GRADIENT-3 Pipeline v1
Identification
- Clé publique
- radar:r-787ce7a2b3fd4c9e70a4799975061d80
- Auteurs
- Pretorius, Eugene
- Année déclarée
- 2026
- Langue déclarée
- en
- Type source
- Software
- Première détection
- 2026-11-13T08:56:58.120135+00:00
- Dernière observation des métadonnées
- 2026-11-13T08:56:58.120135+00:00
- Dernière vérification HTTP
- Non effectuée
Résumé documentaire — langue originale
GRADIENT-3: Zenodo Repository Metadata REGISTERED — April 2026 Discrete Substrate Signatures in the Arabidopsis Circadian Transcriptome: A Triadic Derivation and Empirical Non-Falsification Program for Kinetic Substrate Structure in Circadian RNA-seq — Pipeline, Data, and Pre-Registration Archive Registered Record OSF Pre-registration DOI 10.17605/OSF.IO/B4R3W Pipeline v1 DOI 10.5281/zenodo.19648448 Journal Article IV(a) DOI 10.5281/zenodo.19648470 Pipeline SHA-256 8026f8d074e62e3305dc41ce1b863e6a6add493495641d262fb9d859ad0005cb Registration date 19 April 2026 Contributor Pretorius, Eugene Citations Pre-registration: Pretorius, E. (2026, April). Discrete Substrate Signatures in the Arabidopsis Circadian Transcriptome: A Triadic Derivation and Empirical Non-Falsification Program for Kinetic Substrate Structure in Circadian RNA-seq. Open Science Framework. [10.17605/OSF.IO/B4R3W] Pipeline: Pretorius, E. (2026). GRADIENT-3 Pipeline v1 OFFICIAL [Software]. Zenodo. [10.5281/zenodo.19648448] SHA-256: 8026f8d074e62e3305dc41ce1b863e6a6add493495641d262fb9d859ad0005cb Journal Article IV(a): Pretorius, E. (2026). Gradient Physics: An Ontological-First Derivative Science — Journal Article IV(a) — Discrete Substrate Signatures in the Arabidopsis Circadian Transcriptome: Empirical Instantiation of the Triadic Kernel in the SRP222555 RNA-seq Dataset [Preprint]. Zenodo. [10.5281/zenodo.19648470] Title Discrete Substrate Signatures in the Arabidopsis Circadian Transcriptome: A Triadic Derivation and Empirical Non-Falsification Program for Kinetic Substrate Structure in Circadian RNA-seq — Pipeline, Data, and Pre-Registration Archive Description Overview GRADIENT-3 is the third physical instantiation of Gradient Mechanics, an ontological-first derivative science framework governed by three primitives: Drive (D), Limit (L), and Registration (R). This repository contains the hardlocked computational pipeline (v1 OFFICIAL), the three-regime Gate B audit results across 20,683 Arabidopsis thaliana genes from the circadian RNA-seq dataset SRP222555 (PRJNA566413; GEO: GSE137732; Xin et al. 2019 Plant Cell), and the complete pre-registration metadata to be registered on OSF. The ontological derivation chain is published in full in Journal Article IV(a): Primordial Axiom of Relationality → Triad → {E=4/5, C=7/10, F=3/5} → TI=42/125 → β=13/40 → Φ=0.002 → Δ=0.7016 → r=√(14/9)=1.2472 → H1–H5 + GC-1–GC-5 Every step is derivationally necessitated by the step before it. The corpus precedes the experiment. The experiment tests the corpus. The three primitive values are the unique Diophantine solution to four simultaneous internal constraints on the discrete informational lattice (δ = 1/10). They are not calibrated to data. The universality class exponent β = 13/40 = 0.325 is derived from the Triad's own structural character (d=3, n=1) under the renormalisation group fixed point, with the off-lattice value 1/3 compelled by the discrete lattice to snap to β = 13/40 — the unique lattice-consistent value producing a viable non-dissipative Kinetostatic Margin Φ = 0.002. The three Triadic primitives map exactly to the GRADIENT-3 empirical protocol: Systematization E = 4/5: Drive (D). Forcing magnitude |⟨z_LD⟩| of each gene — the mean absolute departure of a gene's LD expression from its zero-gradient CL baseline across the full 16-ZT time course. Genes are assigned to three equal forcing-magnitude tertiles before any test statistic is computed. Constraint C = 7/10: Limit (L). The zero-gradient CL reference. μ_CL(g) and σ_CL(g) — the mean and standard deviation of gene g in continuous-light samples — constitute the biological Constraint condition. They define the Heaviside threshold against which