Date on Master's Thesis/Doctoral Dissertation
Computer Engineering and Computer Science
Computer Science, MS
Park, Juw Won
Committee Co-Chair (if applicable)
bioinformatics; circular RNA; machine learning; deep learning
Over the past two decades, a circular form of RNA (circular RNA) produced from splicing mechanism has become the focus of scientiﬁc studies due to its major role as a microRNA (miR) ac tivity modulator and its association with various diseases including cancer. Therefore, the detection of circular RNAs is a vital operation for continued comprehension of their biogenesis and purpose. Prediction of circular RNA can be achieved by ﬁrst distinguishing non-coding RNAs from protein coding gene transcripts, separating short and long non-coding RNAs (lncRNAs), and ﬁnally pre dicting circular RNAs from other lncRNAs. However, available tools to distinguish circular RNAs from other lncRNAs have only reached 80% accuracy due to the diﬃculty of classifying circular RNAs from other lncRNAs. Therefore, the availability of a faster, more accurate machine learning method for the identiﬁcation of circular RNAs, which will take into account the speciﬁc features of circular RNA, is essential in the development of systematic annotation. Here we present an End to-End multimodal deep learning framework, our tool, to classify circular RNA from other lncRNA. It fuses a RCM descriptor, an ACNN-BLSTM sequence descriptor, and a conservation descriptor into high level abstraction descriptors, where the shared representations across diﬀerent modalities are integrated. The experiments show that our tool is not only faster compared to existing tools but also eclipses other tools by an over 12% increase in accuracy. Another interesting result found from analysis of a ACNN-BLSTM sequence descriptor is that circular RNA sequences share the characteristics of the coding sequences
Chaabane, Mohamed, "End-to-end learning framework for circular RNA classiﬁcation from other long non-coding RNAs using multi-modal deep learning." (2018). Electronic Theses and Dissertations. Paper 2954.