Computational inference (reverse engineering) |
Infers putative regulatory relationships from gene expression data |
Fast; cheap |
The interactions predicted could be indirect (they do not have to reflect physical interactions); regulators that themselves do not change in expression will be missed; detection limits of mRNA measurements will affect GRN predictions (regulators or genes expressed at low levels will be missed) |
Chromatin immunoprecipitation (ChIP) |
Experimentally identifies physical interactions between TFs and DNA; TF-centered (protein-to-DNA) |
In vivo; can detect TF dimers and complexes |
Condition-dependent interactions can be missed; needs high-quality, highly specific ChIP-grade antibodies; when a universal tag is used for immunoprecipitation, TF is usually overexpressed; peak calling required |
Yeast one-hybrid (Y1H) |
Experimentally identifies physical interactions between DNA and TFs; gene-centered (DNA-to-protein) |
Heterologous; condition- and context-independent |
TFs that require post-translational modifications before binding DNA will be missed; not yet suitable for heterodimers |