Table 1. Published transdifferentiation protocols used to evaluate our genomic model of cell identity.

Source cell	Target cell	Genomic Data		Transdifferentiation Factors
		Mouse	Human
Fibroblast	Myoblast	E P		MyoD1 [12]
Liver	Pancreas	E P H	E P H	Pdx1 [21]
Pancreatic islet	Liver	E P H	E P H	Cebpa or Cebpb [22]
Fibroblast	Hepatocyte	E P H	E P H	Hnf1a and one of: Foxa1, Foxa2, Foxa3 [49]; Gata4, Hnf1a, Foxa3, knockdown p19(Arf)* [50]
Fibroblast	Cardiomyocyte	E P	E P H	Gata4, Mef2c, Tbx5 [51]; Gata4, Tbx5, Baf60c* [52]
Fibroblast	Neuron	E P		Pou3f2 (also known as Brn2), Ascl1, Myt1l [53]
Liver	Neuron	E P		Pou3f2, Ascl1, Myt1l [54]
Fibroblast	Neural stem cell	E P H	E P H +	Sox2, Pou3f2, Foxg1 [14]; SOX2 [26]

We obtained genomic data (E = gene expression RNA-seq, P = Polycomb-associated H3K27me3 ChIP-seq, H = Heterochromatin-associated H3K9me3 ChIP-seq) for pairs of mouse and human tissues with published transdifferentiation protocols. Genomic datasets are listed in Table S1. Asterisks (*) mark genes that were not included in our testing, because they were not transcription factors (Baf60c) or were knocked-down in the protocol (p19(Arf)). Cross (+): two sources of data were used for human neural stem cells: cortex-derived neurospheres and in vitro derived neural progenitor cells. We assume that the published transdifferentiation protocols apply to both human and mouse cells.