Fig. 2. Hit-and-run methylation model.
(A) The schematic of the simulation algorithm. Orange ovals indicate the algorithm start and stop, yellow parallelograms show the input and output, blue boxes correspond to processes, the pink box indicate the “tethered methylation” subprocess, and green diamonds represent logical decisions (see Materials and Methods for further explanations and description of input parameters). (B) The dynamics of methylation pools from a representative hit-and-run simulation under optimized parameters. Here and in (C), simulations started with completely unmethylated nucleosomes. (C) Optimization of base methylation probability (kb) and H3K27me3 stimulation (S3) parameters. The heatmap representation of the sums of absolute differences between the mean fractions of H3K27me0, H3K27me1, H3K27me2, and H3K27me3 forecasted by the model and those measured by mass spectrometry in S2 and Kc cells indicates that parameter settings S3 = 3.5 and kb = 0.625 × 10−4 and S3 = 2.25 and kb = 0.75 × 10−4 (marked with dashed white boxes) result in predictions most closely resembling the experimental measurements (33). The fractions of variously methylated H3K27 positions predicted by the model were calculated on the basis of 10 inspection rounds from 1000 different simulations. The inspection rounds were selected randomly between the 216th and 240th simulation hour (10th cell cycle) when the system has reached the steady state. (D) Side-by-side comparison of forecasted and measured fractions of variously methylated H3K27 molecules. Box plots show the percentage of methylation positions in each state for S3 = 3.5 and kb = 0.625 × 10−4 calculated as above. Here and in all subsequent figures, the box plots indicate the medians (solid lines) and averages (dashed lines) and span interquartile range with whiskers covering the full extent of the data. The average percentages of corresponding methylated H3K27 molecules measured by mass spectrometry (33) are shown with yellow (S2 cells) and purple (Kc167 cells) dots.