Table 5.

Methods for reconstruction of regulatory networks.

Approaches	Highlights	Challenges	Organisms studied	References
Differential equations	Network dynamic over time, regulation and optimization of function	High computational demanding, complex parameter optimization	Mus musculus, Candida albicans	Linde et al., 2015
Boolean	Switch-like behavior, efficient and easy interpretation	Only two states, good in small networks, Only synchronous interactions	H. pylori	Franke et al., 2008
Bayesian*	Robust to deal of disturbances, integrated knowledge to increase the support	Non-dynamical, high computational cost, often used a hybrid method to increase the accuracy	E. coli	Yang et al., 2011
Neural networks	Allows continuous variables over time, very sensitive for regulated systems, noise-resistant	Computational complex, difficult for training, need a lot of input data	Caulobacter crescentus, E. coli, Bacillus subtilis	Yaghoobi et al., 2012; Umarov and Solovyev, 2017
State space model	High computational efficiency, probabilistic framework to simulate the network, determines an optimal threshold value	There are no learning steps	Saccharomyces cerevisiae, Aspergillus fumigatu	Do et al., 2009; Koh et al., 2009

^*To counteract the stationary problem of Bayesian networks, The dynamic Bayesian network approach was developed.