Output wiring similarity gave no clear response-profile prediction gain in 6248 selected worm cell pairs

We tested whether cells with similar chemical output patterns cause similar responses across downstream cells in C. elegans. We linked the wild-type signal propagation atlas to two adult wiring maps. The primary sample had 6,248 source pairs from 115 cells and 68 name-based families. Five-fold validation held out source families. A model with output cosine similarity reduced squared error by 0.272 percent for adult map 7 and 0.219 percent for adult map 8. The paired reductions were 0.00009501 and 0.00007670. Their conditional source-resampling 95 percent limits were -0.00009174 to 0.00029636 and -0.00009377 to 0.00025977. All six planned sensitivity comparisons also had limits that included zero. Thus, the added value remains uncertain in this conditional model. The test uses measured response scales and does not predict unmeasured profiles from anatomy alone. Different anatomy animals, pooled responses, shared targets, fixed-fit resampling and missing geometry controls limit interpretation. Code, source hashes, pair tables, fold predictions and a figure support reproduction.

C. elegans · connectome · signal propagation · shared output · response profiles · source holdout