A Peak Is Not a Mechanism

1. Two claims that require separate evidence

A genome may complete on time without reproducing an observed initiation profile. Conversely, a model may produce a visually convincing peak while failing to complete. The present analysis therefore evaluates initiation-rate shape and full completion in the same parameter condition. It preserves the original genome-scale question while asking which components of the model explain its behavior. [1]

Arbona and colleagues established an important prior modeling context: localized candidate origins, competition for limiting factors, and recycling can produce a rise and fall in initiation rate under suitable conditions. That result motivates a mechanism test here; it does not establish the parameter values or completion guarantee of a different human-scale model. [2]

2. Define the observable before comparing curves

I(t) denotes new initiation events per unit time per unit length of unreplicated DNA. For each ten-minute bin, the calculation divides the sum of initiations across layouts by the sum of integrated unreplicated DNA length over that bin. Event-to-event geometry permits exact integration of this denominator. Using total genome length instead would measure a different quantity.

The saved literature audit obtained 57 Xenopus and 24 budding-yeast numerical points from Arbona et al.'s public source tables. It converted rates from per kilobase to per megabase without plot digitization, smoothing, time shifting, or amplitude fitting. These are published experimental summaries: the frog data derive from DNA-fiber measurements, while the yeast curve is inferred from population replication timing. They are not newly collected experiments or reanalysis of original microscopy. [3] [4]

The supplied uncertainty values were preserved. For the Xenopus table, the column label says standard deviation while the figure caption describes SEM; the audit does not resolve that discrepancy by relabeling it. Replicate identities and covariance are not supplied in these summaries. Their uncertainties therefore do not justify a calibrated goodness-of-fit test for this model.

3. Factorial computational design

The model retains 6.4 billion base pairs, 400,000 candidates, 1,500-base-pair/minute forks, and K=24,000 effective fork units. Each remaining candidate has hazard λ₀×free/K. The study crossed λ₀ values 0.001, 0.1, and 1 per minute; loading time constants 0 and 30 minutes; and return delays 0 and 480 minutes. Twelve conditions shared the same 200 layouts and candidate marks. The 480-minute delay prevents reuse before the comparison deadline, rather than disabling it forever. [1]

The primary shape descriptor uses bins with mean unreplicated exposure at least 1% of the initial genome. An interior 30-minute initiation-rate window must exceed both the first and last supported 30-minute windows by at least 20%. Joint success additionally requires at least 190/200 genomes complete by 480 minutes. This rule was fixed before the main analysis. It is a chosen descriptive criterion, not a biological law or an experimental fit statistic.

4. Successful completion and an unmet joint criterion

With λ₀=1/min, 30-minute loading, and immediate reuse, all 200 layouts completed within eight hours; median completion was 255.94 minutes, or 4.27 hours. The same condition with delayed return completed 0/200 by eight hours. At slower initiation, immediate reuse alone was insufficient. Thus reuse and initiation kinetics interact within this fixed model. [1]

The successful fast condition also had a clear raw rise and fall: I(t) was 2.944 initiations/Mb/min during 200–210 minutes and 0.1765 during 210–220 minutes. The latter bin still contained 10,905 starts across the layouts, but its mean unreplicated exposure was only 0.483%. It lay outside the predefined 1% support range. The calculation therefore did not relax the rule to claim joint success: 0/12 conditions passed both endpoints.

That negative result does not mean the raw curves lack a peak. It means the specific supported-range descriptor and completion target were not jointly satisfied. One slow-initiation, delayed-return condition passed the shape descriptor yet completed 0/200. A peak alone neither identifies recycling nor guarantees completion. Initial loading changes early dynamics, while late depletion of usable candidates can reduce initiation even as resources become available.

5. Normalized shape cannot supply an absolute clock

A later saved identifiability control reused one layout in two deterministic replays. Scaling fork speed and initiation hazard down by two, and time-dependent loading up by two, preserved all 112,973 event correspondences after time rescaling. Completion changed from 263.64 to 527.28 minutes: one replay met the eight-hour reference and the other missed it. This was one paired control, not another population experiment. [1]

The result explains why fitting only normalized time and amplitude cannot independently validate an absolute completion deadline. A matched physical clock or fork-speed measurement can break this symmetry. It does not show that parameters remain unidentifiable after such measurements are supplied, and it does not constitute a completed human calibration.

6. Interpretation and study closeout

Published human-cell perturbations support investigating component reuse: Fan et al. connected impaired disassembly with reduced soluble components and initiation. However, such evidence does not map an effective pool directly to molecule counts or assign this model's binding and return rates. The public frog and yeast tables likewise cannot validate a human capacity boundary. No parameter-matched empirical fit or independent human validation is claimed here. [5]

The saved ablation verifier recorded 372,025 passing checks; the separate statistics audit recorded 1,545. These support event accounting, aggregate calculations, uncertainty calculations, and the fixed descriptor. They do not turn synthetic parameter choices into biological measurements. No new simulations were executed for this article.

This non-peer-reviewed record publishes a compact evidence summary with source hashes, selected results, and links to the original public experimental tables. Complete local raw runs are not included. A future discriminating study would need to pair an absolute initiation time course with candidate positions, available functional factors, and a reuse perturbation in the same cellular system. Agreement across those observables would carry more mechanistic information than matching a peak alone. [1]

Update, 24 October 2026: The subsequent HCT116 public-data audit did not establish the matched S-phase clock, unreplicated-DNA exposure, or functional recovery time needed for that validation. The investigation has closed; no author inquiry or cold-storage restoration was initiated. The requirements above describe what a future validation would need, not an ongoing experiment. See the full investigation and closeout.