Skip to main content
. 2021 Oct 12;16(10):e0258550. doi: 10.1371/journal.pone.0258550

Fig 1. The basic steps taken in Bayesian network reconstruction.

Fig 1

Different steps were followed as: 1) Estimation of DAGs between RBP and miRNA, RBP and its PPI partners based on the expression data and prior information. 2) Identification of significant DAGs from total DAGs considering a suitable convergence criteria. 3) The significant DAGs obtained in previous step were directly modeled via a generalized linear model, assuming a multivariate Gaussian distribution. 4) Parameter estimation between RBP and miRNA, RBP and its PPI partners based on the significant DAGs using a sparse regularized penalty. Different set of rho lambda values were taken into consideration and optimum lambda value were selected. 5) Selection of optimum sparse regularized penalty from an array of penalties that fit best to data. 6) Final estimation of parameters using the optimum penalty and selection of parameters considering certain suitable criteria.