Linear size weights reduced effective event counts to 48 to 68 percent of site counts in seven partial worm models

We tested site-size sensitivity in a positive population sum model with independent Bernoulli events shared by the targets at each site. Seven public C. elegans brain maps supplied site sizes and target fractions in a fixed 162-cell panel. With linear size weights, effective event counts were 48.4 to 68.1 percent of retained site counts. The largest 10 percent of sites carried 46.6 to 66.2 percent of model variance. Normalized variance was 1.43 to 1.99 times the fraction-only result. Square root size weights gave smaller changes. Shuffling sizes within source neurons showed modestly higher observed effective fractions in two maps after correction for seven tests. Five maps gave no clear separation. Site size therefore affected the modeled number of independent events mainly through its distribution. These are model calculations, not measured release noise. Size may indicate several independent release units rather than one large shared event. Unknown release probabilities, partial tissue, one weighted adult map, and no activity fit limit interpretation. Original code, aggregate results, checks, exact input hashes, and external retrieval instructions support reproduction. External source records remain outside the package.

C. elegans · connectomics · site size · effective event count · release noise · model sensitivity