IGPred

IGPred predicts immunoglobulins (antibodies) from protein sequence data to identify antibody sequences for studies of humoral immune function.

Key Features:

• Input: Operates on protein sequence data to classify sequences as immunoglobulins or non-immunoglobulins.
• Feature encoding: Represents protein sequences using pseudo amino acid compositions that incorporate nine physicochemical properties of amino acids.
• Prediction target: Distinguishes immunoglobulins (antibodies) from non-immunoglobulins.
• Evaluation: Performance reported using jackknife cross-validation with 96.3% correct prediction for immunoglobulins and 97.5% for non-immunoglobulins.

Scientific Applications:

• Antibody sequence annotation: Identification of immunoglobulin sequences within proteomes for antibody research and sequence curation.
• Diagnostic and biomedical research: Support for studies that require distinguishing antibody sequences in medical and diagnostic investigations.
• Biotechnology and therapeutic development: Prioritization of candidate antibody sequences for downstream biotechnological and therapeutic workflows.
• Experimental validation support: Generation of predicted immunoglobulin candidates to guide experimental validation of antibody sequences.

Methodology:

Protein sequences are encoded as pseudo amino acid compositions incorporating nine physicochemical properties and the model performance is assessed by jackknife cross-validation (96.3% accuracy for immunoglobulins, 97.5% for non-immunoglobulins).