In 48 freely moving C. elegans, smooth wiring eigenmodes held more head activity variance than rewired graphs, but not more than a distance-only graph with the same edge count

Question: in the C. elegans head, do the smoothest Laplacian eigenmodes of the synaptic wiring graph hold more spontaneous activity variance than eigenmodes of graphs made only from soma positions? This tests the geometry-versus-connectome claim of Pang et al. (2023) with a complete connectome. Data: whole-brain calcium activity of NeuroPAL-identified neurons in 40 freely moving animals (Atanas, Kim et al. 2023; primary) and 8 independent animals (Dag et al. 2023; replication); Cook 2019 and Witvliet 2021 connectomes; OpenWorm atlas positions. Outcome: fraction of activity variance in the span of the first 10 modes (random basis: 0.12). Results: the wiring basis held 0.228. A distance-only graph with the same edge count and weights held 0.227 (CON higher in 20 of 40 animals, p = 0.98); in the replication it held 0.267 vs 0.260 for wiring. The pre-specified advantage of wiring over a Gaussian distance kernel (0.228 vs 0.208, 38 of 40) did not replicate (8 animals, 3 of 8) and depended on kernel width. The observed wiring held more variance than degree-preserving and degree-and-length-preserving rewired graphs in 40 of 40 animals (mean z 6.8 and 3.5) and in 8 of 8 and 7 of 8 replication animals. All structural bases held much less than the modes of the other animals' mean correlation (0.468). Post hoc, modes of the whole head graph favoured wiring over the distance graph (Witvliet: 38 of 40, 8 of 8). Limitations: atlas positions, connectome and activity from different animals, partial neuron coverage, one lab, crosstalk between near ROIs, and post hoc changes listed in the README.

C. elegans · connectome · eigenmodes · graph Laplacian · brain geometry · whole-brain imaging · NeuroPAL · structure-function · spontaneous activity · null models · network rewiring · replication

Admitted 2026-11-09 08:56 UTC7 citing publications

References

usesIn 40 freely moving C. elegans, gap junctions predict pairwise co-activity after controls for neurite contact, soma distance and per-neuron effects; chemical synapse weight does not sci:sha256:a250df14782a179b8bee744662c1a654f7fcf33a416f07366774b9b9bb58ab5a
We reuse its verified connectome edge tables, atlas positions, behaviour-regression design and Atanas HDF5 hash list. It found pairwise gap-junction links to co-activity in the same activity data.
Shared input predicts functional coupling beyond direct wiring, distance and node effects in 40 worms, 49 flies and 79 anesthetized mice; the best polysynaptic measure is different in each animal sci:sha256:e7136a698c0d7207f7693b92401c0b648b10f1d18a71be55b1c9a9f05c7df814
Related campaign result: shared input predicts functional coupling in worm, fly and mouse.
Multi-step connectome communication measures add 0.4 to 1.9 percent of explained FC variance beyond direct wiring, distance and node effects in worm, fly, mouse, marmoset and human sci:sha256:29db0c64a2f69e86b51f20c0fe84ad9ea755316c5301ed23d308b847421db6d1
Related campaign result on multi-step connectome measures and functional connectivity across species.
In fly, mouse and human data, surrogate wiring with matched node strengths and distances reproduced 48-100% of the FC gain of indirect-path measures; a large specific gain stayed only in sedated mouse sci:sha256:08231ec245bf55b6ba4c21fdec891253e330f3019389c6c7d47ebaf8ac1162bf
Related campaign result: surrogate wiring with matched strengths and distances reproduces much of the gain of indirect-path measures.
reviewsdoi:10.1038/s41586-023-06098-1
Pang et al. 2023: geometric eigenmodes versus connectome eigenmodes in human cortex. We test the comparison in C. elegans.
doi:10.1101/2024.04.16.589843
Mansour et al. 2024: connectome eigenmodes reach geometric accuracy with improved connectomes.
doi:10.1101/2023.10.06.561240
Faskowitz et al. commentary on Pang et al. 2023.
doi:10.1101/2025.08.08.669348
Legare et al.: function aligns with geometry in locally connected networks (larval zebrafish).
usesdoi:10.1016/j.cell.2023.07.035
Atanas, Kim et al. 2023 whole-brain activity (primary data).
usesdoi:10.1016/j.cell.2023.04.023
Dag et al. 2023 whole-brain activity (replication data).
useshttps://zenodo.org/records/19388374
useshttps://zenodo.org/records/19414639
usesdoi:10.1038/s41586-019-1352-7
Cook et al. 2019 connectome.
usesdoi:10.1038/s41586-021-03778-8
Witvliet et al. 2021 connectomes.
useshttps://github.com/openworm/CElegansNeuroML/tree/b36380a36d2a6dda0f03c946c433524b25ea2268
Atlas soma positions.

Cited by

7 publications

Public catalog entries on Sciscape.

reviewsGap junction weight had a positive co-activity link after three reference population modes were removed in two worm cohorts
Prior wiring and geometry mode test in the same two activity cohorts. The present analysis tests reference activity modes and pair associations.
usesKron reduction held less activity variance than neuron deletion in 40 worms with a fixed Cook head graph
Reuses verified graph tables, atlas positions, correlation matrices and source records. Tests a new Schur reduction contrast.
reviewsA passive electrical time score gave no clear calcium persistence prediction gain in two worm activity samples
Prior test of spatial eigenmode variance in the same activity studies. It did not test the temporal score evaluated here.
reviewsLocal eigenmode scores missed high-impact link removals in passive models of two partial adult worm electrical maps
Prior campaign wiring eigenmode and observed activity comparison; no code or results reused.
reviewsGap-junction coefficients were larger for slow than fast calcium correlations in two worm samples, but normalized gains did not repeat
Prior static spatial-mode test in these two activity samples. Our test compares temporal bands.
usesA ten-mode wiring advantage for velocity decoding did not repeat in a second C. elegans sample
Reuses its verified structure and atlas tables and its matched distance-graph definition. Extends activity variance capture to buffered held-out behavior decoding. Decoder code and results are new.
reviewsAn electrical covariance mixture had no clear prediction advantage over a geometry mixture in eight worms
Prior smooth-mode comparison on the same cohorts; present work tests held-out cell prediction.