LD forcing residuals are measured. Registration Frequency F = 3/5: Registration (R). Discrete ZT expression events per gene per chronon in the circadian time course. The derived signal z(g,ZT) = [log₂(TPM_LD(g,ZT)+1) − μ_CL(g)] / σ_CL(g) is the primary Registration observable — the circadian forcing residual in units of the zero-gradient noise floor. The Derived Biological Consequences IMPORTANT: This preregistration does not disclose contingent probe parameters. All predictions below are derived from {E=4/5, C=7/10, F=3/5, δ=1/10} by exact rational arithmetic before any expression data was examined. They are necessitated by the Triadic structure, not chosen to fit the data. Where the OSF preregistration format invites disclosure of calibration constants or contingency parameters, this submission distinguishes: (a) Hardlock-derived constants [H] — immutable, zero free parameters; (b) pipeline engineering choices [C] — standard methodological procedures pre-declared before data contact; (c) the derived probe logic — which is not contingent. Derived from the Hardlock constants before the SRP222555 dataset was examined: GC-5 — Wegner Power Law: kw ~ |⟨z_LD⟩|^(11/10). Predicted log-log slope = 11/10 = 1.100 [H] (exact rational, derivable before data contact). κ_exp = 5 × α_Wegner = 5 × 11/50 = 11/10. No framework in circadian biology predicts how gene kurtosis scales with forcing magnitude. GC-3 — N_sat = 25 Substrate Clustering: k-means on the 20,683-gene × 16-ZT z_LD matrix yields elbow at k = N_sat = 25 ± 2. N_sat = 1/φ = 1/(σ×δ) = 25 [H]. Exact rational, zero free parameters. GC-2 — Snap Phase Clustering: Alpine gene snap events (|z_LD(g,ZT)| > r = 1.2472 [H]) cluster at ZT15–ZT21 (dark-phase onset and mid-dark phase in LD 16:8). Phase preference driven by clock output cascade timing. 8 pre-registered ZT phases, Bonferroni correction. GC-4 — DC Likelihood on Harmonic Residuals: Δℓ > 0 on residuals after subtracting best-fit harmonic regression (24h + 12h fixed periods). Harmonic periods are fixed by circadian biology — not free parameters. The DC formula uses Λ, Δ, r [all H], zero free parameters. GC-1 — GO-Term Independent Regime Validation: H1–H5 ordering holds when gene regime is assigned from Gene Ontology annotation (GO:0007623, circadian rhythm) rather than |⟨z_LD⟩|. This provides a fully independent regime axis and confirms structural independence of the probe from the expression signal. Preliminary Biological Confirmation The following results are audit-locked Gate B values (16 April 2026). They are not preliminary; they are the pre-registered test outcomes for the Gate B hypotheses. Gate C items remain blinded pending OSF timestamp. H3 Snap Fraction — r = √(14/9) [H] (Gate B confirmed): Predicted threshold r [H] 1.247219 = √(14/9); derived from σ_ratio² = 1 + ε_snap²/Φ = 14/9 CL null rate (empirical) 21.05% (theoretical N(0,1): 21.23%) — calibration CONFIRMED Alpine snap rate 77.55% | excess 3.68× null Flat snap rate 44.14% | excess 2.10× null CoLocated snap rate 18.43% | excess 0.88× null (BELOW null — actively buffered) Gradient order Alpine > Flat > CoLocated CONFIRMED Free parameters ZERO — r [H] derived before data contact H4 DC Likelihood Ratio — Λ, Δ, r [H] (Gate B confirmed): Λ (Lambda) [H] 0.298447 = 1 − TI^β Δ (Delta) [H] 0.701553 = TI^β r [H] 1.247219 = √(14/9) DC exp coefficient [H] 5/28 = (1−1/r²)/2 exact rational Alpine DC-win rate 100.0% (6,894 / 6,894 genes) Flat DC-win rate 99.4% (6,857 / 6,895 genes) CoLocated DC-win rate 20.4% (1,403 / 6,894 genes) — null-consistent Free parameters ZERO — all constants [H] derived before data contact These results confirm that the Dirac Comb substrate structure is present in the forcing residuals of the SRP222555 circadian RNA-seq dataset, that the snap fraction in Alpine genes exceeds the Hardlock-derived threshold r by 3.68× the zero-gradient null rate, and that the Dirac Comb model with fixed Hardlock constants separates the gene regime classes with complete discriminability. None of these predictions were adjusted after data contact. Hypotheses (H1-H5 Foundational Lock + Gate C Biological