Unified mRNA Subcellular Localization Predictor based on machine learning techniques.

Area of Science:

• Molecular Biology
• Bioinformatics
• Computational Biology

Background:

• Messenger RNA (mRNA) subcellular localization is vital for gene expression, cellular migration, and adaptation.
• Experimental methods for determining mRNA localization are costly, time-consuming, and labor-intensive.

Purpose of the Study:

• To develop an efficient and accurate computational method for predicting mRNA subcellular localization.
• To address the limitations of current experimental techniques through an in silico approach.

Main Methods:

• Utilized a machine learning (ML) based approach named Unified mRNA Subcellular Localization Predictor (UMSLP).
• Employed an in silico strategy incorporating four feature sets: kmer, pseudo k-tuple nucleotide composition, nucleotide physicochemical attributes, and Z-curve transformation.
• Validated the model on a benchmark dataset across five distinct subcellular locales: nucleus, cytoplasm, extracellular region (ExR), mitochondria, and endoplasmic reticulum (ER).

Main Results:

• UMSLP achieved high prediction performance with over 87% precision, 94% specificity, and 94% accuracy on an independent testing dataset.
• UMSLP significantly outperformed existing tools like mRNALocator, mRNALoc, and SubLocEP in average prediction accuracy.
• SHapley Additive exPlanations revealed that k-mer features are dominant for most localizations, while Z-curve features are crucial for mitochondria prediction.

Conclusions:

• UMSLP provides a powerful and accurate tool for predicting mRNA subcellular localization.
• The developed in silico method offers a cost-effective and efficient alternative to experimental approaches.
• The study highlights the importance of integrated feature sets for robust prediction of mRNA localization.