Loc4Lnc: Accurate prediction of long noncoding RNA subcellular localization via enhanced RNA sequence representation

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• Long noncoding RNAs (lncRNAs) are key regulators of gene expression, chromatin structure, and cellular processes.
• Their subcellular localization is critical for understanding biological functions and disease mechanisms.
• Existing prediction methods struggle with capturing long-range sequence dependencies.

Purpose of the Study:

• To develop an advanced deep learning framework, Loc4Lnc, for accurate prediction of lncRNA subcellular localization.
• To address the limitations of current methods in handling long-range interactions within RNA sequences.

Main Methods:

• Developed Loc4Lnc, a deep learning model integrating convolutional layers and transformer blocks.
• Utilized TextCNN for classification after capturing local motifs and long-range dependencies.
• Constructed a benchmark dataset from the RNALocate v2.0 database covering five subcellular locations.

Main Results:

• Loc4Lnc achieved a prediction accuracy of 0.636 on an independent test set.
• The model significantly outperformed existing feature extraction methods and state-of-the-art predictors.
• Demonstrated robustness and effectiveness in classifying lncRNAs across cytoplasm, nucleus, cytosol, chromatin, and exosome.

Conclusions:

• Loc4Lnc provides an effective tool for predicting lncRNA subcellular localization by capturing crucial long-range interactions.
• The framework enhances understanding of lncRNA roles in gene regulation and cellular functions.
• Lays groundwork for future research into lncRNA-mediated biological processes and disease.