Predictions) H1: Per-gene kurtosis of z_LD(g) ordered monotonically by Drive regime (CoLocated < Flat < Alpine). Derived from Heaviside threshold on L producing non-Gaussian temporal forcing profiles in strongly forced genes. Test: Jonckheere-Terpstra, α = 0.001, one-tailed. H2: Super-white variance scaling (α > −1.000) on the gene forcing-magnitude axis. Derived from non-zero gene-gene covariance forced by circadian module co-regulation — the biological analog of Registration footprint autocorrelation. H3: Per-observation snap fraction using threshold r = √(14/9) = 1.2472 [H] follows Alpine > Flat > CoLocated, with empirical CL null rate ≈ 21.2% consistent with z_CL ∼ N(0,1). Zero free parameters. H4: Per-gene Dirac Comb log-likelihood ratio: DC-win fraction (sum Δℓ > 0) follows Alpine > Flat > CoLocated. Λ = 0.2984 [H], Δ = 0.7016 [H], r = 1.2472 [H]. Classical Gaussian null demolished for Alpine gene class. H5: The z-field is continuously distributed (n_unique/n_total > 0.99). N-GM grain-size bound: f_lattice > 1 ZT chronon (3h) for all genes. No quantisation at the measurement cadence. GC-5: kw ~ |⟨z_LD⟩|^(11/10). Predicted log-log slope 1.100 [H] inside 95% CI of empirical OLS slope. GC-3: k-means elbow at k = N_sat = 25 ± 2 on the z_LD gene×ZT matrix. N_sat = 25 [H]. GC-2: At least 80% of Alpine snap events cluster at ZT15–ZT21 (dark-phase onset). 8 pre-registered ZT phases, Bonferroni correction. GC-4: Δℓ > 0 on per-gene z_LD residuals after subtraction of best-fit harmonic model (24h + 12h fixed periods). Zero free parameters. GC-1: JT ordering CoLoc < Flat < Alpine holds when gene regime is assigned from GO:0007623 circadian rhythm annotation independently of |⟨z_LD⟩|. Three-Regime Endpoints Alpine (High Drive, top forcing tertile): 6,894 genes. Forcing magnitude |⟨z_LD⟩| ≥ 1.5352. N observations = 110,304 (6,894 genes × 16 ZT). Snap events = 85,537 (77.55%, excess 3.65× Gaussian null). Constitutive Non-Responsive genes = 0. High Drive — maximum Dirac Comb accumulation. Flat (Low Drive, mid forcing tertile): 6,895 genes. Forcing magnitude 0.9777 ≤ |⟨z_LD⟩| < 1.5352. N observations = 110,320. Snap events = 48,694 (44.14%, excess 2.08×). CNR genes = 2. Intermediate Drive. CoLocated (Zero Drive, bottom forcing tertile): 6,894 genes. Forcing magnitude |⟨z_LD⟩| < 0.9777. N observations = 110,304. Snap events = 20,326 (18.43%, excess 0.87× — below null, actively buffered). CNR genes = 3. Zero-Drive control — reference process. Total archive: 27,655 genes (kallisto quantification). Retained after CPM filter: 20,693. Retained after σ_CL filter: 20,683. Total LD observations: 330,928 (20,683 genes × 16 ZT). Total CL observations: 330,928. Constitutive Non-Responsive class: 5 genes total (CoLoc=3, Flat=2, Alpine=0). Frozen snap vectors: 330,928 per-observation binary flags at pipeline SHA-256. v1 OFFICIAL Pipeline Three-Regime Gate B Results All pre-registered Gate B tests confirmed. The Triadic model is not falsified by any metric in the SRP222555 dataset. H1: Per-Gene Kurtosis Drive Ordering Alpine mean kw: −0.535 | Flat mean kw: −0.532 | CoLoc mean kw: −0.431 JT: J=75,525,173, Z=9.1213, p<10⁻⁸ | Ordering CoLoc < Flat < Alpine: CONFIRMED | Alpine normality p=0.0000: NON-GAUSSIAN | Verdict: CONFIRMED CNR class: 5 genes (CoLoc=3, Flat=2, Alpine=0) excluded from JT — constitutive non-responsive, zero temporal variance in z_LD. Counted as separate derived biological class. Note on kurtosis direction: All three regimes show negative (platykurtic) mean kw, in contrast to the leptokurtic kw in GRADIENT-1 (full excess κ=2.17 Alpine clocks) and GRADIENT-2 (Alpine mean kw=86.49 pulsars). This is a derived consequence of deterministic sustained circadian forcing (Alpine genes displaced at large |z| at every ZT) versus stochastic rare-event snap accumulation in physical substrates. Non-Gaussianity is confirmed in both cases; the direction of departure encodes the nature of the gradient. H2: Super-White Variance Scaling α = −0.4863 (z_stat=15.84, p<0.000001) on gene forcing-magnitude axis | n_genes=20,683 | n_scales=10 | Verdict: CONFIRMED H3: Per-Observation Snap Fraction | r = √(14/9) = 1.2472 [H] Empirical CL null rate: 21.05% (theoretical N(0,1): 21.23%) — calibration CONFIRMED Alpine 77.55% (excess 3.68×) | Flat 44.14% (excess 2.10×) | CoLocated 18.43% (excess 0.88×, below null) Full gradient order confirmed | Verdict: CONFIRMED H4: Per-Gene Dirac Comb Likelihood Ratio Λ=0.298447 [H] | Δ=0.701553 [H] | r=1.247219 [H] | DC exp coeff=5/28 [H] Alpine 100.0% (6,894/6,894) | Flat 99.4% (6,857/6,895) | CoLocated 20.4% (1,403/6,894) Full gradient order Alpine > Flat > CoLocated CONFIRMED | Zero free parameters | Verdict: CONFIRMED H5: N-GM Grain-Size Bound n_unique=656,921 | n_total=661,856 | frac_unique=0.9925 | continuous=True | f_lattice > 1 ZT chronon | Verdict: CONFIRMED GC-5: Wegner Power Law kw ~ |⟨z_LD⟩|^(11/10) Pending. Requires: gradient3_stage3_h1_kurtosis.csv (available). Predicted slope [H]: 1.100 = 11/10 (exact rational). Gate C. GC-3: N_sat = 25 Substrate Clustering Pending. Requires: gradient3_stage2_z_LD.csv (available). k-means elbow at k=25±2. N_sat=25 [H]. Gate C. GC-2: Snap Phase Clustering by ZT Pending. Snap vectors frozen at pipeline SHA-256. 85,537 Alpine snap events. 8 pre-registered ZT phases (ZT0,3,6,9,12,15,18,21), Bonferroni correction. Gate C. GC-4: DC Likelihood Ratio on Harmonic Residuals Pending. Requires harmonic model fit (24h + 12h fixed periods). Δℓ > 0 prediction uses Λ, Δ, r [all H]. Zero free parameters. Gate C. GC-1: GO-Term Independent Regime Validation Pending. Requires GO:0007623 annotation file for TAIR10 gene identifiers. JT ordering prediction unchanged. Gate C. Summary H1 Per-gene kurtosis ordering: J=75,525,173, Z=9.1213, p<10⁻⁸. CONFIRMED. H2 Super-white variance: α=−0.4863, z_stat=15.84. CONFIRMED. H3 Snap fraction r [H]: Alpine 77.55%, excess 3.68× null, order confirmed. CONFIRMED. H4 DC likelihood ratio: Alpine 100.0% > Flat 99.4% > CoLoc 20.4%, order confirmed. CONFIRMED. H5 N-GM grain-size bound: frac_unique=0.9925, continuous=True. CONFIRMED. GC-5 Wegner power law: Predicted slope 1.100 [H]. GATE C. GC-3 N_sat=25 clustering: k=25±2 predicted [H]. GATE C. GC-2 Snap phase clustering: ZT15–21 predicted [H]. GATE C. GC-4 DC on harmonic residuals: Δℓ>0 predicted [Λ,Δ,r H]. GATE C. GC-1 GO-term regime validation: JT ordering under independent regime. GATE C. Snap Detection Alpine: 85,537 snap observations across 6,894 genes (77.55% snap rate, 3.65× Gaussian null rate). Maximum snap density concentrated in strongly clock-regulated genes with |⟨z_LD⟩| >> 1.5352 (deep Alpine tertile). All 110,304 Alpine per-observation snap flags frozen at pipeline SHA-256. Flat: 48,694 snap observations across 6,895 genes (44.14% snap rate, 2.08× Gaussian null rate). Intermediate forcing. All 110,320 Flat per-observation snap flags frozen at pipeline SHA-256. CoLocated: 20,326 snap observations across 6,894 genes (18.43% snap rate, 0.87× Gaussian null — below the null rate). Constitutive genes are actively buffered against variation, confirming the zero-Drive reference behaves as a quieter-than-random reference process. This mirrors the G1 co-located baseline result (1.1× ≈ Gaussian). All 110,304 CoLocated per-observation snap flags frozen. Total: 154,557 snap observations across 20,683 genes across all three regimes. 330,928 total observations. Frozen snap vectors (per-observation binary: gene × ZT, snap=1 if |z_LD|>r [H]) are stored in gradient3_stage3_h4_dc_likelihood.csv and gradient3_snaps_by_regime.csv. These constitute the Gate C GC-2 input. Threshold r = 1.2472 [H] — not a calibration parameter. The CoLocated snap rate of 18.43% (below the theoretical Gaussian null of 21.23%) confirms the zero-Drive gene class is actively buffered against expression variation — not merely near-Gaussian. This is the strongest confirmation that the CoLocated regime is a genuine zero-Drive reference process, not just the weakest Drive regime. Gate System Gate A — COMPLETE (pre-dates all data contact): Derivation of all Hardlock constants (Essays I through XIII of Gradient Cybernetics). Biological predictions GC-2 (snap phase clustering), GC-3 (N_sat=25), GC-5 (Wegner slope 11/10), and the H3 snap threshold