Locate-R

Locate-R predicts the subcellular localization of long non-coding RNAs (lncRNAs) using nucleotide composition features to inform their cellular roles and disease relevance.

Key Features:

• Feature Selection: Uses n-gapped l-mer and l-mer nucleotide composition features and selects a subset of 655 features for modeling.
• Machine Learning Model: Implements locally deep support vector machines (SVMs) for classification of lncRNA subcellular localization and reports improved accuracy compared to existing state-of-the-art methods.

Scientific Applications:

• Disease Identification: Predicts lncRNA localization to aid identification of potential biomarkers associated with various diseases, particularly cancers.
• Functional Insights: Provides localization information to support inference of lncRNA functional roles and mechanisms of action within cells.

Methodology:

Extracts n-gapped l-mer and l-mer nucleotide composition features, performs feature selection to retain 655 features, and applies locally deep support vector machines (SVMs) for localization prediction.