Identification of piRNA disease associations using deep learning

Abstract

Piwi-interacting RNAs (piRNAs) play a pivotal role in maintaining genome integrity by repression of transposable elements, gene stability, and association with various disease progressions. Cost-efficient computational methods for the identification of piRNA disease associations promote the efficacy of disease-specific drug development. In this regard, we developed a simple, robust, and efficient deep learning method for identifying the piRNA disease associations known as piRDA. The proposed architecture extracts the most significant and abstract information from raw sequences represented in a simplicated piRNA disease pair without any involvement of features engineering. Two-step positive unlabeled learning and bootstrapping technique are utilized to abstain from the false-negative and biased predictions dealing with positive unlabeled data. The performance of proposed method piRDA is evaluated using k-fold cross-validation. The piRDA is significantly improved in all the performance evaluation measures for the identification of piRNA disease associations in comparison to state-of-the-art method. Moreover, it is thus projected conclusively that the proposed computational method could play a significant role as a supportive and practical tool for primitive disease mechanisms and pharmaceutical research such as in academia and drug design. Eventually, the proposed model can be accessed using publicly available and user-friendly web tool athttp://nsclbio.jbnu.ac.kr/tools/piRDA/.

Keywords: Convolutional Neural Network; Deep learning; Positive unlabeled learning; Reliable negative sample; Sequence analysis; Web-server; piRNA disease associations.

Graphical abstract

Fig. 1

The overall workflow of proposed Architecture piRDA for identifying piRNA disease associations.

Fig. 2

Illustration of reliable negative selection. (a) Positive and unlabeled data samples. (b) Training with random negative. (c) Unlabeled samples according to their prediction scores.

Fig. 3

Illustrating detailed architecture of proposed method piRDA where the convolutional block comprises convolution layer with ReLU as an activation function along with group normalization and max-pooling layers. The dense block consists of two fully connected layers along with dropout probability ReLU as an activation function and sigmoid activation function for prediction associated scores.

Fig. 4

Illustration of the five folds success rate (ROC), with associated calculation of prediction quality (AUC) and standard deviation error.

Fig. 5

Illustration of the five folds PRC together with (AUPRC) and standard deviation error.

Fig. 6

Illustration of ROC along with AUC and standard deviation error of sub 10-fold cross-validation.

Fig. 7

Illustration of PRC along with AUC and standard deviation error of sub 10-fold cross-validation.

Fig. 8

Clusters of positive and negative piRNA disease associations features of the proposed method obtained from hidden layer activation using UMAP.

Fig. 9

Illustration of evaluation measures comparision of piRDA with existing methods for identifying piRNA disease associations.