Carboxylator

Carboxylator predicts γ-carboxyglutamate (Gla) formation by identifying carboxylation sites on glutamate residues in proteins to support studies of γ-glutamyl carboxylase–mediated modifications relevant to the blood clotting cascade and bone growth.

Key Features:

• RBFN model: Constructs a predictive model using a radial basis function network (RBFN) to identify carboxylation sites.
• AA20D sequence encoding: Uses AA20D amino acid sequence representation as an input feature.
• Amino acid composition: Incorporates amino acid composition as an input feature.
• Solvent-accessible surface area (ASA): Uses ASA to capture residue surface accessibility associated with carboxylation preference.
• Cross-validation performance: Evaluated with five-fold cross-validation achieving an accuracy of 0.874.
• Independent testing: Validated on independent datasets not included in cross-validation.
• Target modification: Predicts conversion of glutamate residues to γ-carboxyglutamate (Gla) catalyzed by γ-glutamyl carboxylase.

Scientific Applications:

• Site identification: Identification of potential γ-carboxylation sites on glutamate residues in protein sequences.
• Functional studies: Support investigations of γ-glutamyl carboxylase–mediated modifications in the blood clotting cascade and bone growth.
• Experimental complement: Provide preliminary in silico screening that complements experimental approaches such as mass spectrometry.

Methodology:

Builds a predictive model using a radial basis function network (RBFN) with AA20D sequence encoding, amino acid composition, and solvent-accessible surface area (ASA) as input features; model performance was assessed by five-fold cross-validation (accuracy 0.874) and independent testing.