gShapeLnoc: A Graph Network Incorporating Shapelet Embedding Model for LncRNA Subcellular Localization.

Area of Science:

• Genomics and Molecular Biology
• Bioinformatics and Computational Biology

Background:

• Subcellular localization of Long non-coding RNAs (LncRNAs) is critical for understanding gene regulation and disease mechanisms.
• Existing machine learning methods often overlook global LncRNA features or fail to account for mutations.
• Deep learning models improve feature integration but still have limitations in capturing comprehensive LncRNA characteristics.

Purpose of the Study:

• To develop an advanced computational method for accurate prediction of LncRNA subcellular localization.
• To integrate both global and local feature representations for enhanced LncRNA classification.
• To address limitations of previous methods, including the potential impact of mutations.

Main Methods:

• Utilized the Shapelet model to extract representative local k-mer features from LncRNAs.
• Developed the gShapeLnoc algorithm, which combines global sequence features with local Shapelet-derived features.
• Evaluated the algorithm's performance on a real-world dataset for LncRNA subcellular localization prediction.

Main Results:

• The gShapeLnoc algorithm demonstrated superior performance compared to existing state-of-the-art methods.
• The combination of global and local features significantly improved prediction accuracy for LncRNA subcellular localization.
• The method effectively captures essential LncRNA characteristics for robust classification.

Conclusions:

• The gShapeLnoc algorithm represents a significant advancement in predicting LncRNA subcellular localization.
• Integrating diverse feature types (global and local) is key to improving classification accuracy.
• This approach offers a more comprehensive and accurate tool for LncRNA research.