Differential architecture search in deep learning for genomic applications

The data explosion caused by unprecedented advancements in the field of genomics is constantly challenging the conventional methods used in the interpretation of the human genome. The demand for robust algorithms over the recent years has brought huge success in the field of Deep Learning (DL) in solving many difficult tasks in image, speech and natural language processing by automating the manual process of architecture design. This has been fueled through the development of new DL architectures. Yet genomics possesses unique challenges as we expect DL to provide a super human intelligence that easily interprets a human genome. In this paper, the state-of-the art DL approach based on differential search mechanism was adapted for interpretation of biological sequences. This method has been applied on the splice site recognition task on raw DNA sequences to discover high-performance convolutional architectures by automated engineering.The discovered architecture achieved comparable accuracy when evaluated with a fixed Recurrent Neural Network (RNN) architecture. The results have shown a potential of using this automated architecture search mechanism for solving other problems in genomics.