Using kinetic modeling to compare pre-mRNA-based and mRNA-based GRN inference methods. (A) The inference of regulatory activity, g(t), using target gene expression level (either pre-mRNA level or mRNA level) can be affected by various factors. These factors include the dynamics of the regulatory activity (i.e., Ton), the transcription rate of the target gene (α), the splicing rate of the pre-mRNA (β), and the degradation rate of the mRNA (γ). (B) Heatmaps showing the relative accuracy between pre-mRNA-based and mRNA-based inference of the regulatory activity in the absence of stochasticity. The model (Supplemental Fig. S1A) was analytically solved using indicated parameter combinations and the relative inference accuracies between pre-mRNA-based and mRNA-based methods were calculated (Methods). The redder the color, the more accurate the pre-mRNA-based method is. (C) Heatmap showing the relative accuracy between pre-mRNA-based and mRNA-based inference of the regulatory activity in the presence of stochasticity. The model (Supplemental Fig. S1A) was stochastically solved using indicated parameter combinations, and the relative inference accuracies between pre-mRNA-based and mRNA-based methods were calculated. Note that under low gene expression levels and slow regulatory dynamics, the pre-mRNA-based method underperforms compared to the mRNA-based method.