LabCaS

LabCaS predicts calpain substrate cleavage sites from amino acid sequences to identify calpain-mediated proteolytic events of the calcium-dependent cysteine protease family.

Key Features:

• Sequence-level labeling: Labels potential cleavage sites directly from entire protein sequences without fragmenting sequences into short peptides.
• Algorithm: Implements a conditional random field (CRF) algorithm to label cleavage sites.
• Machine learning: Leverages machine learning techniques for cleavage-site prediction.
• Feature integration: Integrates multiple amino acid features and sequence-derived data for site recognition.
• Cross-calpain applicability: Designed to recognize cleavage sites across various calpain proteins.
• Benchmark validation: Validated on a set of 129 benchmark proteins with an AUC of 0.862 in a jackknife test.

Scientific Applications:

• Cleavage-site mapping: Identification of calpain-mediated cleavage sites for studies of substrate processing.
• Mechanistic studies: Investigation of calpain substrate cleavage mechanisms and specificity.
• Biological implication analysis: Study of calpain functions and their implications in health and disease.

Methodology:

Uses machine learning with a conditional random field algorithm to label cleavage sites from full protein sequences, integrating multiple amino acid and sequence-derived features; performance was evaluated by a jackknife test on 129 benchmark proteins yielding an AUC of 0.862.