In recent years, deep learning has shown progressive growth in its application to bioinformatics and has investigated excitingly auspicious power to mine the complicated interconnection hidden in large-scale biological and biomedical data. The most impressive research areas in bioinformatics using deep learning are Omics, Biomedical image processing, Biomedicine, and Drug Discovery.

Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Graph Convolutional Networks (GCN), Variational Auto-Encoder (VAE), Generative Adversarial Networks (GAN), and Deep Active Learning are the deep learning architectures applied for bio-informatics application. Symbolic reasoning empowered deep learning, Meta-learning, Deep generative models, Few-shot learning, Reinforcement learning, and Attention mechanism are the modern deep learning techniques in bioinformatics.

Identifying enzymes using multi-layer neural networks, Gene expression regression, RNA-protein binding sites prediction with CNN, DNA sequence function prediction with CNN and RNN, Biomedical image classification using transfer learning and ResNet, Graph embedding for novel protein interaction prediction using GCN, Biology image super-resolution using GAN and High dimensional biological data embedding and generation with VAE are the recent application scenarios of deep learning in bioinformatics.

Some limitations that need to be resolved in bioinformatics using deep learning are reducing computational requirement and model compression, catastrophic forgetting, uncertainty scaling, interpretability, imbalanced data, overfitting, and lack of data.

A number of comprehensive reviews and surveys have been published on deep learning in bio-informatics that describe the details, such as different deep learning architectures, essential concepts, applications, limitations, significant challenges, development tools, benchmark datasets, ongoing trends, and future perspectives.