r=√(14/9) specified before any data access. Derivational chain from Primordial Axiom to empirical consequences fully documented in Journal Article IV(a). All constants certified immutable in gradient3_gate_a_hardlock_certification.csv. Gate B — COMPLETE (v1 OFFICIAL pipeline, April 2026): Statistical audit. All H1–H5 CONFIRMED. All values locked at SHA-256: 8026f8d074e62e3305dc41ce1b863e6a6add493495641d262fb9d859ad0005cb. Gate C — PROSPECTIVE (commences after OSF timestamp): GC-5: Wegner power law kw~|⟨z_LD⟩|^(11/10). Requires: gradient3_stage3_h1_kurtosis.csv (available). GC-3: N_sat=25 substrate clustering. Requires: gradient3_stage2_z_LD.csv (available). GC-2: Snap phase clustering at ZT15–21. Input: 330,928 frozen per-observation snap flags. 8 pre-registered ZT phases, Bonferroni correction. GC-4: DC likelihood on harmonic residuals. Requires harmonic model fit (24h+12h fixed periods). GC-1: GO-term independent regime validation. Requires GO:0007623 annotation for TAIR10. Pipeline Scope and Scalar Invariance GRADIENT3_pipeline_v1.py is not an instrument built for RNA-seq data. It is the detection instrument for the necessary consequences of Drive (D), Limit (L), and Registration (R). The SRP222555 Arabidopsis thaliana circadian RNA-seq dataset is the third physical domain in which these consequences are tested, following the GRADIENT-1 confirmation in the ROCIT 2022 optical clock network and the GRADIENT-2 confirmation in the NANOGrav 15-Year Pulsar Timing Array. The mathematical core — the Jonckheere-Terpstra trend test, the super-white variance scaling test, the snap fraction test against r [H], the DC likelihood ratio, the N-GM grain-size bound, the snap vector generation — contains no domain-specific assumptions. These functions operate on any ordered observable with an associated Drive magnitude. They were derived from the structure of D, L, R, not fitted to circadian expression data. Scalar invariance is not a property that needs to be engineered into the pipeline. It is a property inherited from the derivation. The same derivational chain that produced JT Z = 14.84 in optical clock networks (GRADIENT-1) and JT Z = 4.911 in millisecond pulsars (GRADIENT-2) is tested in this repository against 20,683 Arabidopsis genes. Drive in GRADIENT-1 was geopotential height differential Δh. Drive in GRADIENT-2 was observation baseline T_i. Drive in GRADIENT-3 is forcing magnitude |⟨z_LD⟩|. All three instantiate the same triadic primitive D from the Kernel {E, C, F}. The detection logic is unchanged. SHA-256: 8026f8d074e62e3305dc41ce1b863e6a6add493495641d262fb9d859ad0005cb This hash is the hardlock across all instantiations. Any execution of this pipeline on any domain that produces a result does so under the same locked code that produced the SRP222555 Gate B results. Reproducibility Protocol np.random.seed(42) is applied for the sub-sample in H1 normality testing (up to 5,000 gene kw values) to avoid over-sensitivity of the normality test on large samples. All other statistical tests — the JT statistic, the OLS variance scaling slope, the snap fraction, the DC log-likelihood sum, the N-GM unique count — are pure deterministic functions of the input data. Input data: SRP222555 FASTQ files (32 paired-end RNA-seq runs). BioProject PRJNA566413. Available from NCBI SRA: https://trace.ncbi.nlm.nih.gov/Traces/study/?acc=SRP222555 To reproduce: 1. Download GRADIENT3_pipeline_v1.py [10.5281/zenodo.19648448]. Verify SHA-256: 8026f8d074e62e3305dc41ce1b863e6a6add493495641d262fb9d859ad0005cb 2. Download SRP222555 FASTQ files from NCBI SRA (32 paired-end runs, ~75.5 GB). Place at the path specified in FASTQ_DIR in the pipeline header. 