Random contact sampling reduces connection count and reciprocity differences among 162 neurons in eight C. elegans brain maps

We tested how contact count affects two network measures in eight published C. elegans brain maps. The main test kept 162 neurons marked embryonic in the source table. Original connection counts ranged from 675 to 1,632. We sampled each map without replacement to 1,152 contacts. Exact expected connection counts then ranged from 658.18 to 694.20. The range was 96.2 percent smaller. The range of mean reciprocal fractions was 69.8 percent smaller. At hatching, 675 observed connections were within the sampling ranges of both adults. Adult means were 689.87 and 678.19. The adult sampling model did not fit all maps. Larvae at 16 and 27 hours had fewer connections than both adult models. These deficits remained after feasible source neuron output counts were matched. A comparison between the two adults also fell outside the model ranges. Thus, model mismatch alone cannot identify a developmental cause. Adding the 18 neurons marked for later brain entry changed the hatching result. The findings apply to the fixed neuron scope and the two tested measures. They do not show that development adds contacts at random. The package preserves aggregate inputs, exact formulas, code, source hashes, sampling results, figures, and repeat steps. Original source files remain external. The main results can be repeated offline from the aggregate inputs.

C. elegans · developmental connectome · contact count · rarefaction · reciprocity · null model · reproducibility