Deep neural networks identify sequence context features predictive of transcription factor binding

Transcription factors bind DNA by recognizing specific sequence motifs, which are typically 6–12 bp long. A motif can occur many thousands of times in the human genome, but only a subset those sites actually bound. Here we present machine-learning framework leveraging existing convolutional neural network architectures and model interpretation techniques to identify interpret context features most important for predicting whether particular instance will be We apply our predict binding at motifs 38 transcription lymphoblastoid cell line, score importance sequences base-pair resolution characterize predictive binding. find that choice training data heavily influences classification accuracy relative such as open chromatin. Overall, enables novel insights into factor is likely inform future deep learning applications non-coding genetic variants. The process highly complex while short recognized well known, less known about determines its motif. Zheng colleagues method uses networks help transcribing proteins their target DNA.