3. Install dependencies: numpy, pandas, scipy, astropy, kallisto v0.50.1 4. Clear the GRADIENT3_checkpoints/ directory for a clean official run. 5. Execute: python GRADIENT3_pipeline_v1.py The pipeline self-hashes at execution and prints the SHA-256 to confirm pipeline integrity. The hash printed at execution must match 8026f8d074e62e3305dc41ce1b863e6a6add493495641d262fb9d859ad0005cb. Output: GRADIENT3_output/ directory containing 13 CSV data provenance files (gradient3_gate_a_hardlock_certification.csv through gradient3_gate_b_audit.csv), the pipeline SHA-256 lock file, the Gate B audit report, and the Gate C quarantine register. Contingency Register PREREGISTRATION NOTE ON CONTINGENCIES: The OSF preregistration format assumes the registrant is disclosing contingent probe parameters — calibration constants, analysis thresholds, or modelling choices that could have been adjusted to fit the data. This preregistration does not contain such contingencies in the probe logic. The Hardlock constants (Λ, Δ, r, N_sat, κ_exp) are derived from {E, C, F, δ} by exact rational arithmetic and are immutable. The following register documents the only points where the pipeline meets contingent data properties — these are all engineering choices [C], not probe parameter disclosures [H]. None of the following introduces a free parameter into the probe. C-1 — CPM expression filter: Raw TPM values span six orders of magnitude. Genes expressed below 1.0 CPM in fewer than 4 samples cannot contribute reliable CL noise floor estimates (σ_CL). Resolved: CPM filter (>1.0 CPM in ≥4 samples) is standard RNA-seq practice [C, pre-declared before data contact]. Retained 20,693 / 27,655 genes. No free parameters — the threshold 1.0 CPM and minimum sample count 4 are standard methodological minimums, not optimised to outcome. C-2 — σ_CL floor filter: σ_CL < 0.01 produces near-zero denominators in z(g,ZT) = [log₂TPM_LD − μ_CL] / σ_CL. These genes have no CL temporal variance and cannot carry a meaningful forcing signal. Resolved: σ_CL > 0.01 filter [C, pre-declared]. Retained 20,683 / 20,693 genes. No free parameters — σ_CL = 0.01 is a numerical floor, not a biological threshold. C-3 — Constitutive Non-Responsive class: 5 genes (CoLoc=3, Flat=2, Alpine=0) have temporal variance of z_LD below machine epsilon (< 10⁻⁸). Kurtosis is undefined (0/0 division) for these genes. Resolved: CNR genes are excluded from the JT kurtosis test and counted as a separate derived biological class. The exclusion threshold 10⁻⁸ is a machine-precision floor [C], not an analytical choice. Zero free parameters — CNR genes are classified, not removed. Their count by regime is reported in the audit. C-4 — Paired-end kallisto quantification: SRP222555 is paired-end RNA-seq. kallisto pseudo-alignment against TAIR10 cDNA (Ensembl Plants release 57) is standard and produces transcript-level TPM aggregated to gene level. No alternative quantification was evaluated. [C, pre-declared] No free parameters. C-5 — ZT time point assignment: The ZT time point sequence (ZT0,0,3,3,6,6,9,9,12,12,15,15,18,18,21,21) follows the published study design from SRP222555 metadata. Assignment is derived from the experimental record, not from expression data. Gene regime assignment uses forcing magnitude |⟨z_LD⟩| — an expression-derived quantity — which introduces a partial first-moment circularity with respect to the test statistics (fourth moment, binary threshold, log-likelihood). This is documented and queued for Gate C resolution via GO-term independent regime assignment (GC-1). No free parameters. Foreknowledge Summary Directional predictions — per-gene kurtosis ordering CoLocated < Flat < Alpine, super-white variance on the gene forcing axis, snap fraction excess at r = √(14/9) [H], DC-win rate ordering, N-GM continuity — were all derived from {E, C, F, δ} before any gene expression data was examined. What changed between pipeline versions (v1 through v5 of the development series, condensed into the single official v1 release) were execution mechanics (signal derivation, gene regime assignment, H1 CNR class), not predicted directions or predicted values. Version (development) What changed Implication v1 (dev): H1-H5 with std-whitened signal std-whitened per gene destroyed kurtosis signal before probe ran (kw~−0.4 all regimes). Signal construction error, not probe change. H1/H2 FAIL. Corrected in v2. v2 (dev): Mean-centered signal, gene-rank H2 H1: cross-sectional kurtosis genuinely inverted (CL more leptokurtic than LD — biological property). H2: sorted gene-rank axis produced alpha~0 