Fig. 5: Performance of the RB-CME solver for the repressilator.
From: Advanced methods for gene network identification and noise decomposition from single-cell data
A Diagram of the repressilator with three gene expression systems whose proteins cyclically repress each other. Our method classifies all the mRNAs as leader-level species and all the proteins as follower-level species. B Convergence of the Monte Carlo method and the RB-CME solver (with the filtered FSP as the chosen filtering approach). We depict the error by the sum of the L1 errors in estimating the leader system and the follower system. The exact probability distribution is approximated by the Monte Carlo method with 3 × 109 samples. Given the same sample size or the same time cost, the RB-CME solver is much more accurate than the Monte Carlo method. C Performance of the RB-CME solver (with 104 samples) and the Monte Carlo method (with 105 samples) in estimating the marginal distributions. Both approaches have relatively the same computational time, and they both accurately estimate the marginal distributions of individual species. The superior performance of the RB-CME solver is attributed to the estimation of the follower system, which dominates the whole estimation problem. Source data are provided as a Source Data file.
