Transcription factor interactions explain the context-dependent activity of CRX binding sites

We assessed how the effects of transcription factor binding sites (TFBSs) on the activity of a cis-regulatory sequence (CRS) depend on the local sequence context. We constructed a library of synthetic CRSs composed of TFBSs enriched in mouse photoreceptors and assayed them by massively parallel reporter assay in live mouse retina. We used the resulting data to train neural network-based models that predict CRS activity from sequence. A library of 6600 barcoded reporter gene plasmids were transfected into mouse retinas. Three replicates were performed. We measured barcoded reporter gene activity by RNA-seq, normalized by the DNA-seq reads of the input plasmid pool.