Gene Expression and Protein Function: A Survey of Deep Learning Methods

Deep learning methods have found increasing interest in recent years because of their wide applicability for prediction and inference in numerous disciplines such as image recognition, natural language processing, and speech recognition. Computational biology is a data-intensive field in which the types of data can be very diverse. These different types of structured data require different neural architectures. The problems of gene expression and protein function prediction are related areas in computational biology (since genes control the production of proteins). This survey provides an overview of the various types of problems in this domain and the neural architectures that work for these data sets. Since deep learning is a new field compared to traditional machine learning, much of the work in this area corresponds to traditional machine learning rather than deep learning. However, as the sizes of protein and gene expression data sets continue to grow, the possibility of using data-hungry deep learning methods continues to increase. Indeed, the previous five years have seen a sudden increase in deep learning models, although some areas of protein analytics and gene expression still remain relatively unexplored. Therefore, aside from the survey on the deep learning work directly related to these problems, we also point out existing deep learning work from other domains that has the potential to be applied to these domains.