https://w3id.org/np/RAK2W5mDAQwecEd127ZGV4l5DEMAtBHMi_JO7eM6DPFas
.trig | .trig.txt | .jelly | .jelly.txt | .jsonld | .jsonld.txt | .nq | .nq.txt | .xml | .xml.txt
@prefix this: <https://w3id.org/np/RAK2W5mDAQwecEd127ZGV4l5DEMAtBHMi_JO7eM6DPFas> .
@prefix sub: <https://w3id.org/np/RAK2W5mDAQwecEd127ZGV4l5DEMAtBHMi_JO7eM6DPFas/> .
@prefix np: <http://www.nanopub.org/nschema#> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix prov: <http://www.w3.org/ns/prov#> .
@prefix npx: <http://purl.org/nanopub/x/> .
@prefix dc: <http://purl.org/dc/terms/> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
sub:Head {
this: a np:Nanopublication;
np:hasAssertion sub:assertion;
np:hasProvenance sub:provenance;
np:hasPublicationInfo sub:pubinfo .
}
sub:assertion {
<https://w3id.org/delphi/resource/assessment/levin-2023-competency-04--claim-test>
a <https://w3id.org/delphi/ontology/Assessment>;
dc:description "analysis.py re-implements the model in vectorised numpy without importing the authors' code: per generation it records the best genotypic fitness, develops each embryo by restricted bubble sort, selects the top 10 on phenotypic fitness, refills to 100 by single-point crossover and point-mutates each individual with probability 0.6. Hardwired and competency 20, 100 and 400 populations are each run 100 times for 250 generations from fixed seeds. For each Table 1 threshold it reports the first generation at which the mean best phenotypic fitness reaches it, and runs one-sided Welch t-tests of each competent population against hardwired at generations 2, 10 and 20.";
<http://www.w3.org/2000/01/rdf-schema#comment> "Measured: 4 generations for the mean best phenotypic fitness of the competency-400 population to reach 1.0 (n = 100, Monte Carlo; fixed seed, deterministic for a given numpy version). Interpretation: This is a simulation-result claim, so the test is replication: re-run the described model and check that the reported dynamics follow from it. Parameters are taken from Supplement S1/S2.1 of the paper (array of 50 cells, values 1-50 drawn with repetition, 100 embryos, top 10% selected on phenotypic fitness, crossover, per-individual point-mutation probability 0.6, competency levels 20/100/400) and unspecified details from the authors' code (github.com/Niwhskal/CellularCompetency, commit 2be29f96): fitness 9^x/9 where x is the normalized non-inversion count; restricted bubble sort that swaps strictly inverted adjacent pairs, visiting positions in a fresh random order each sweep, up to the competency limit; selected parents are kept and children are added by single-point crossover (cut in [11, 50)) until the population is refilled, then every individual is point-mutated with probability 0.6. 'Reaches a threshold' follows the paper's Table 1: the first generation at which the mean over repeats of the best individual's fitness is >= the threshold. The re-implementation is written independently (vectorised numpy) and does not import the authors' code.";
<http://www.w3.org/2000/01/rdf-schema#label> "Claim test of levin-2023-competency-04: supports, confidence 0.9";
<http://www.w3.org/2000/01/rdf-schema#seeAlso> <https://github.com/fractastical/delphisci/blob/bf2af6ee0a84e23ffa6414ad7a0a50a8c276506e/test-data/claim-tests/levin-papers/levin-2023-competency-04/result.json>;
prov:wasAttributedTo <https://w3id.org/delphi/resource/agent/delphisci-claim-tester-claude-opus-5-5>;
<https://w3id.org/delphi/ontology/admissibleDatasets> "Planned datasets: competency-ga-reimplementation-runs";
<https://w3id.org/delphi/ontology/assessmentType> "claim-test/0.1.0";
<https://w3id.org/delphi/ontology/dataKind> "computational";
<https://w3id.org/delphi/ontology/disconfirmingObservation> "Pre-registered: at the 0.97 or 1.0 threshold some higher competency level is strictly slower than a lower one (including hardwired faster than any competent level), or a competent population is not better than hardwired at generation 10 (p >= 0.05 or lower mean). Outcomes meeting neither the pass nor the fail criterion (ordering holds but the magnitudes in (b) or the significance in (c) are not met) are pre-registered as 'mixed'.";
<https://w3id.org/delphi/ontology/evidenceAgainst> "'Under 6 generations at competency 400' is reproduced (4); the claim's 'over 200 for hardwired' is reproduced (230), but that number is sensitive to the 1.0 threshold, which requires every one of the 100 repeats to be perfectly sorted in the same generation.",
