We tested whether a molecular peptide map adds information about neural responses after anatomical wiring and gene expression breadth are known. We joined four anatomical exports, a 92-pair molecular map and the wild type (WT) and unc-31 signal propagation atlas. All 301 neurons with anatomical records could act as relays inside their own maps. The main sample had 1,278 directions with at least three finite trial values in each genotype. Five folds withheld source expression groups and removed their labels from both training roles. Adding a peptide-receptor match count raised Brier loss from 0.082753 to 0.084248 in WT and from 0.052261 to 0.052552 in the mutant. WT gains changed sign under two other fold assignments. All planned descriptive ranges for added gains included zero. Of 130 WT detections, 45 met only the mutant equivalence criterion, while 64 met neither mutant criterion. The test did not establish added prediction value for this count. It did not directly test biological response loss. Sparse mutant data, a weak mutant baseline, fixed models and data from different animals limit the result. The package gives the paired benchmark, channel lists, source versions and hashes, original code, fitted states, forecasts, figures and reproduction steps.
Filters: arxiv: q-bio.MN (Molecular Networks) ×
We tested whether known peptide and receptor gene matches explain a difference in response detection between wild-type and unc-31 C. elegans. We joined a 302-neuron peptide network to a published signal atlas. The main test used 1,342 directed pairs with at least three finite trials in each strain. The model controlled for wiring from four anatomical exports, neuron expression groups, trial counts, soma distance and the pharynx. The peptide score had a partial correlation of 0.0326 with wild-type minus mutant detection. Controls kept each gene interaction count fixed and changed the molecular matches. The one-sided rank estimate was p = 0.137 from 1,000 control networks. The fitted effect was 0.99 percentage points per residual score standard deviation. A descriptive neuron profile weight range was -2.03 to 4.01 points. This is not an animal confidence interval. A planned check outside the pharynx gave r = 0.0629, with p = 0.050 and 0.058 in two control samples. Response size gave almost no association. These results show no clear excess for the aggregate score in the main test. They do not exclude individual peptide pathways. Detection labels do not test a response difference between strains. The package preserves the full molecular input, joined pair data, all control graphs, source hashes, code, figures and repeat instructions.
We tested whether a dense high-degree core in an inferred peptide map exceeds a reference that keeps complete receptor profiles. We used the February 2024 C. elegans map, expression threshold 4, 302 neuron labels, 127 expression groups and 92 molecular gene matches at 500 nM. The main graph excluded equal endpoints and pairs within one expression group. It had 30,824 edges. The reference exchanged whole receptor profiles only between groups with equal receptor counts and equal neuron copy counts. Peptide profiles, molecular matches and anatomy stayed fixed. Each graph selected its own 151-neuron core, with exact averaging at degree ties. Core density divided by whole-map density was 2.0431. In 20,000 independent assignments, the reference mean was 2.1031 and its central 95 percent range was 2.0585–2.1513. The recorded ratio was 2.85 percent below that mean. The two-sided reference tail value was 0.00580. Absolute core density stayed inside its reference range. The ratio was also below its range when within-group pairs were kept or broad cell type was matched. It was inside the range at 100 nM and in mid- and long-range checks. Thus this result depends on the setting. All 80,000 sampled core scores passed a second calculation. A clean process reproduced all numerical records. The package preserves source files, provenance, code, assignments, per-neuron results and figures. This reference changes neuron degrees and does not repeat a fixed-degree rich-club test. The results describe potential interactions in one map, not measured signaling or differences between animals.
We tested whether two-way peptide links exceed a reference that keeps complete receptor profiles and expression breadth. We used the February 2024 C. elegans map, 302 neuron labels, 127 distinct source expression groups, and 92 molecular gene pairs at 500 nM. The main short-range graph excluded pairs within the same expression group. A control exchanged whole receptor profiles only between groups with equal receptor gene counts and equal neuron copy counts. This kept receptor gene frequency and co-expression at both group and neuron levels. The graph had 30,824 directed edges and 9,736 unordered two-way pairs. The fraction of edges with a reverse edge was 63.17%. In 20,000 independent assignments, its mean was 62.52% and its central 95 percent range was 60.87–64.14%. The two-sided reference tail value was 0.437. Each of seven planned checks also placed the fraction inside its reference range. At 100 nM, the edge count exceeded its reference range, while the two-way fraction did not. Exact count means, exhaustive small cases, all saved scores, full adjacency checks, and a clean numerical repeat passed. This result concerns a conditional expression model. It does not establish equivalence, physiological feedback, or variation between worms. The package preserves source files and hashes, code, all 160,000 assignments, results, figures, and repeat instructions.
