An enhance cnn-rnn model for predicting functional non-coding variants

• Main Authors: Mohd. Kamarudin, Jalilah Arijah, Ahmad Ahyad, Nur Afifah, Abdullah, Afnizanfaizal, Sallehuddin, Roselina
• Format: Article
• Language: English
• Published: Little Lion Scientific 2018
• Subjects: QA76 Computer software
• Online Access: http://eprints.utm.my/id/eprint/84181/1/RoselinaSalleh2018_AnEnhanceCnn-RnnModelForPredicting.pdf; http://eprints.utm.my/id/eprint/84181/; http://www.jatit.org/volumes/Vol96No11/17Vol96No11.pdf
• Institution: Universiti Teknologi Malaysia

In the era of big data, deep learning has advanced rapidly particularly in the field of computational biology and bioinformatics. In comparison to conventional analysis strategies, deep learning method performs accurate structure prediction because it can handle high coverage biological data such as DNA sequence and RNA measurement using high-level features. However, predicting functions of non-coding DNA sequence using deep learning method have not been widely used and require further study. The purpose of this study is to develop a new algorithm to predict the function of non-coding DNA sequence using deep learning approach. We propose an enhanced CNN-RNN model to predict the function of non-coding DNA sequence. In this model, we train an algorithm to automatically find the optimal initial weight and hyper-parameter to increase prediction accuracy which outperforms other prediction models.