A Genetic Algorithm for Finding Discriminative Functional Motifs in Long Non-coding RNAs

摘要

Long non-coding RNAs (lncRNAs), each with >200 nucleotides in length, constitute a large portion of the human transcriptome. Although recent studies indicate that lncRNAs play key roles in gene regulation, development and disease, the RNA functional motifs are still poorly understood. Most of the existing algorithms for motif finding are severely limited in scalability with regards to sequence and motif size. In this study, we propose a novel genetic algorithm for discriminative motif identification capable of handling large input sequences and motif sizes by utilizing genetic operators to learn and evolve in response to the input sequences. We utilize our method on long non-coding RNA (lncRNA) transcripts as a test case to identify functional motifs associated with subcellular localization. Our methodology shows high accuracy and the ability to identify functional motifs associated with subcellular localization in lncRNAs, which recapitulates a previous experimental study.