Extended Data Fig. 1: Simulation to test the relationship between the relative deadenylation rate and the local poly(A) density at a steady state.

a. The production (π) of mRNAs and their poly(A) tails, which includes transcription and 3′ end processing, was modeled with a Gaussian distribution. Deadenylation (δ) and decay (φ) were modeled with logistic functions of the poly(A) length. Deadenylation rates were varied with three different parameters (i)–(iii): uniform, decreasing, and increasing along the poly(A) length (long-to-short), respectively. The number of transcripts over the cycles is shown on the third column to ensure that the system reaches a steady state. The light green-to-black lines in the plots in the right-most column represent the poly(A) length distribution over the cycles. b. Another computer simulation with a different production model. The initial distribution of poly(A) length in the production (π) step is modeled with the negative binomial distribution to account for a more realistic scenario based on the experimental observation³⁴.