Shared input gave no clear phase separation beyond muscle position in 20 motor output pools from one larval fly wiring release

We tested whether shared premotor input marks outputs that act in the same crawl phase. We used the released wiring and phase labels of Zarin et al. (2019). We kept one body segment and combined motor neurons with shared target muscles into 20 output pools. Input cosine similarity gave area under the receiver operating characteristic curve (AUC) 0.573 for forward crawl and 0.597 for backward crawl. We enumerated all 51,840 paired-label arrangements that keep six muscle position groups fixed. Exact upper-tail probabilities were 0.0380 and 0.9097. Holm values for the two primary tests were 0.0761 and 0.9097. Neither test passed a 0.05 threshold. The forward result gave a small positive signal with uncertain support after correction. The backward score was below its position reference mean of 0.630. Separate motor rows gave AUCs near 0.64; equal weight for each output-pool pair gave 0.576 and 0.599. Thus, output weights and muscle position affected the apparent structure-function match. These results apply to one reduced wiring release and coarse phase labels from other animals. They do not show that wiring has no role in crawl timing. The package preserves new code, derived data, complete reference scores, figures, provenance, an internal plan, 39 independent checks, and a complete replay. The pinned source data stay external.

connectome · Drosophila larva · motor output · shared input · crawl phase · muscle position · exact permutation test · structure and function