"All pre-registered criteria hold: strict ordering at fitness 1.0 (230 > 72 > 33 > 4 generations), weak ordering at every threshold, competency 400 at 1.0 by generation 4, hardwired after generation 100, and every Welch test p < 1e-10.",
"Genotypic fitness plateaus lower the higher the competency (0.97, 0.87 and 0.59 at generation 250 vs 1.0 hardwired), consistent with the companion claim levin-2023-competency-05; that claim is not tested here.",
"The re-run is consistently somewhat faster than Table 1 (e.g. competency 20 reaches 1.0 at 72 vs 93, hardwired at 230 vs 250). The authors' plotting code counts hardwired crossings from 0 and competent crossings from 1, and their committed configuration uses 10 repeats rather than the 100 stated in the supplement; either could account for small offsets. The qualitative claim does not depend on them.",
"This is a replication of a simulation, not an empirical test: it shows the reported dynamics follow from the stated model, not that real cellular competency accelerates evolution.";
<https://w3id.org/delphi/ontology/expectedObservation> "Pre-registered: with 100 repeats of 250 generations per condition, (a) the generation at which the mean best phenotypic fitness first reaches 1.0 is strictly ordered hardwired > competency 20 > competency 100 > competency 400, and the same weak ordering (>=) holds at every Table 1 threshold (0.65, 0.75, 0.8, 0.9, 0.97, 1.0); (b) competency 400 reaches 1.0 by generation 10 and hardwired does not reach it before generation 100; (c) at generations 2, 10 and 20 each competent population's best phenotypic fitness exceeds hardwired with Welch t-test p < 0.001.";
<https://w3id.org/delphi/ontology/failCriterion> "Pre-registered: at the 0.97 or 1.0 threshold some higher competency level is strictly slower than a lower one (including hardwired faster than any competent level), or a competent population is not better than hardwired at generation 10 (p >= 0.05 or lower mean). Outcomes meeting neither the pass nor the fail criterion (ordering holds but the magnitudes in (b) or the significance in (c) are not met) are pre-registered as 'mixed'.";
<https://w3id.org/delphi/ontology/hasConfidence> "0.9"^^xsd:double;
<https://w3id.org/delphi/ontology/hasEvidence> <https://w3id.org/delphi/resource/evidence/levin-2023-competency-04--claim-test--dataset-1>;
<https://w3id.org/delphi/ontology/hasVerdict> "supports";
<https://w3id.org/delphi/ontology/hypothesis> "In an independent re-implementation of the Shreesha & Levin (2023) artificial-embryogeny genetic algorithm, the best individual's phenotypic fitness (mean over repeats) crosses each fitness threshold sooner the higher the competency level, with competency 400 reaching fitness 1.0 within a few generations and the hardwired population taking far longer.";
<https://w3id.org/delphi/ontology/inferenceMethod> "analysis.py re-implements the model in vectorised numpy without importing the authors' code: per generation it records the best genotypic fitness, develops each embryo by restricted bubble sort, selects the top 10 on phenotypic fitness, refills to 100 by single-point crossover and point-mutates each individual with probability 0.6. Hardwired and competency 20, 100 and 400 populations are each run 100 times for 250 generations from fixed seeds. For each Table 1 threshold it reports the first generation at which the mean best phenotypic fitness reaches it, and runs one-sided Welch t-tests of each competent population against hardwired at generations 2, 10 and 20.";
<https://w3id.org/delphi/ontology/passCriterion> "Pre-registered: with 100 repeats of 250 generations per condition, (a) the generation at which the mean best phenotypic fitness first reaches 1.0 is strictly ordered hardwired > competency 20 > competency 100 > competency 400, and the same weak ordering (>=) holds at every Table 1 threshold (0.65, 0.75, 0.8, 0.9, 0.97, 1.0); (b) competency 400 reaches 1.0 by generation 10 and hardwired does not reach it before generation 100; (c) at generations 2, 10 and 20 each competent population's best phenotypic fitness exceeds hardwired with Welch t-test p < 0.001.";
<https://w3id.org/delphi/ontology/preregistered> "false"^^xsd:boolean .
<https://w3id.org/delphi/resource/claim/levin-2023-competency-04> <https://w3id.org/delphi/ontology/hasAssessment>
<https://w3id.org/delphi/resource/assessment/levin-2023-competency-04--claim-test> .
