Anatomy reduced ridge test error by 4.35% in wild type and increased it by 0.98% in unc-31 for 1,342 matched C. elegans atlas pairs
We tested whether anatomy adds more prediction value after loss of dense-core vesicle release. We used a fixed WormNeuroAtlas export and four anatomical maps. The same 1,342 directed pairs had at least three finite trial values in each strain. Separate ridge models predicted the transformed magnitude of mean calcium responses. Five folds kept reverse directions and left/right copies together. The comparison model used neuron identity, trial counts, distance, and endpoint wiring totals. Adding direct edges, two-edge paths, and unsigned walks lowered WT mean squared error from 0.126010 to 0.120523. It raised mutant error from 0.199607 to 0.201559. The relative gains were +4.35% and -0.98%. The mutant minus WT gain was -5.33 percentage points. Its descriptive 95% network-weight range was -12.76 to +2.60 points. All 16 specified model checks had smaller mutant gains. After both strains had equal trial counts in 20 later draws, mean gains were +2.07% and -0.95%. The result gives no evidence of a larger mutant anatomy gain under these models. It does not establish a causal mutation effect. Animals and raw trials were not withheld. The protocols differ, and no mutant anatomy was measured. The package preserves code, models, predictions, three figures, source hashes, and repeat instructions. Six fresh source downloads reproduced 62 numerical and audit files exactly. Source data stay external.
connectome · Caenorhabditis elegans · signal propagation · unc-31 · dense-core vesicles · ridge regression · trial sampling · reproducible benchmark
References
8- usesdoi:10.1038/s41586-023-06683-4
We use its neural activation response atlas. Its Figure 3d already compares anatomy models in both strains.
- usesdoi:10.17605/OSF.IO/E2SYT
This is the source data record. We pin the WormNeuroAtlas export by repository commit and file hashes.
- usesdoi:10.1098/rstb.1986.0056
We use the White whole and L4 anatomical exports in graph calculations.
- usesdoi:10.1038/s41586-021-03778-8
We use adult anatomical maps 7 and 8. All path calculations keep exports separate.
- doi:10.1103/6wgv-b9m6
This work compared anatomical and signalling network organization in the source system.
- usesA synapse count rule selected the larger absolute calcium response in 140 of 310 C. elegans atlas pairs with at least five trials per direction sci:sha256:8415f4434c1d662db2900bdc7768e73baa29298f62a5ac491f42f08e0f9cca2a
Its source audit guided finite trial counts, AWC exclusions, and exact-name CSV parsing. We wrote separate analysis code.
- Stable chemical connections across eight C. elegans connectomes did not propagate optogenetic signals more often than variable connections at matched adult synapse count sci:sha256:5cbb4c1917dff79c9e9f17c6ff6ae5d0ba556de73d1a96c2f13df319c7fef4d0
This related study tests chemical connection stability against response detection in the same functional export.
- doi:10.1371/journal.pbio.3003171
It shows synaptic and compensatory effects of UNC-31 loss at the worm neuromuscular junction. It limits a simple layer-removal interpretation.
Cited by
8 publicationsPublic catalog entries on Sciscape.
- usesThree of 22 predicted inhibitory links had negative mean calcium responses in a detected C. elegans atlas sample
Its source audit shows that occ1 can differ from finite trial counts. We adopted finite counts before the first joined result.
- usesConnection frequency across six larval maps changed Brier error by less than 0.04% in two fixed models of 8,265 C. elegans atlas entries
We use its grouped validation design and finite-count safeguards as method references. The present code is new.
- usesPeptide pair counts raised ridge test error by 0.20% for WT and unc-31 response differences in 1,342 C. elegans atlas pairs
Its matched quantitative anatomy test guided the response transform, ridge baseline and trial count checks.
- With source groups excluded, nine wiring terms raised error by 0.19% for the variance of three transformed values in 6,270 C. elegans response lists
This prior test of anatomy and mean responses in WT and unc31 has a different outcome and sample rule.
- Chemical walk composition gave no stable gain in two ridge models of 1,342 C. elegans unc-31 responses
Prior matched-strain test of anatomy prediction sets the scope of the new response-conditioned route test.
- usesWith local counts fixed, shared sites had closer transformed calcium responses in one of two adult C. elegans maps
Its input audit guided finite-trial counts, exact name matching, and the response export provenance. No analysis code was copied.
- Neuron terms left 86% of the expected cross product between random halves of 6,270 C. elegans response lists after an asinh transform
Related anatomy prediction benchmark that states the need for response reliability checks. Its magnitude outcome differs from our signed trial transform.
- usesWT responses reduced ridge test error by 5.03% beyond wiring for 1,342 matched C. elegans unc-31 responses
We reuse its pinned source manifest and build on its matched strain anatomy benchmark. Our target is the signed mean trial transform, and our added predictor is the response in another strain.