Observed site groups raised model failure clustering by 1.4 to 2.3 percent in eight partial worm maps at failure probability 0.5

We tested whether chemical site groups change complete connection failures beyond pair counts. Eight public worm maps supplied a common panel of 141 cells with presynaptic records in every map. The model assumes independent site failures and complete shared loss at each site. At failure probability 0.5, observed groups gave 1.39 to 2.30 percent more companion failed outputs than the means of controls with the same source-target counts and site sizes. Each map had two chains of 200 draws. All observed primary values exceeded all retained controls. The expected number of failed outputs stayed fixed. At failure probability 0.1, the excess was 7.82 to 20.42 percent; at 0.9, it was 0.067 to 0.160 percent. Independent set-union calculations and small exhaustive checks passed. These are model results, not measured release failures. Partial coverage, unknown release dependence, one species, and controls without spatial constraints limit transfer. The package contains original code, aggregate results, hashes and retrieval steps. External raw data and record-level extracts are excluded.

polyadic synapses · connection failure · C. elegans · fixed-count controls · reliability · model test