fiDPD

fiDPD predicts protein functional sites (PFSs) and their associated physicochemical interactions from sequence to inform identification of biochemical roles and evolutionary conservation.

Key Features:

• Functional site and physicochemical interaction annotation: Uses the fiDPD domain-profile database annotated for functional sites and physicochemical interactions, constructed from protein domains sourced from the Protein Data Bank (PDB).
• Sequence-based prediction methodology: Performs sequence-based analysis to identify PFSs and predict associated physicochemical interactions on protein surfaces.
• Validation and performance metrics: Validated on 13 target proteins from CASP10/11, achieving a Matthews correlation coefficient (MCC) of 0.66 for PFS prediction and 80% recall for physicochemical interaction prediction.
• Conservation insight: Reveals conservation of both PFSs and their physicochemical interactions across homologous proteins.

Scientific Applications:

• Rational drug design: Identifies PFSs and interaction properties that can guide selection of therapeutic targets.
• Side-effect assessment: Characterizes protein interaction sites that may mediate off-target biochemical reactions relevant to adverse effects.
• Biochemical reaction insight: Provides information on the physicochemical interactions at PFSs to infer biochemical roles of proteins.

Methodology:

Sequence-based prediction using the fiDPD functional site- and physicochemical interaction-annotated domain-profile database built from PDB domains; performance evaluated on 13 CASP10/11 targets (MCC = 0.66 for PFS prediction; 80% recall for interaction prediction).