Pf-Phospho

Pf-Phospho predicts phosphorylation sites in Plasmodium proteins (including Plasmodium falciparum and Plasmodium berghei) to enable computational analysis of phospho-signaling networks.

Key Features:

• Machine Learning Approach: Employs Random Forest classifiers trained on 12,096 known phosphosites from Plasmodium falciparum and Plasmodium berghei with a 75% training/validation and 25% test split derived from phosphoproteomics datasets.
• Predictive Performance: Reports sensitivity 84%, specificity 75%, and precision 78% for kinase-independent phosphosite prediction and includes kinase-specific predictions for plasmodial kinases PfPKG, PfPKA, PfPK7, and PbCDPK4.
• Integration with External Resources: Linked with PlasmoDB, MPMP, Pfam, and AlphaFold2-predicted structures to support contextual analyses with metabolic and protein–protein interaction networks.

Scientific Applications:

• Phospho-signaling network analysis: Enables mapping and computational analysis of phosphorylation-mediated signaling pathways in Plasmodium species.
• Regulatory mechanism exploration: Facilitates investigation of kinase-dependent and kinase-independent regulatory mechanisms in Plasmodium biology.
• Target prioritization for malaria research: Supports identification of candidate phosphorylation sites and kinases for functional studies and potential therapeutic intervention.

Methodology:

Uses Random Forest classifiers trained on 12,096 phosphosites from P. falciparum and P. berghei derived from phosphoproteomics data with a 75% training/validation and 25% test split; models produce kinase-independent and kinase-specific predictions for PfPKG, PfPKA, PfPK7, and PbCDPK4 and leverage integrations with PlasmoDB, MPMP, Pfam, and AlphaFold2-predicted structures.