DRpred: A Novel Deep Learning-Based Predictor for Multi-Label mRNA Subcellular Localization Prediction by...

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• Subcellular localization of messenger RNA (mRNA) is crucial for understanding gene expression regulation and its link to human diseases.
• Existing mRNA localization predictors show limitations, particularly in multi-label prediction accuracy.
• Accurate prediction of mRNA localization patterns is vital for biological and medical research.

Purpose of the Study:

• To develop a novel, high-performance multi-label predictor for mRNA subcellular localization.
• To improve upon the predictive performance of existing state-of-the-art methods.
• To provide a tool offering profound insights for biological and medical research.

Main Methods:

• Utilized Bayesian networks to model dependencies among localization labels.
• Employed Word2vec for feature extraction from mRNA sequences.
• Developed a neural network integrating BiLSTM and attention mechanisms for enhanced prediction.

Main Results:

• DRpred demonstrated competitive performance in multi-label mRNA subcellular localization prediction.
• Achieved high accuracies for single subcellular localizations, outperforming existing methods.
• Specific accuracies reported for chromatin (82.14%), cytoplasm (93.02%), cytosol (80.37%), exosome (94.00%), membrane (90.58%), nucleolus (84.53%), nucleoplasm (82.01%), nucleus (79.71%), and ribosome (85.67%).

Conclusions:

• DRpred offers a significant advancement in predicting mRNA subcellular localization, especially for multi-label scenarios.
• The predictor's performance suggests its utility in advancing biological and medical research.
• The integration of Bayesian networks, Word2vec, BiLSTM, and attention mechanisms provides a robust framework for RNA localization prediction.