pseapred

pseapred predicts proteins secreted by the malaria parasite Plasmodium falciparum into infected erythrocytes to enable identification of potential vaccine and drug targets.

Key Features:

• Support Vector Machine (SVM) models: Support Vector Machine classifiers were used to distinguish secretory from non-secretory proteins.
• Dataset: A balanced dataset comprising 252 secretory and 252 non-secretory proteins was used for training and testing.
• Sequence representations: Features included amino acid composition, dipeptide composition, split-amino acid composition, and split-dipeptide composition.
• PSSM profiles: Position-Specific Scoring Matrix (PSSM) profiles derived from PSI-BLAST searches were used as input features.
• Performance metrics: Reported results were: amino acid composition SVM MCC 0.72 (85.65% accuracy); dipeptide MCC 0.74 (86.45%); split-amino acid MCC 0.74 (86.40%); split-dipeptide MCC 0.77 (88.22%); PSSM-based SVM MCC 0.86 (92.66% accuracy).
• Validation: Model performance was assessed using 5-fold cross-validation.
• Biological insight: Analyses emphasized the importance of residue composition and indicated that multiple sequence information (via PSSM) is more informative than single-sequence composition alone.

Scientific Applications:

• Secretory protein prediction: Prediction of Plasmodium falciparum proteins secreted into infected erythrocytes.
• Target identification: Identification and prioritization of potential vaccine and drug targets among secretory proteins.
• Parasite biology studies: Investigation of roles of secretory proteins in parasite growth, survival, and interactions with infected erythrocytes.

Methodology:

Support Vector Machine classifiers were trained and evaluated on a dataset of 252 secretory and 252 non-secretory proteins using features from amino acid composition, dipeptide and split compositions, and PSSM profiles obtained via PSI-BLAST, with performance assessed by 5-fold cross-validation.