1 ELSE 0 END) AS n_indexed

Returns [754, 4].

Attached curation worksheet

754-row CSV with all the data the curator needs to assign indices:

~/Documents/Claude/Projects/Claude CoWork/reports/VFB_00120000-painted-domain-curation-worksheet.csv

Columns: current_ie_index, channel_short_form, painted_individual_short_form, painted_individual_label, painted_class_short_form, painted_class_label, wlz_folder_url.

Once index values are assigned, a one-shot Cypher write-back keyed on
channel_short_form can land the property on the right relationships in pdb.

Related code-side prep (already ready, not pushed yet)

VFBquery v1.14.6 (run.sh at
~/Documents/Claude/Projects/Claude CoWork/projects/vfbquery-v1.14.6-roi-tree-walk-expansion/)
expands the relationship walk used by get_template_roi_tree so once
ie.index is populated on the leg painted-domain channels, the tree-walking
step can reach the painted leaves from prothoracic leg via
has_sensory_dendrite_in | fasciculates_with | attached_to_part_of | sends_synaptic_output_to_cell (necessary because the leaves are sensory /
motor / muscle classes, not part_of the leg in the ontology).

The patch does not alter the WHERE exists(ie.index) filter — fix belongs
in pdb data.

Wider context

This is one instance of a broader ROI-browser-coverage gap affecting most
templates. Tracked separately at #405.

How this was found

User-reported missing painted domains on
v2-dev.virtualflybrain.org/.../geppetto?q=FBbt_00006048,TransgeneExpressionHere
(the leg template's ROI browser). Live audit against pdb confirmed the
WHERE exists(ie.index) filter is the gate. Robbie Court suggested fixing
upstream in pdb rather than working around in VFBquery; this ticket is the
upstream fix request.

[Robbie1977]

VFB_00120000-painted-domain-curation-worksheet.csv