PRED-SIGNAL

PRED-SIGNAL predicts signal peptides and their cleavage sites in archaeal proteins to identify secretory proteins and distinguish them from cytoplasmic and transmembrane proteins.

Key Features:

• Curated dataset: Uses a dataset of 69 experimentally verified archaeal proteins with signal peptides as the training and analysis basis.
• Archaeal signal peptide characteristics: Identifies distinctive features including a unique amino acid composition in the hydrophobic region with a higher occurrence of isoleucine and cleavage site patterns resembling those of Gram-positive bacteria.
• Hidden Markov Model (HMM): Employs a Hidden Markov Model trained on the curated dataset to predict presence of signal peptides and their cleavage sites and to distinguish secretory from cytoplasmic and transmembrane proteins.
• Performance metrics: Reports sensitivity of 100%, specificity of 98.41%, and a Matthews' correlation coefficient of 0.964, validated by a 35-fold cross-validation procedure.
• Genome-scale predictions: Applied to 48 completely sequenced archaeal genomes to identify 9,437 putative signal peptides.
• Comparative performance: Outperforms predictors that rely on eukaryotic or bacterial sequences for signal peptide prediction in archaea.

Scientific Applications:

• Genome annotation: Annotation of newly sequenced archaeal genomes by predicting signal peptides and cleavage sites for protein-coding genes.
• Protein secretion studies: Investigation of archaeal protein secretion pathways through identification of secretory proteins.
• Comparative genomics: Comparative analyses of signal peptide properties across archaeal species and versus bacterial and eukaryotic sequences.
• Proteome characterization: Distinguishing secreted proteins from membrane and cytoplasmic proteins in proteome annotations.

Methodology:

Training of a Hidden Markov Model on a curated set of 69 experimentally verified archaeal signal peptides, validation by 35-fold cross-validation, and genome-scale scanning of 48 completely sequenced archaeal genomes to predict 9,437 putative signal peptides.