<https://w3id.org/delphi/resource/evidence/levin-2023-competency-04--claim-test--dataset-1>
a <https://w3id.org/delphi/ontology/EvidenceItem>;
dc:description "best-individual genotypic and phenotypic fitness per generation for hardwired and competency 20/100/400 populations; mean curves in derived/mean_curves.json";
dc:license "";
<http://www.w3.org/2000/01/rdf-schema#label> "Dataset: Re-implementation of Shreesha & Levin (2023) Experiment 1, 100 repeats x 250 generations x 4 populations (generated by analysis.py)" .
}
sub:provenance {
<https://w3id.org/delphi/resource/agent/delphisci-claim-tester-claude-opus-5-5> a
prov:SoftwareAgent;
<http://www.w3.org/2000/01/rdf-schema#label> "delphisci claim tester (claude-opus-5.5; cursor local session)";
prov:actedOnBehalfOf <https://orcid.org/0009-0007-4948-9192>;
<https://w3id.org/delphi/ontology/usesModel> <https://w3id.org/delphi/resource/model/claude-opus-5-5> .
<https://w3id.org/delphi/resource/model/claude-opus-5-5> a <https://w3id.org/delphi/ontology/LanguageModel>;
dc:identifier "claude-opus-5.5" .
sub:assertion prov:wasAttributedTo <https://w3id.org/delphi/resource/agent/delphisci-claim-tester-claude-opus-5-5>;
prov:wasDerivedFrom <https://github.com/fractastical/delphisci/blob/bf2af6ee0a84e23ffa6414ad7a0a50a8c276506e/test-data/claim-tests/levin-papers/levin-2023-competency-04/result.json>;
prov:wasGeneratedBy sub:test .
sub:plan prov:hadPlan <https://github.com/fractastical/delphisci/blob/bf2af6ee0a84e23ffa6414ad7a0a50a8c276506e/test-data/claim-tests/README.md> .
sub:test a prov:Activity;
prov:qualifiedAssociation sub:plan;
prov:wasAssociatedWith <https://w3id.org/delphi/resource/agent/delphisci-claim-tester-claude-opus-5-5> .
}
sub:pubinfo {
this: dc:created "2026-11-08T16:24:31.879Z"^^xsd:dateTime;
dc:creator <https://orcid.org/0009-0007-4948-9192>;
dc:license <https://creativecommons.org/licenses/by/4.0/>;
npx:hasNanopubType <https://w3id.org/delphi/ontology/Assessment>;
npx:introduces <https://w3id.org/delphi/resource/assessment/levin-2023-competency-04--claim-test>;
<http://www.w3.org/2000/01/rdf-schema#label> "Claim test of levin-2023-competency-04: supports, confidence 0.9";
<https://w3id.org/delphi/ontology/domain> <https://w3id.org/delphi/resource/domain/levin-papers> .
sub:sig npx:hasAlgorithm "RSA";
npx:hasPublicKey "MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAyFHtPB6p0mCjDozxlOY9CmO3nBm7Te3jFbXyLmM+G1tGe47Da0jYIbhkl9Oo1ZjxmCa3LCwlEqK8gj6hEIcL4ahW30JXh9HycfKVk1bugkzMEV15tJdTzAU3sGMFnbNsn43xY2LFLy45tApkRm5byEiM7oRRrooIdwoQKnP4Wzq/bfAoJ9qvW+MGkelyOTd1jbSt+KfmT7xKmDfWzAHddUlWwMX2QJOLxDaffOteTop6ZRgPiNmPJKA/S62Gac9XtSYT3oDkZ6/nr5cxxNLpOyxTf5FPISiCe11xYO/unYjY+mwnOfA8yhJIyx1l0O9I0vpOpn+E5TvyHJV+oSTrvwIDAQAB";
npx:hasSignature "ivsWzvB+nM+JlX73uR2WebkR56/BnwqBfJLMTKHlQ+HRh2iTijWtffUsdygDRir+U1BIBK0mjN8k6Qh/k/GtzR7iolPONxsv7UdZJnSS7ANerIppJ0efo4kiwDNxW99hSMkOpC4tKu4fFH7i1h2dMdcLQT2DvvDpebtkiPHrYc6G95NkiElGHYP1soBdZefCdbI1QboszJ8dmlcxJ4BOVnUYg5TDUeGGnozBbgfpcnzLtkyhw2emDPmFAMsxX/K0Nxpfv1zaZJxc8ZYvncaJaVM4xgGt0N5DzvL+k3SD1v7OXVaVQfh2rQCm/+AtfTjIKTewDyjPB1E7YuybDjnvCQ==";
npx:hasSignatureTarget this:;
npx:signedBy <https://orcid.org/0009-0007-4948-9192> .
}