Predictive modeling of gene expression and localization of DNA binding site using deep convolutional neural networks

Abstract

Here, we developed a deep learning approach—called DARSI—that leverages these massively parallel reporter assays to predict levels of gene expression from DNA sequences and help locate these important binding sites. By training our model to recognize DNA sequence patterns that affect gene expression, our method not only finds known binding sites with high accuracy, but also predicts new binding sites that call for future experimental scrutiny.Author summary Understanding how genes are turned on and off is a fundamental question in biology. This process is often controlled by transcription factor proteins that bind to specific regions of DNA to activate or repress nearby genes. Identifying where these proteins bind is key to decoding how genes are regulated. Recently, large experimental datasets stemming from so-called massively parallel reporter assays in which thousands of slightly different DNA sequences are tested have opened the door to the identification of these binding sites in high-throughput.