LocTree2

LocTree2 predicts subcellular localization of proteins across Archaea, Bacteria, and Eukaryota to inform protein function.

Key Features:

• Domain-Specific Classification: Classifies proteins into 3 classes for Archaea, 6 classes for Bacteria, and 18 classes for Eukaryota.
• Protein Fragment Support: Provides predictions applicable to protein fragments as well as complete sequences.
• Hierarchical Support Vector Machines: Employs a hierarchical system of support vector machines for localization prediction.
• Membrane vs. Non-Membrane Differentiation: Separates membrane-bound proteins from non-membrane proteins in predictions.
• Performance Metrics: Reports Q18 = 65% for Eukaryota and Q6 = 84% for Bacteria.

Scientific Applications:

• Functional Genomics: Infers potential protein roles and interactions by predicting cellular localization.
• Protein Engineering: Informs design of proteins with targeted cellular destinations.
• Comparative Biology: Enables comparative studies of protein localization across Archaea, Bacteria, and Eukaryota.

Methodology:

Uses a hierarchical cascade of support vector machines to assign proteins, including fragments, to domain-specific localization classes and to distinguish membrane versus non-membrane proteins.