Fig 1.
Flowchart of the time-varying scale-free graphical lasso method for dynamic scale-free gene network modeling.
Fig 2.
Analysis of data replicates generated by the R package huge [32] (a scale-free model) using different glasso variants and tvsfglasso at .
Different panels represent different binary classification metrics.
Fig 3.
Analysis of data replicates generated by simulator described in [18] (a scale-free model) using different glasso variants and tvsfglasso at .
Different panels represent different binary classification metrics.
Fig 4.
Analysis of data replicates generated by simulator described in [12] (a smooth scale-free model) using different glasso variants and tvglasso and tvsfglasso at .
Different panels represent different binary classification metrics.
Fig 5.
Analysis of data replicates generated by simulator described in [33] (a super-hub model) using different glasso variants and tvglasso and tvsfglasso at .
Different panels represent different binary classification metrics.
Fig 6.
Analysis of data replicates generated by simulator described in [13] (non-scale-free network) using different glasso variants and tvglasso and tvsfglasso at .
Different panels represent different binary classification metrics.
Fig 7.
Analysis of data replicates generated by simulator described in [12] (smooth Erdős-Rényi graph) using different glasso variants and tvglasso and tvsfglasso at .
Different panels represent different binary classification metrics.
Fig 8.
Estimated degree of the time-varying TFs network constructed using tvsfglasso at (A) for the whole network and (B) for the TFs with the greatest increases in degree, as determined by the highest slope coefficients
,
+
time. Different line types represent different TFs.
Fig 9.
Density estimates of degree distributions of the time-varying Drosophila melanogaster TF networks constructed with tvsfglasso at at time-points (A) 0h, and (B) 20h.
Fig 10.
The neighborhood of BLIMP-1 in the Drosophila melanogaster networks at times-points (A) 16h, (B) 18h, and (C) 20h when the network is constructed using tvsfglasso at .