Prime editing nicks one DNA strand and uses the pegRNA's 3′ extension as a template for reverse transcriptase, writing the edit directly onto a new 3′ single-strand "flap." That edited 3′ flap must displace the original, unedited 5′ flap before the cell's endogenous flap-resolution machinery (FEN1-mediated excision, followed by ligation and mismatch repair) permanently installs the edit. This handoff is the rate-limiting, stochastic step that sets real-world editing efficiency.
Flap stability score S(L,GC) = 0.55·(L/34) + 0.45·(GC/100)
Predicted efficiency P_3' = P0 · exp(-L / 16) · (1 + 0.6·GC_frac) · (PE3 ? 2.6 : 1)
Per-trial outcome 3'-flap wins with probability p = clamp(P_3', 0.01, 0.98)
Indel estimate PE3 off: ≈1.5% PE3 on: ≈4 + 3·GC_frac %
- Flap length L — longer RT extensions (bigger edits) take longer to synthesize and compete less effectively against the pre-existing 5′ flap, so predicted efficiency decays roughly exponentially with L — matching the qualitative trend reported for prime editing (Anzalone et al., Nature 2019): small edits install far more efficiently than long insertions.
- GC content — a GC-richer flap forms a more stable heteroduplex once reannealed, nudging the competition toward the edited strand.
- PE3 — nicking the complementary, non-edited strand forces the cell's mismatch-repair machinery to use the edited strand as the correction template, roughly tripling efficiency (PE3 vs PE2) at the cost of a higher indel rate from the two simultaneous nicks (Anzalone et al. 2019; Chen et al., Cell 2021, describe PE4/PE5 MMR-evasion designs built to fix exactly this trade-off).
- Run Trial — draws one stochastic outcome from the current probability and animates the losing flap dissolving while the winning flap ligates into the backbone; Auto-run fires trials continuously so the measured frequency can be compared to the predicted curve.
This is a simplified, pedagogically-tuned model of the qualitative competition — real efficiency also depends on PBS length, cell type, chromatin state and repair-pathway usage — but the shape of the trends (shorter edits win more, PE3 helps at an indel cost) matches the published prime-editing literature.