MULocDeep

MULocDeep predicts protein localization across 10 major subcellular compartments and 45 distinct suborganellar annotations while providing residue-level interpretations to support studies of protein sorting and function.

Key Features:

• Comprehensive Localization Prediction: Predicts multiple localizations for a single protein across 10 major subcellular compartments and leverages a dataset with 45 distinct suborganellar annotations.
• Interpretability at Residue Level: Assesses each amino acid's contribution to localization, yielding residue-level interpretations that aid understanding of protein sorting and function.
• Experimental Validation: Evaluated on experimentally generated mitochondrial protein datasets from Arabidopsis thaliana cell cultures, Solanum tuberosum tubers, and Vicia faba roots.

Scientific Applications:

• Protein Function Analysis: Predicts precise localization to assist in assigning and interpreting protein functional roles within cellular compartments.
• Mechanism Insights: Provides residue-level interpretation to investigate mechanisms governing protein sorting, localization, dynamics, and interactions.

Methodology:

Employs a deep learning-based framework trained on extensive localization datasets to predict subcellular and suborganellar localizations and to provide residue-level interpretations.