Multi-tissue Analysis of Co-expression Networks by Higher-Order Generalized Singular Value Decomposition Identifies Functionally Coherent Transcriptional Modules
Figure 1
Illustration of the C3D method.
Graphical summary of the main steps of the C3D method: (1) data initialization, (2) HO-GSVD based algorithm and (3) cluster nodes selection and validation. Input data can be either gene expression or co-expression matrices (graphs) and the output include information about the identified clusters (cluster density, formatted network file), the conditions where the clusters are detected and the cluster significance (p-value). To retrieve significant clusters, the user can specify (i) the misclassification error rate (MER) for inclusion of genes in the cluster and (ii) the empirical p-value for significance of the cluster.