artifact. Correct signal; wrong aggregation axis. Corrected in v3. v3 (dev): Fold-change kurtosis, random subsampling H2 H1: FC kurtosis between consecutive ZT. Unicode log file crash. Encoding error; H1 still wrong axis. Corrected in v4. v4 (dev): Derived z-signal; sample-level ZT regime Correct signal z(g,ZT)=[log₂TPM_LD−μ_CL]/σ_CL. But regime split on SAMPLE axis (ZT time points) — no within-LD triadic gradient at sample level. H1/H3 FAIL (trivially saturated). Regime axis corrected in v5. v5 (dev): Gene-level regime from |⟨z_LD⟩|; H1 CNR class Correct unit of analysis (gene, not sample). CNR genes identified and separated. Gate C quarantine formalised. 5/5 PASS. Condensed into v1 OFFICIAL. v1 OFFICIAL: Single locked release All development work condensed. Gate A/B/C structure formalised. SHA-256 locked. No developmental content. This is the registered pipeline. The SRP222555 FASTQ files required for Gate C items (GC-1: GO annotation; GC-2: snap phase ZT clustering; GC-3: k-means on z_LD; GC-4: harmonic residuals; GC-5: Wegner regression) have been accessed in full to produce Gate B quantification results. However, the Gate C test statistics have not been computed. The per-observation snap vectors and the z_LD matrix are frozen at the pipeline SHA-256 and constitute the Gate C input for GC-2 and GC-3 respectively. Files in This Repository GRADIENT3_pipeline_v1.py: Hardlocked official pipeline. [10.5281/zenodo.19648448] SHA-256: 8026f8d074e62e3305dc41ce1b863e6a6add493495641d262fb9d859ad0005cb GRADIENT3_OSF_Preregistration_COMPLETE.docx: Registered OSF pre-registration. [10.17605/OSF.IO/B4R3W] Journal_Article_IV_a_GRADIENT_BIOLOGY.docx: Journal Article IV(a) preprint. [10.5281/zenodo.19648470] GRADIENT3_v1_GateB_Audit_Report.docx: Full three-regime Gate B audit report. SHA-256 lock confirmed. All test results, contingency register, data provenance. gradient3_dataset_FINAL.csv: Canonical GRADIENT-3 dataset. 32 rows × 9 columns. Per-sample DC log-likelihood and snap statistics. gradient3_gate_a_hardlock_certification.csv: Gate A certification. 16 Hardlock constants with exact rational, float value, derivation, gate_a_certified=True, mutable=False. gradient3_stage0_sample_manifest.csv: Stage 0. 32 samples: run, condition, zt_hours, rep. Locked before expression data. gradient3_stage1_tpm_matrix.csv: Stage 1. 27,655 genes × 32 samples. TPM values from kallisto pseudo-alignment. gradient3_stage2_signal_provenance.csv: Stage 2. Per-sample signal statistics: mean_z, std_z, μ_CL mean, σ_CL mean. gradient3_stage2_z_LD.csv: Stage 2. 20,683 genes × 16 LD samples. The derived forcing signal z(g,ZT). LOCKED before tests. gradient3_stage2_z_CL.csv: Stage 2. 20,683 genes × 16 CL samples. The null reference z_CL(g,t). LOCKED before tests. gradient3_stage2_gene_regimes.csv: Stage 2. Per-gene regime assignment (gene, forcing_mag, tertile boundaries, regime, σ_CL, μ_CL). LOCKED before H1–H5 tests. gradient3_stage3_h1_kurtosis.csv: Stage 3. Per-gene kw, regime, forcing_mag, n_zt, constitutive_nonresponsive flag. gradient3_stage3_h2_variance_scaling.csv: Stage 3. α, z_stat, p, super_white, n_genes, n_scales. gradient3_stage3_h3_snap_fraction.csv: Stage 3. r_hardlock, null_rate_CL_empirical, null_rate_CL_theory, snap_rate per regime, gradient_order_full. gradient3_stage3_h4_dc_likelihood.csv: Stage 3. Per-gene: regime, n_zt, delta_logL_total, n_pos_zt, frac_pos_zt, dc_wins. gradient3_stage3_h5_ngm_bound.csv: Stage 3. n_unique, n_total, frac_unique, max_bin_excess, continuous, bound. gradient3_snaps_by_regime.csv: Diagnostic. Per-regime: n_obs, n_snaps, rate_pct, gauss_pct, excess. Frozen for Gate C GC-2. gradient3_gate_b_audit.csv: Machine-readable Gate B audit record. All FC1–FC5 verdicts, key statistics, Gate C status. gradient3_gate_b_audit_report.txt: Full text audit report. gradient3_gate_c_quarantine.csv: Gate C quarantine register. 