Table 2.

General model structure.

Represented property	Update rule	Probability	Explanation
Auto-activation of gene modules	$m_e^A\left(t+1\right)={\mathbf{\text{m}}}_{\mathbf{\text{e}}}^{\mathbf{\text{A}}}\mathbf{\text{(t)}}\wedge \neg \left(m_e^B\left(t\right)\vee m_e^P\left(t\right)\right)\wedge \neg m_{m∕hc}^A\left(t\right)$	0.5/0.5	Regulatory proteins are closely coregulated and are often connected by positive feedback loops. (Boyer et al., 2005; Chickarmane and Peterson, 2008; MacArthur et al., 2008)
Pluripotency module activating DNA methylation through variable DNMT expression	$dnmt\left(t+1\right)={\mathbf{\text{m}}}_{\mathbf{\text{e}}}^{\mathbf{\text{P}}}\mathbf{\text{(t)}}\vee m_e^E\left(t\right)\vee dnmt\left(t\right)$	0.99	DNMT3 coregulated with Pluripotency genes. DNMT3 methylates unspecifically (Adewumi et al., 2007; Mah et al., 2011)
Mutual inhibition of gene modules	$m_e^A\left(t+1\right)=m_e^A\left(t\right)\wedge \neg \left(m_e^B\left(t\right)\vee m_e^P\left(t\right)\right)\wedge \neg m_{m∕hc}^A\left(t\right)$	0.5/0.5	Master Regulators inhibit other master regulators, competing lineages repress each other (Niwa et al., 2005; Ralston and Rossant, 2005; MacArthur et al., 2008)
Heterochromatin increases probability for DNA methylation	$m_m^A\left(t+1\right)=m_m^A\left(t\right)\vee dnmt\left(t\right)\wedge {\mathbf{\text{m}}}_{\mathbf{\text{hc}}}^{\mathbf{\text{A}}}\mathbf{\text{(t)}}$	0.05	Interaction via G9a complex: DNMT3A/B bind to nucleosomes with methylated histones such as H3K9me and methylates DNA (Cedar and Bergman, 2009)
Heterochromatin formation is inhibited by appropriate gene module	$m_{hc}^A\left(t+1\right)=m_{hc}^A\left(t\right)\vee m_m^A\left(t\right)\wedge \neg m_e^A\left(t\right)$	0.11	G9a binds specific sequences (Epsztejn-Litman et al., 2008)
DNA methylation increases probability for heterochromatin formation	$m_{hc}^A\left(t+1\right)=m_{hc}^A\left(t\right)\vee {\mathbf{\text{m}}}_{\mathbf{\text{m}}}^{\mathbf{\text{A}}}\left(\mathbf{\text{t}}\right)\wedge \neg m_e^A\left(t\right)$	0.17	Promotes chromatin inheritance after mitosis (Thomson et al., 2010)
DNA demethylation slower than other factors	$m_m^A\left(t+1\right)=m_m^A\left(t\right)\wedge \mathbf{\text{demeth(t)}}$	0.02	Passive cell cycle dependent demethylation through variable DNMT1 activity after mitosis (Li et al., 1992)
DNA demethylation is faster in euchromatin	$m_m^A\left(t+1\right)=m_m^A\left(t\right)\wedge \left(\mathbf{\text{demeth(t)}}\vee {\mathbf{\text{m}}}_{\mathbf{\text{hc}}}^{\mathbf{\text{A}}}\right)$	0.03	Histone deacetylase (HDAC) inhibitor TSA induces global and specific DNA demethylation (Ou et al., 2007)
Methylation not necessary to downregulate retroviral gene expression	$m_e^E\left(t+1\right)=\neg m_{hc}^E\left(t\right)\neg \vee m_m^E\left(t\right)$	0.5	Retroviral silencing is DNMT3A/B independent in the first 10 days of reprogramming (Pannell et al., 2000)
Retroviral gene demethylation is very slow in absence of DNMT3A/B or DNMT1	$m_m^E\left(t+1\right)=m_m^E\left(t\right)\wedge \left(\neg demeth\left(t\right)\vee dnmt\left(t\right)\right)$	0.001
Retroviral gene heterochromatin dynamics	$m_{hc}^E\left(t+1\right)=m_{hc}^E\left(t\right)\vee m_e^P\left(t\right)$	0.1	A complex between HDAC and NANOG (NODE complex responsible for the silencing of developmental genes) could account for retroviral silencing (Hotta and Ellis, 2008; Liang et al., 2008)

In bold, we represent the part of the variable’s update rule that reflects the modeled property referenced in the column Explanation. The column P contains the probabilities of the update rule.