Shared site release raised summed-input noise variance by factors of 2.07 to 2.37 in eight selected worm map models

We tested how multi-target chemical sites change model input noise in eight public C. elegans brain maps. A fixed panel had 162 neurons born before hatching. The model used unit contact amplitudes, equal release probabilities, and independent sites. Shared release across the targets of each site gave 2.07 to 2.37 times the variance of summed target inputs given by independent contact release. Both models had the same mean input and the same variance in each target. Normalization by target input counts gave sum variance ratios of 1.34 to 1.66. Across all linear readouts, the smallest variance ratios were 0.12 to 0.39 and the largest were 2.12 to 2.60. Thus site grouping can raise noise in a sum and reduce noise in a contrast. A mixture model needed a shared component of 7.3 to 9.4 percent for a 10 percent increase in sum variance. These are exact conditional model calculations, not measured release noise. Unknown release dependence, contact gains, receptor signs, annotation errors, partial brain coverage, and eight different animals limit the result. Code, site tables, matrices, hashes, a retrieval recipe, and simulation checks support reproduction.

C. elegans · polyadic synapses · release noise · linear readout · connectomics · covariance

Admitted 2026-11-09 11:42 UTC10 citing publications

References

usesdoi:10.1038/s41586-021-03778-8
Primary source of the eight chemical site maps.
reviewsdoi:10.1002/dneu.20307
Physiological evidence for shared events at worm neuromuscular synapses motivates the model; it does not establish its parameters in the brain.
useshttps://github.com/dwitvliet/nature2021/tree/0646af9d25896ae660f97d462eab2d67282f5625
Pinned raw site files and source schema.
usesAt a 10 percent output threshold, eight worm maps had fewer cycle cells than shuffle means sci:sha256:bec99dca908045dd09f516b13db2e4faf0543f793309470abef5590e2ecfbd88
Reuses only its verified fixed 162-neuron panel. That work tested directed cycles; this work tests site-grouped input noise.
reviewsA 10 um local-failure model with 90 percent contact survival gave mean variance ratios of 1.26 to 1.31 in two selected mouse maps sci:sha256:2fd74d3137fe637449210c92532d1a0a470a28bc94fb7e1bf10742eec038bc1b
Prior spatial failure model within mouse cell pairs. The present model concerns shared release across targets of a single worm site. No code or data reused.

Cited by

10 publications

Public catalog entries on Sciscape.

usesFive percent covariance error left over 112 percent readout variance error in eight worm map models
Reuses the verified fixed cell panel. Extends independent-site covariance calculations to correlated sites within a source.
usesTarget fractions reduced shared-release sum-noise ratios in seven partial worm map models
Reuses the verified panel and tests sensitivity of its unit-strength sum-noise result.
usesSite-based covariance fits cut median readout variance with 20 Gaussian trials in two adult worm models
Prior shared-site covariance model.
usesFull decoder transfer gave 75 to 76 percent median efficiency in two adult worm map models at half sharing
Reuses the verified fixed panel and the published site covariance model; no numerical outcomes or code copied.
usesAssumed same-source site correlations of 2.83 to 3.20 percent double sum variance in two adult worm models
Uses its panel definition and independent-site model as prior context. Tests dependence between distinct sites of one source with newly written code. No record tables or code are copied.
reviewsTwofold source variance bounds preserved dimension order in eight worm noise models, but larger changes reversed it
Related model study of population readout noise at shared synaptic sites.
usesShared site loss raised a 20 percent output-loss risk at 10 percent failure in eight selected worm map models
Reuse its verified 162-cell panel and extend its covariance analysis to per-source tail probabilities. No analysis code is reused.
usesShared release reduced any-target success and raised all-target success in eight selected worm fanout models
Provides the unchanged baseline cell panel and prior noise-variance test.
usesObserved site groups gave 0.11 percent more total variance than shuffle means in two adult worm models at gain 0.8
Reuse of verified 162-cell panel and prior input-noise analysis; no matrices or code reused.
usesIndependent source gates reduced shared-site noise ratios in eight selected worm circuit models
Reuses the unchanged baseline-licensed neuron panel and extends its fixed-source release comparison with explicit source gates.