TAPPred

TAPPred predicts peptide binding affinity to the transporter associated with antigen processing (TAP) to support identification of cytotoxic T lymphocyte (CTL) epitopes.

Key Features:

• Dataset: Analysis of 409 peptides with varying affinities for the human TAP transporter.
• Positional analysis: Examination of amino-acid distribution across nine positions (P1 to P9) within high-, intermediate-, and low-affinity TAP binders leading to derivation of predictive rules.
• Quantitative matrix: A scoring matrix based on position- and residue-specific contributions that correlated predicted and experimental affinities (r = 0.65).
• SVM model: A support vector machine using sequence information that improved correlation between predicted and experimental affinities (r = 0.80).
• Cascade SVM: A cascade SVM incorporating both amino acid features and sequence data that achieved a correlation of r = 0.88 and was validated by jackknife testing.
• Biological context: Models peptide processing and transport steps relevant to antigen presentation, including proteasome-generated peptides and transport to the endoplasmic reticulum via TAP.

Scientific Applications:

• CTL epitope identification: Support for identifying peptides likely to be transported by TAP and presented as cytotoxic T lymphocyte epitopes.
• Immunological research: Analysis of peptide-TAP interactions to study antigen processing mechanisms.
• Vaccine development: Prioritization of peptide candidates based on predicted TAP transport propensity.
• Therapeutic design: Informing design of peptide-based therapeutics by assessing TAP-binding potential.

Methodology:

Analysis of 409 peptides and positional amino-acid distributions across P1 to P9 to derive predictive rules; construction of a quantitative position/residue matrix (r = 0.65); development of an SVM using sequence information (r = 0.80); and refinement with a cascade SVM combining amino acid features and sequence data validated by jackknife testing (r = 0.88).