FaaPred

FaaPred predicts fungal adhesins and adhesin-like proteins using Support Vector Machine (SVM) classifiers trained on compositional features and Position-Specific Scoring Matrix (PSSM) data from PSI-BLAST to facilitate identification and study of adhesion-related roles in fungal pathogenesis.

Key Features:

• Support Vector Machine (SVM) classifiers: Uses SVM models as the core predictive algorithm.
• Amino acid composition: Incorporates amino acid composition as an input feature.
• Dipeptide frequencies: Includes dipeptide frequency profiles for sequence-level information.
• Multiplet fractions: Utilizes multiplet fraction features derived from sequence composition.
• Charge distribution: Accounts for residue charge distribution across proteins.
• Hydrophobicity profiles: Employs hydrophobicity-based features.
• PSSM matrices (PSI-BLAST): Integrates Position-Specific Scoring Matrix matrices obtained from PSI-BLAST searches.
• Combined-feature performance: Best-performing classifiers combine compositional properties and PSSM to achieve 86% overall prediction accuracy.

Scientific Applications:

• Adhesin identification: Predicts candidate fungal adhesins and adhesin-like proteins from protein sequences.
• Pathogenesis research: Facilitates study of adhesion-related roles in fungal pathogenesis, including host cell attachment and mating.
• Aggregation and biofilms: Supports investigation of homotypic and xenotypic aggregation, foraging, and biofilm formation mechanisms.
• Experimental prioritization: Helps prioritize candidates for experimental characterization of novel adhesins.
• Proteome annotation: Enables annotation of fungal proteomes for adhesin-like proteins.

Methodology:

Support Vector Machine classifiers trained on features including amino acid composition, dipeptide frequencies, multiplet fractions, charge distributions, hydrophobicity profiles, and Position-Specific Scoring Matrix (PSSM) matrices derived from PSI-BLAST, with best-performing models combining compositional properties and PSSM.