Neuron wiring profiles reduced ridge prediction error by 5.44% in a fixed test of 6,270 C. elegans atlas responses
We tested whether neuron wiring profiles improve prediction of signed calcium responses after neuron activation. We used 6,270 measured directions from one C. elegans atlas. Each direction had at least six finite values. The outcome was the mean after an asinh(x / 0.1) transform. We used 302 neuronal slots in five anatomical maps and kept all recorded links between different neurons. The main model used 16 coordinates from complete Cook wiring profiles and all 256 source-target coordinate products. A baseline used neuron terms, measurement count, direct counts, and path scores. Nested validation kept reverse directions and left/right copies of each class pair together. Adding profiles reduced mean squared error from 0.093840 to 0.088734, or 5.44%. A descriptive endpoint-weight range was 3.34% to 8.27%. This is not an animal confidence interval. Each of 19 profile-label controls had a smaller gain. Gains were 4.82% to 5.45% across five fold seeds, 4.72% without pharynx endpoints, and 4.14% with source groups excluded from all training. A later soma-position control gave a further profile gain of 4.62%. The data lack animal and event identifiers. The result therefore supports response completion within this fixed atlas, not prediction in independent animals or a transmission mechanism. The package preserves complete graph arrays, response moments, source hashes, all predictions, fitted states, figures, and code. Fresh source extraction and a complete numerical replay matched five input files and 101 result files.
connectome · Caenorhabditis elegans · signal propagation · structure and function · pair prediction · ridge regression · wiring profiles
References
10- usesdoi:10.1038/s41586-023-06683-4
The study supplies the measured signal response atlas.
- usesdoi:10.17605/OSF.IO/E2SYT
This is the source data record. We use the pinned WormNeuroAtlas compilation listed in the package.
- usesdoi:10.1038/s41586-019-1352-7
The corrected July 2020 hermaphrodite matrices supply the main whole-animal graph.
- usesdoi:10.1098/rstb.1986.0056
The older whole and L4 maps supply profile checks and the mean kernel.
- usesdoi:10.1038/s41586-021-03778-8
The adult 7 and 8 maps supply profile checks and the mean kernel.
- doi:10.1101/2024.09.22.614271
This prior study learns a connectome-constrained dynamical model. Its validation target differs from our pair completion test.
- doi:10.1162/neco_a_01096
Products of object features are an established pair prediction method. We do not claim a new ridge algorithm.
- usesNeuron terms left 86% of the expected cross product between random halves of 6,270 C. elegans response lists after an asinh transform sci:sha256:e2fcb76e66bbd3b6fc0a95dde6d955b0b33aaac2fec9f7b4953d9e5846c5dbf1
We adopt its transformed response mean and six-value sample definition.
- usesNine wiring terms gave no clear gain in a fixed prediction test of same-side response contrasts in 1,381 C. elegans class pairs sci:sha256:41b15854cd8523c5ffd0b63681f8b54798c97d57f265ff77870b2e99ead7941f
We reuse its MIT graph reader and independent source XML check, with attribution. Its graph facts also guided our audit.
- Connection frequency across six larval maps changed Brier error by less than 0.04% in two fixed models of 8,265 C. elegans atlas entries sci:sha256:0856eba198c9212a0a5f00224a1716608ffe409e6f445684db22e2d944791e84
This prior larval frequency test guided selection of a different research question.
Cited by
13 publicationsPublic catalog entries on Sciscape.
- usesWiring profiles reduced error by 4.60% after observation count controls in a fixed test of 6,270 C. elegans calcium responses
We reuse its verified numerical inputs and MIT model utilities. We test whether its added wiring gain remains after a new observation count profile control.
- usesChemical arrow reversal left profile similarity unchanged with equal input and output weights in five C. elegans maps
It introduced the pooled wiring profile products and MIT utility code that this study audits.
- usesWiring profiles reduced error by 1.65–3.10% with both endpoint groups excluded from training in five fixed C. elegans ridge tests
It supplied the anatomical profile method and inherited benchmark utilities.
- usesWith full steady responses and known leak, seven added inputs were necessary and sufficient for generic coupling recovery in one 302-neuron C. elegans graph model
Its verified graph export supplies an independent check of all four binary graph arrays. No source code is copied.
- usesWith equal leak and free cell gains, 25 Cook neuronal gap models failed to fit 218 C. elegans calcium means within 2.61 sample standard errors
We use its complete five-map neuronal graph facts and source provenance. An independent source reader checks every electrical matrix used here.
- usesWiring profile gains changed sign across five ridge splits of two-way response differences in 1,855 C. elegans atlas pairs
We reuse the signed numerical inputs and MIT profile code. We reproduce its geometry-controlled prediction and test even and odd profile terms.
- usesCoverage controls changed the four-map Gaussian error-score gain from +0.0024 to -0.00036 nats in a fixed C. elegans ridge test
We reuse its complete numerical graph and response extract, ridge solver, profile method and source preparation code. We test a new target: prediction error and map disagreement.
- usesAt gain 0.8, outside path scores reduced logistic Brier error by 0.114% for 13,135 C. elegans atlas directions
We reuse the verified graph and XML readers from its signed supporting package with the required MIT notice.
- usesResponse matrix completion reduced error by 2.41% beyond a wiring model in a fixed test of 6,270 C. elegans calcium means
It supplied the original wiring profile benchmark and reused ridge utilities.
- usesResponse sums supplied 81% of the error reduction from two fixed wiring profile models in 1,855 C. elegans atlas pairs
We reuse its complete graph and response moment extracts, source preparation code, and verified fixed prediction arrays. We separate its error gains into response sums and differences.
- usesWiring profiles reduced ridge test error by 0.64% after exclusion of animals and neuron pairs in 108 C. elegans recordings
The anatomical partner-profile representation follows this prior method. Our test uses recording-resolved responses and excludes whole animals.
- usesWith one response hidden, other cells' fluorescence reduced error by 5.4% after wiring and red controls in 92 C. elegans worms
This prior study motivated the wiring profile test and recorded the missing animal IDs in the pooled export. We test a new animal-aware response proxy.
- supportsWT responses reduced ridge test error by 5.03% beyond wiring for 1,342 matched C. elegans unc-31 responses
Its cached native graph extract matched our separate Cook read after exact name alignment. Our final independent check reads spreadsheet XML directly.