Figure 1
From: Gene regulatory networks in plants: learning causality from time and perturbation
An experimental/computational systems-biology cycle using different data types and feedback. Starting from many possible edges, different data types and their analyses successively reduce the size of the network, while increasing confidence in edges. (1) Correlation leads to pairwise associations of genes. (2) Transgenic manipulation permits the determination of the effect of mutations and overexpression of single genes. (3) Binding experiments (for example, Chip-Seq) reveals physical connectivity of a source gene to a target. (4) Time-series experiments along with machine-learning techniques lead to a weighted network where the weight on the edge from A to B determines the extent of influence of A on B. (5) Subsequent predictions followed by validations can then suggest the need for new experimentation, thus refueling the systems-biology cycle.