5 items: GC-1 through GC-5. Status=QUARANTINED. Related Identifiers OSF Pre-registration: [10.17605/OSF.IO/B4R3W] Pipeline v1 Software: [10.5281/zenodo.19648448] Journal Article IV(a) Preprint: [10.5281/zenodo.19648470] SRP222555 Input Data (NCBI SRA): https://trace.ncbi.nlm.nih.gov/Traces/study/?acc=SRP222555 GRADIENT-2 Pre-registration: https://doi.org/10.17605/OSF.IO/Z5NDM GRADIENT-2 Pipeline v3: https://doi.org/10.5281/zenodo.19165903 GRADIENT-2 Journal Article II: https://doi.org/10.5281/zenodo.19165633 NANOGrav 15-Year Input Data: https://doi.org/10.5281/zenodo.16051178 GRADIENT-1 Pre-registration: https://doi.org/10.17605/OSF.IO/39SDP GRADIENT-1 Pipeline v5: https://doi.org/10.5281/zenodo.18749096 GRADIENT-1 Journal Article I: https://doi.org/10.5281/zenodo.18750277 ROCIT 2022 Input Data: https://doi.org/10.5281/zenodo.17107693 Corpus Paper I: https://doi.org/10.5281/zenodo.18457320 Corpus Paper II: https://doi.org/10.5281/zenodo.18474271 Corpus Paper III: https://doi.org/10.5281/zenodo.18487518 Corpus Paper IV: https://doi.org/10.5281/zenodo.18495883 Corpus Paper V: https://doi.org/10.5281/zenodo.18510075 Corpus Paper VI: https://doi.org/10.5281/zenodo.18518126 Corpus Paper VII: https://doi.org/10.5281/zenodo.18525123 Corpus Paper VIII: https://doi.org/10.5281/zenodo.18561473 Corpus Paper IX: https://doi.org/10.5281/zenodo.18598479 Corpus Paper X: https://doi.org/10.5281/zenodo.18609937 Corpus Paper XI: https://doi.org/10.5281/zenodo.18624178 Corpus Paper XII: https://doi.org/10.5281/zenodo.18632185 Corpus Paper XIII: https://doi.org/10.5281/zenodo.18643513 Corpus Paper XIV: https://doi.org/10.5281/zenodo.18649194 Keywords Arabidopsis thaliana · Circadian Clock · Circadian Transcriptome · Dirac Comb · Drive-Limit-Registration · Forced Gene Expression · Gradient Mechanics · Jonckheere-Terpstra · Kinetostatic Margin · Kinetic Substrate · Kurtosis · N-GM Bound · Non-Gaussianity · Ontological-First Derivative Science · RNA-seq · Scalar Invariance · Snap Fraction · SRP222555 · Triadic Derivation · Zeitgeber Time License CC BY 4.0 Contributor Pretorius, Eugene GRADIENT-3 v1 OFFICIAL · Registered April 2026 Pipeline SHA-256: 8026f8d074e62e3305dc41ce1b863e6a6add493495641d262fb9d859ad0005cb OSF: [10.17605/OSF.IO/B4R3W] | Pipeline: [10.5281/zenodo.19648448] | Article: [10.5281/zenodo.19648470] Input Data: https://trace.ncbi.nlm.nih.gov/Traces/study/?acc=SRP222555 | PRJNA566413 | GEO: GSE137732
Identifiants
- DOI
- 10.5281/zenodo.19648447
Repères documentaires
- Famille documentaire
- unknown
- Types déclarés par les sources
- Software
- Éditeurs
- Zenodo
Les mentions d’entités ne constituent pas une preuve de reconnaissance juridique. La juridiction étudiée est distincte du pays d’affiliation des auteurs.
Sources
DataCite · Document 1
Citer cette référence
Pretorius, Eugene. GRADIENT-3 Pipeline v1. 2026. https://doi.org/10.5281/zenodo.19648447
JSON · CSL-JSON · BibTeX · RIS · CSV · JSON-LD ·
APA, Chicago, ISO 690 : formats bibliographiques fondés sur les champs disponibles. OSCOLA / Bluebook : données structurées uniquement ; aucune conformité juridique complète n’est revendiquée sans vérification des éléments nécessaires.
- APA
- Pretorius, Eugene. (2026). GRADIENT-3 Pipeline v1. https://doi.org/10.5281/zenodo.19648447
- Chicago
- Pretorius, Eugene. GRADIENT-3 Pipeline v1. 2026. https://doi.org/10.5281/zenodo.19648447
- ISO 690
- Pretorius, Eugene. GRADIENT-3 Pipeline v1. 2026. https://doi.org/10.5281/zenodo.19648447
Warning: Undefined array key "last_link_verified" in /home/clients/e27d81e12f04d2bfd747aa264540ac56/sites/_UDTR_APP/legal_library_reference.php on line 31
UDTR — Bibliothèque juridique mondiale des droits de la Nature. GRADIENT-3 Pipeline v1. Notice documentaire: https://www.declarationuniverselledesdroitsdelarbre.org/global-rights-of-nature-radar/reference/r-787ce7a2b3fd4c9e70a4799975061d80/ · Dernière vérification HTTP : non effectuée
Données de cette version figée
Dossiers et relations documentaires
Rôles des sources
Source bibliographique / publication d’origine · https://zenodo.org/doi/10.5281/zenodo.19648447
Un lien vers une autorité ne prouve pas à lui seul une affirmation juridique. Les résumés et les textes intégraux restent soumis aux conditions de leur source.
Statut documentaire
Pertinence à examiner
Bibliothèque
Cette identité n’est pas encore intégrée à la Bibliothèque juridique.