What neuromodulator data does VFB hold for neurons?

Asked at the VFB table, NeuroFly 2026 (Cologne), 9 September 2026.

The question

“What information do you have on neuromodulators for neurons?”

A visitor working on aminergic and peptidergic modulation wanted to know whether VFB records which transmitter or neuropeptide a neuron releases, and where that information comes from.

How we answered it

1. search_terms for “neuromodulator” — returns nothing. There is no term or search facet with that name, so the first job was to find how the concept is actually encoded.
2. list_search_facets with contains="ergic" — lists the nine transmitter facets the search index carries, with the number of records under each.
3. get_term_info on peptidergic neuron (FBbt:00004101) — shows the curated pattern: the class is defined as any neuron capable of peptide secretion (GO:0002790), with 312 subclasses for specific peptides.
4. get_term_info on individual EM reconstructions from the hemibrain, male CNS and FlyWire — shows the second pattern: a per-neuron capable of relationship carrying a prediction confidence and a citation.

What we found

Transmitter identity lives at two levels, both using the same relationship — capable of (RO:0002215) pointing at a Gene Ontology secretion process — so a single query pattern finds both.

Curated, at the class level. Ontology classes carry the assertion, and it is inherited by every instance. The umbrella peptidergic neuron class has 312 subclasses, most of them peptide-specific: allatostatin A and C, CCAP, CCHamide-1/2, corazonin, Dh44, Hugin, ITP, leucokinin, Mip, NPF/sNPF, Pdf, PTTH, tachykinin, natalisin, GPA2/GPB5, eclosion hormone and so on, at adult and larval stages. The classical amines follow the same pattern (190 dopaminergic classes, plus octopaminergic, tyraminergic, serotonergic and histaminergic ones).

Predicted, per EM neuron. Where a connectome ships transmitter predictions, VFB attaches them to each reconstruction as capable of with the classifier’s confidence and its source — for example the hemibrain LPN_R reads “43% capable of dopamine secretion, neurotransmission (Eckstein et al., 2024)”, and the male CNS AstA1_L reads “74% capable of GABA secretion (Berg et al., 2025)”. A neuron’s tags reflect both routes, so AstA1_L is Peptidergic through its allatostatin A class and GABAergic through the prediction.

These are counts of VFB records, not of neurons in a fly: the same cell type appears once per connectome it was reconstructed in.

Around the identity itself: class-level connectivity aggregates work on these classes (DownstreamClassConnectivity on peptidergic neuron returns 56,562 rows), 203 driver-line expression patterns are annotated to peptidergic classes, 81 scRNAseq clusters map to them (Davie et al., 2018 and others), and there are 45 neurotransmitter-receptor gene entries.

Worth being honest about. Neuropeptides are never predicted from EM — the classifiers only cover the small molecules — so peptidergic identity on a connectome neuron comes solely from its mapping to an FBbt class (FlyWire’s ITP cell-type label becomes adult ipc-1 neuron, for instance). The peptidergic facet is generic, so “which peptide?” means walking the subclass tree. Per-neuron predictions cover the hemibrain, FlyWire and male CNS but not the older FAFB/CATMAID skeletons, and the receptor side is the 45 gene entries with no per-neuron receptor expression. Finally, the predictions are attributed evidence, and they can disagree between connectomes for the same cell — see the DNp32 entry for a case where three volumes give three different monoamines.

Sources

• FBbt:00004101 — peptidergic neuron
• Eckstein et al., 2024, Cell 187(10): 2574–2594 (FlyWire/hemibrain transmitter predictions, FlyBase FBrf0259490 )
• Berg et al., 2025, bioRxiv 10.1101/2025.10.09.680999 (male CNS)
• Davie et al., 2018, Cell 174(4): 982–998 (scRNAseq clusters)