DeepSig

DeepSig predicts secretory signal peptides and cleavage sites in protein sequences to support accurate protein localization and functional characterization.

Key Features:

• Deep learning-based approach: DeepSig uses deep learning models to analyze protein sequences for signal peptide presence and cleavage-site positions.
• Joint prediction: Performs both signal peptide detection and precise cleavage-site identification in the same framework.
• Benchmark performance: Comparative benchmarks on an updated independent dataset of proteins report improved performance relative to existing state-of-the-art tools.
• Training data: Models are trained on comprehensive datasets of known signal peptides and cleavage sites.

Scientific Applications:

• Proteomics and protein localization: Facilitates annotation of secretory proteins and determination of subcellular targeting from primary sequences.
• Protein secretion and signaling studies: Supports investigation of protein secretion mechanisms and cellular signaling pathways that depend on signal peptide processing.
• Disease-related research: Aids studies of disease pathogenesis where mislocalization or altered secretion of proteins is implicated.

Methodology:

DeepSig trains deep learning models on comprehensive datasets of known signal peptides and cleavage sites and validates model performance against independent datasets for signal peptide and cleavage-site prediction.