Right after in-vitro transcription, every mRNA molecule races through a two-step enzymatic pathway before it can be translated efficiently — while a competing exonuclease pathway degrades whatever is not yet capped. This is a continuous-time Markov chain on four states per molecule (U → uncapped, C0 → Cap0, C1 → Cap1, D → degraded):
dU/dt = -(k1 + kdeg)·U (GTase caps it, or exonuclease chews it)
dC0/dt = k1·U - k2·C0 (guanylyltransferase adds GpppN)
dC1/dt = k2·C0 (2'-O-methyltransferase → Cap1, immune-silent)
dD/dt = kdeg·U (5'→3' exonuclease/decapping loss)
Each of the N particles in the reactor is simulated as an independent trial of this chain: every frame, the probability of a transition is rate × dt, so across the whole population the fractions converge to the analytic exponential solution above. Capped RNA (Cap0 or Cap1) is protected from the exonuclease — only uncapped molecules can be degraded, which is why racing k1 ahead of kdeg is the single biggest lever on final yield.
- k1 (capping enzyme) — how fast the vaccinia-derived guanylyltransferase/triphosphatase adds the 5' GpppN cap. Raise it to outrun degradation.
- k2 (2'-O-methyltransferase) — converts Cap0 to Cap1. Cap1 is what actually evades innate-immune sensors (RIG-I/IFIT1) and drives high translation — this is the fraction quality control cares about most.
- kdeg (exonuclease) — degradation pressure on still-uncapped molecules; representative of reaction time, enzyme purity and buffer conditions.
- Temperature — scales all three rates together via a Q10 = 2 rule,
rate(T) = rate₀ · 2^((T-37)/10), the same rule of thumb used to reason about enzymatic reaction speed in GMP process development.
The QC panel flags the batch as passing once Cap1 yield clears 85% and degraded loss stays under 15% — a simplified stand-in for the potency and integrity specifications a real mRNA manufacturing release protocol checks by LC-MS and gel electrophoresis.