A fixed model gave a 0.000009 mean Brier gain from two-step electrical paths in 15,384 C. elegans atlas pairs

We tested whether two-step electrical paths add information about detected responses to neural stimulation. We used 15,384 ordered pairs from the WT signal propagation atlas in WormNeuroAtlas 0.0.7.3. We formed paths within two separate adult anatomy maps. Five folds kept opposite directions and left/right copies of a class pair together. The fixed logistic model controlled for direct wiring, trial count, neuron identity, bilateral homology, endpoint degree, and other two-step path types. Adding electrical paths raised Brier error by 0.000047576 in map 7 and reduced it by 0.000065852 in map 8. The mean reduction was 0.000009138, or 0.020 percent. Its central 95 percent node-weight range was -0.000159678 to 0.000207933. It includes zero. All planned checks also gave ranges across zero. A wider White whole-map test used 23,384 measured pairs. Its range also included zero. Raw detection rates were about twice as high with an electrical path among pairs without direct wiring or bilateral homology. Thus the raw association did not give a clear added benefit under the full model. The ranges keep predictions fixed and do not measure variation across animals. The test does not establish equivalence or identify routes used by neural signals. The package preserves code, derived inputs, predictions, model states, source hashes, figures, and rebuild instructions. Independent path counts and score calculations passed. A fresh source download reproduced the numerical results.

C. elegans · connectome · gap junction · signal propagation · two-step paths · structure and function · prediction benchmark