A systems view of cellular reprogramming.

modulation to improve reprogramming outcomes. Daley, Collins and colleagues illustrated this function in the transcription factor–mediated conversion of murine B cells to macrophages. Finally, they used CellNet to argue that previously reported murine iHep cells should be reclassified as endodermal progenitors rather than as liver cells. As the researchers recognize, there are many potential confounding effects in analyses with CellNet. The gene expression data used to train the classifiers are mainly derived from in vivo tissue, which contains many primary cell types, whereas CellNet is used to assess single cell types, albeit also of unknown functional homogeneity, generated and cultured in vitro. Although the authors rule out crippling effects of these confounders for some cell types, these may prove more problematic in other cases. Also, CellNet is still only available for a limited number of cell types and for analysis of data obtained with microarrays; the researchers plan to extend it to RNA-seq data in the future. Even with a cell-specific GRN in hand, these studies illustrate that engineering cell identity is far from easy, not surprisingly given the molecular complexity involved. Its limitations notwithstanding, CellNet should be a very useful tool to evaluate fate conversion methods and the cells that result. Natalie de Souza