A Machine Learning Approach to Motif Finding for LncRNA Sub-cellular Localization

Motifs are functional elements in DNA, RNA, and protein sequences. Motif finding in molecular sequences is well studied. However, applying machine learning techniques specifically integrating cytoplasmic/nuclear relative concentration index (CN-RCI) of genes across multiple cell lines to uncover motifs associated with lncRNA subcellular localization is still a new area of study. In this work, we start with the CN-RCI values from genes in multiple cell lines to develop a machine learning approach for identifying lncRNA motifs that are associated with lncRNA subcellular localization. The reliability of the approach is analyzed, first by using the identified motifs in experiments on classification of lncRNA subcellular localization, and then by comparison of the motifs against a library of known localization-related sequence fragments.