A generative model for constructing nucleic acid sequences binding to a protein

BMC Genomics

Table 1 Part of positive data from [19] used for training and testing DeepBind

Protein	Type	Species	Family	Experiment	AUC in test data
TEAD4	TF	H. sapiens	TEA	ChIP-seq	0.990
NFATC1	TF	H. sapiens	Rel	SELEX	0.909
DRGX	TF	H. sapiens	Homeodomain	SELEX	0.897
GCM1	TF	H. sapiens	GCM	SELEX	0.841
NFKB1	TF	H. sapiens	Rel	SELEX	0.771
OLIG1	TF	H. sapiens	bHLH	SELEX	0.733
RXRB	TF	H. sapiens	Nuclear receptor	SELEX	0.720
SOX2	TF	H. sapiens	Sox	SELEX	0.605
BHLHE23	TF	H. sapiens	bHLH	SELEX	0.557
MTF1	TF	H. sapiens	C2H2 ZF	SELEX	0.538
FOXP3	TF	H. sapiens	Forkhead	SELEX	0.499
MBNL1	RBP	H. sapiens	Znf	RNAcompete	–

Negative data was prepared by shuffling dinucleotides. TF: transcription factor. RBP: RNA binding protein. AUC of DeepBind for MBNL1 is not available

ISSN: 1471-2164