Time 0 s Fork 0 bp Okazaki fragments 0 ssDNA gap 0 bp Active lagging pols 0
Hero Fork Lagging Top
Parental strands blue ; new DNA green ; RNA primers orange . Helicase ring unwinds ahead; the leading strand is copied continuously, the lagging strand in Okazaki fragments (each primed near the fork and extended away from it). Bases A gold, T pink, G violet, C cyan. Drift tokens are decorative.
WebGL is not available here — the models and graphs on the right still work.
Replication Fork, Okazaki Fragments, Origin Firing and Meselson–Stahl
Three linked views of DNA copying: the molecular fork (3D), the whole genome with many origins, and the 1958 experiment that proved replication is semiconservative.
1. Fork 2. Genome 3. Meselson–Stahl
Presets
Pause Restart Quality: High
Readouts
Lagging-strand gap (single-stranded template) over time
Fork checks
Display units: seconds and base pairs are scaled for watching (real bacterial forks run about 1000 bp/s with ~1000–2000 nt fragments; eukaryotic about 30–50 bp/s with ~100–200 nt).
Organism presets (typical orders of magnitude)
Even, simultaneous Random, simultaneous Stochastic firing
New random trial
Space–time diagram of one trial
Fraction replicated versus time
Readouts
Genome checks
Bacteria were grown for generations on heavy nitrogen (15 N), then switched to light (14 N). DNA from each generation is spun in a CsCl gradient: heavier DNA settles lower. Which pattern appears depends on how DNA is copied.
Semiconservative Conservative Dispersive
Generations after the switch
Readouts
Experiment: which model survives?
Pick the model that nature uses (the rows above show what each predicts), then compare what you see at generations 1 and 2.
Meselson–Stahl checks
Educational models, not a lab protocol. Genome numbers are typical values, rounded; fork animation is schematic (base-pair count per Okazaki fragment shortened).
Controls Drag / one finger: orbit; wheel / pinch: zoom; Hero / Fork / Lagging / Top jump to camera views. Space pauses while the 3D view is focused. Tabs switch the panel; the 3D fork keeps running in the background. The science
The fork. A helicase opens the double helix at speed v (bp/s). DNA polymerase can only add nucleotides to a 3′ end, so the leading strand is copied continuously behind the fork while the lagging strand is copied discontinuously: primase lays a short RNA primer (about 10 nt) every F bases, polymerase extends it away from the fork at speed vp , and later primers are replaced and the nicks sealed by ligase. These pieces are the Okazaki fragments (about 1000–2000 nt in bacteria, 100–200 in eukaryotes).
Lagging-strand bookkeeping. Fragment k is primed when its top base is exposed, at tk = (k+1)F/v + tprime and finished at tk + F/vp . The single-stranded template between the fork and the latest primer has length g(t) that grows at v and drops by F at each priming: its time average is ⟨g⟩ = F/2 + v tprime . By Little's law the average number of fragments being extended simultaneously is λ W = (v/F)(F/vp ) = v/vp .
Genome scale. One origin with two forks needs T = L/(2v): for E. coli (4.6 Mbp, ~1000 bp/s) about 40 minutes. Eukaryotes use many origins: n equally spaced origins firing together finish in L/(2nv); random positions leave random gaps, 1 − f = e−2vnt/L ; origins firing at a constant rate I per bp per second on unreplicated DNA obey the Kolmogorov–Johnson–Mehl–Avrami law 1 − f = exp(−I v t2 ), which reproduces the S-phase profile qualitatively. Passive replication removes origins before they fire (“phantom” origins).
Meselson–Stahl (1958). Semiconservative: each daughter has one old and one new strand. After g generations a fraction 2/2g of duplexes is hybrid (¹⁵N/¹⁴N) and the rest light. Conservative keeps the parental duplex intact: 1/2g heavy and the rest light, no hybrid. Dispersive predicts one band whose density moves toward light by half each generation. Generation 1 (one hybrid band) rules out conservative; generation 2 (hybrid plus light) rules out dispersive. In every model the 15 N is only diluted, so the mean heavy fraction is 2−g . Real bands are broad (diffusion); the Gaussian width here is illustrative, band spacing is exaggerated (real Δρ of about 0.014 g/mL between ¹⁵N and ¹⁴N DNA).
Honest simplifications. Constant speeds; no replication-fork stalling, topoisomerase, proofreading, nucleosomes or origin licensing; genome simulation is 1D on a lattice of 2000 sites (dimensionless units, then scaled); organism numbers are rounded typical values to be spot-checked.
Models built with Blender (polymerase, helicase, primase, bases, tokens; baked textures) and rendered with Three.js.