Recent Advances of Deep Learning in Bioinformatics and Computational Biology

Abstract

Extracting inherent valuable knowledge from omics big data remains as a daunting problem in bioinformatics and computational biology. Deep learning, as an emerging branch from machine learning, has exhibited unprecedented performance in quite a few applications from academia and industry. We highlight the difference and similarity in widely utilized models in deep learning studies, through discussing their basic structures, and reviewing diverse applications and disadvantages. We anticipate the work can serve as a meaningful perspective for further development of its theory, algorithm and application in bioinformatic and computational biology.

Keywords: algorithm; application; bioinformatics; computational biology; deep learning.

Figure 1

The network structure of a deep learning model. Here we select a network structure with two hidden layers as an illustration, where X nodes constitute the input layer, Hs for the hidden layers, Y for the output layer, and f (·) denotes an activation function.

Figure 2

The general analysis procedure commonly adopted in deep learning, which covers training data preparation, model construction, hyperparameter fine-tuning (in training loop), prediction and performance evaluation. Basically, it still follows the requisite schema in machine learning.

Figure 3

Illustrative structure diagram of Recurrent Neural Network, where X, Y, and W are defined the same as above; L_i denotes the loss function between the actual Y_i and predicted Ŷ_i (i ∈ N).

Figure 4

The LSTM network structure and its general information flow chart, where X, Y, and W are defined the same as above.

Figure 5

The basic architecture and analysis procedure of a CNN model, which illustrates a classification procedure for an apple on a tree.

Figure 6

The illustrative diagram of an autoencoder model. (A) Basic processing structure of autoencoder, corresponding to the input, hidden, and output layers; (B) Processing steps in encoding; (C) Processing steps in decoding.

Figure 7

Illustrative network structures of RBM and DBN. (A) The structure of RBM. (B)Take the hidden layer of the trained RBM to function as the visible layer of another RBM. (C) The structure of a DBN. It stacks several RBMs on top of each other to form a DBN.

Figure 8

The schematic illustration of transfer learning. Given source domain and its learning task, together with target domain and respective task, transfer learning aims to improve the learning of the target prediction function, with the knowledge in source domain and its task.

Figure 9

Transfer learning has several derivatives categorized by the labeling information and difference between the target and source.