Our methodology for quantifying the perturbations of biological networks provides a coherent framework for multifaceted results. a) Biological network model: the functional layer, or the backbone, is shown in orange, whereas the transcriptional layer is shown in black. Each causal edge is directed and signed. A given gene can be modulated by several nodes of the backbone model as depicted by several black arrows linking several nodes of the functional layer to a single node of the transcriptional layer. b) The fundamental paradigm shift between forward and backward reasoning; while the former considers the gene transcript abundance as a direct surrogate for its associated protein (or protein function), the latter considers the changes in gene transcription as the consequence of the activity of the upstream biological entity described by a node in the functional layer. c) Our new methodology provides a coherent framework connecting Network Perturbation Amplitude quantification (top panel), with mechanistic explanations and identification of the leading nodes in a network (middle panel) and finally the extraction of functional features (i.e., biomarkers) which can be used to stratify patient populations (bottom panel, points are individual samples in the new feature space and colors indicate a different phenotype).