Abstract:Knowledge of the sub-cellular localization of the most diverse class of transcribed RNA, long non-coding RNAs (lncRNAs) will lead us to identify different types of cancers and other diseases as lncRNAs play key role in related cellular functions. In recent days with the exponential growth of known records, it becomes essential to establish new machine learning based techniques to identify the new one due to faster and cheaper solutions provided compared to laboratory methods. In this paper, we propose Locate-R, a novel method for predicting the sub-cellular location of lncRNAs. We have used only n-gapped l-mer composition and l-mer composition as features and select best 655 features to build the model. This model is based locally deep support vector machines which significantly enhance the prediction accuracy with respect to exiting state-of-the-art methods. Our predictor is readily available for use as a stand-alone web application from: http://locate-r.azurewebsites.net/.

Locate-R: Subcellular Localization of Long Non-Coding RNAs Using Nucleotide Compositions.