Peptide channel terms gave an uncertain 1.93% timing error reduction in a nested ridge test of 739 C. elegans kernels

We tested whether possible peptide channels help predict fitted calcium response timing beyond wired structure, cell identity, and expression breadth. We joined a pinned C. elegans signal atlas, corrected Cook wiring, and the February 2024 peptide map. The main panel had 739 directions and 151 neurons. A nested ridge model added 92 channel terms to controls that included channel count. Mean squared log-time error fell from 0.272068 to 0.266815, a 1.93% reduction. A central 95% range from 199 complete network weight refits was -5.56% to +6.89%, where positive values mean lower error. One alternate split gave a loss. The known map exceeded 198 of 199 altered specificity maps, for descriptive rank 0.010. A later baseline with indicators for each observed channel count left a 1.70% gain, with refit range -4.24% to +6.84% and specificity rank 0.020. The count-shape check did not explain the full point gain, but a stable prediction benefit remains uncertain. Timing describes fitted calcium filters, not synaptic delay or receptor kinetics. The ranges are conditional on one atlas and exclude animal variation and kernel-fit uncertainty. The package preserves the joined data, source versions and hashes, code, plans, fitted states, predictions, all refit and control scores, figures, and repeat instructions.

Caenorhabditis elegans · connectome · structure and function · neuropeptide · response timing · molecular specificity · grouped validation · open benchmark