MN9 ranked joint second by extension AUC among 85 motor read-outs in a fixed FlyWire test of 58 cell types with excitatory labels

We tested whether an anatomical score for rostrum extension was specific to motor neuron MN9. We used all 127,400 neurons and 52,793,639 synapses in the FlyWire v630 source graph for the Shiu et al. brain model. A fixed score summed excitatory walks of one to three edges, with each edge divided by all input synapses to its target. We matched 85 motor neurons by exact root ID to a pinned annotation. The main sample had 58 stimulated cell types with at least one excitatory model label and ten positive outcomes. MN9 had AUC 0.9667, with a pointwise 95% bootstrap interval of 0.9167 to 1.0000. It ranked joint second of 85 motor read-outs. The largest alternative AUC was 0.9729, and 25 read-outs had AUC at least 0.90. No maximum from 5,000 label permutations reached the observed MN9 AUC. Across all 106 types, MN9 AUC was 0.7702. Four positive types had only inhibitory labels and received zero excitatory scores. The result links short anatomical walks to a relevant output, but the score does not select a unique motor neuron. The score was chosen in earlier work on this screen. This is an output identity test in one connectome, with incomplete motor ID coverage, not an independent test cohort.

connectome · Drosophila · FlyWire · structure and function · motor neurons · rostrum extension · output specificity · signal propagation