We tested whether a route can survive each single gene loss and still depend on a pair of genes. We used the 2 February 2024 C. elegans peptide map, expression threshold 4, the 500 nM assay cutoff, and the short-range model. A gene loss removed every channel that used that gene. Other map entries stayed fixed. We tested all 100 single losses and all 4,950 pairs of losses. Of 73,517 ordered neuron pairs that kept a route after each single loss, 23,034 (31.33 percent) lost every route after some pair of losses. For 1,581 pairs, at least one such cut left an output at the source and an input at the target. Two gene combinations accounted for 1,578 of these pairs. Both removed all 18 inward edges at the pharynx boundary. For 270 pairs, every original source output edge and target input edge remained. The broader range models had no cuts of this type at 500 nM. The package gives complete cut tables, graph certificates, code, input hashes, and repeat instructions. Independent graph checks and a complete repeat passed. These are exact results for fixed support maps. They do not show that mutant neurons lose a physiological response.
We tested how exact peptide and receptor gene pairs affect the order of highly connected neurons. We used the February 2024 C. elegans peptide map: 302 neuron labels, 49 peptide genes, 51 receptor genes, and 92 pairs at 500 nM. Every altered map kept each gene's partner count, all expression profiles, and all spatial limits fixed. In 8,000 altered maps, an average of 84.6% of molecular pairs changed. The short-range neuron degree ranks had mean Spearman correlation 0.939 with the source map (central reference range 0.904–0.965). Expected overlap with the original top 25 was 88.0%; overlap with the top six was 52.7%. Uniform tie selection was included in these scores. Density within the high-degree half was 0.698 in the source and had a reference mean of 0.695 (range 0.653–0.758). Within-family changes and a 100 nM cutoff gave the same broad pattern. The full long-range rank correlation was 0.852, and ranks within its original top quarter were much less stable. Independent sampling and arithmetic checks passed. These results separate stable overall degree order from more sensitive leading-neuron selection in this fixed model. The altered molecular matches are controls, not measured binding pairs. The ranges do not describe variation between worms. The package preserves all source inputs, hashes, 18,000 altered molecular graphs, scores, code, figures, and repeat instructions.
We tested how gene loss changes directed reachability in the fixed 2024 C. elegans peptide map. At expression threshold 4 and a 500 nM biochemical cutoff, 83 of 100 single gene removals leave all reachable ordered pairs intact in the short-range graph. All 12 links that leave its 20-neuron pharyngeal set use one channel, NLP-13 to DMSR-2, and end at RIPL or RIPR. Removing these links, NLP-13, or DMSR-2 loses the same 5,478 reachable pairs. These include all 5,440 pairs from the 20 pharyngeal neurons to 272 neurons outside that set. No neuron loses all inputs or all outputs. Removal of DMSR-2 support at both RIP cells gives the full route effect; removal at either cell alone loses no pair. The mid-range and long-range models retain all reachable pairs after either global gene removal at cutoffs of 50 to 500 nM. Source reporter tables support expression in RIP and three of the four source classes. Removing support from the class with no detected reporter signal leaves the route result unchanged. The package contains all gene results, boundary certificates, source records, code, and figures. The result describes possible routes under fixed expression and diffusion rules. It does not establish a functional peptide cascade or a biological gene-loss effect.
We tested gene deletion resistance in the published 2024 C. elegans neuropeptide map. The primary scope was 302 neurons, 92 gene couples, short-range diffusion, and a 500 nM EC50 cutoff. We kept gene expression and all other support fixed. A connection retained support if any listed peptide precursor and receptor gene couple remained. We applied the established bipartite matching theorem to find the smallest gene deletion set for each direct connection. We excluded self connections. Of 20,278 connections with multiple channels, 5,407 (26.7%) lost all listed support after one gene loss. Shared receptor genes accounted for 3,959 cases. Shared peptide precursor genes accounted for 1,448 cases. DMSR-7 accounted for 2,786 cases. The fractions were 26.5% and 26.4% in the mid-range and long-range models. In the short-range model, the fraction fell to 8.5% at 10 nM. Thus, the level of vulnerability depends on the biochemical cutoff. We preserve connection tables, support sets, minimum deletion sets, and matching certificates. Checks against published aggregates, exhaustive small graphs, independent software, and exact numerical repeats passed. These findings concern listed support in a fixed map. They do not measure neural or behavioral effects of gene deletion.