NetCTL

NetCTL predicts human cytotoxic T lymphocyte (CTL) epitopes by integrating proteasomal cleavage, TAP transport, and MHC class I affinity predictions to support epitope identification for vaccine and immunotherapy research.

Key Features:

• Integrated prediction components: Combines proteasomal cleavage likelihood, TAP transport efficiency, and MHC class I affinity predictions into a single scoring framework.
• CTL epitope identification: Predicts potential human CTL epitopes from protein sequences.
• Versioned benchmark: NetCTL-1.2 version was evaluated and reported in comparative benchmarks.
• Benchmarking metrics: Uses ranking comparisons (RANK measure), specificity at predefined sensitivity levels, and the percentage of known epitopes within the top 5% highest-scoring peptides as performance measures.
• Comparative evaluation: Directly compared against EpiJen, MAPPP, MHC-pathway, and WAPP using curated epitope datasets.
• Large-scale benchmark: Evaluated on a set of 216 known HIV epitopes across all 12 recognized HLA supertypes, achieving sensitivity above 0.72 among the top 5% scoring peptides in that benchmark.
• Statistical comparison note: Demonstrated higher predictive performance than the compared methods across measures, with superiority over EpiJen and MHC-pathway not statistically significant on every measure.

Scientific Applications:

• Vaccine design: Prioritizes candidate CTL epitopes for inclusion in vaccine constructs.
• Immunotherapy research: Identifies antigenic peptides for CTL-based immunotherapies.
• Epitope discovery: Narrows experimental search space for CTL epitope validation from protein sequences.
• Method benchmarking: Serves as a reference in comparative evaluations of epitope-prediction methods using HIV epitope datasets and HLA supertype coverage.

Methodology:

Integrates computational predictions of proteasomal cleavage likelihood, TAP transport efficiency, and MHC class I affinity; evaluated using RANK measure, specificity at predefined sensitivity levels, and percentage of known epitopes within the top 5% highest-scoring peptides on two HIV epitope datasets and a 216-epitope benchmark across 12 HLA supertypes.