Sephiroth

Sephiroth predicts disulfide connectivity patterns in proteins for structural and functional annotation of cysteine residues.

Key Features:

• Evolutionary unsupervised approach: Uses an evolutionary-based unsupervised framework to infer disulfide connectivity from sequence variation.
• Multiple Sequence Alignments (MSAs) via HHblits: Generates high-quality MSAs using HHblits to capture homologous sequence information.
• Coarse-grained cluster-based modelization: Applies a coarse-grained cluster-based modelization of tandem cysteine mutations within protein families.
• Prediction target: Predicts intra-chain disulfide bond connectivity patterns starting from known cysteine bonding states.
• Improved accuracy: Reports a 25-27% performance improvement over state-of-the-art unsupervised predictors.
• Reduced homolog requirement: Lowers the number of aligned homologous sequences required for accurate prediction from ~10^4 to ~10^3.

Scientific Applications:

• Protein structural and functional annotation: Supports annotation of cysteine residues and disulfide bonds in protein structure databases and studies.
• Protein folding and stability studies: Informs analyses of protein folding, stability, and the role of disulfide bonds in structural integrity.
• Evolutionary analysis of cysteine mutations: Enables investigation of tandem cysteine mutation patterns within protein families.

Methodology:

Sephiroth uses HHblits to generate high-quality MSAs and applies an evolutionary-based unsupervised approach with a coarse-grained cluster-based modelization of tandem cysteine mutations to predict intra-chain disulfide connectivity from known cysteine bonding states.