TCRGP

TCRGP predicts T cell receptor (TCR) recognition of specific epitopes using a Gaussian process framework to model sequence-based TCR specificity.

Key Features:

• Gaussian Process Methodology: Employs a Gaussian process framework to model complex relationships in TCR sequence data for epitope recognition prediction.
• Utilization of CDR Sequences: Analyzes amino acid sequences from complementarity determining regions (CDRs) of both TCRα and TCRβ chains and identifies which CDRs are essential for epitope recognition.
• Single-Cell Data Integration: Processes single-cell data, including combined scRNA and TCRαβ sequencing datasets.
• High Prediction Accuracy: Demonstrates superior prediction performance on epitope-specific TCR sequencing data as quantified by higher average AUROC scores compared to existing methods.

Scientific Applications:

• Understanding Immune Status: Assesses TCR specificity to inform individuals' immune status across various disorders.
• Epitope-Specific Analysis: Predicts which CDRs are crucial for recognizing specific epitopes to elucidate TCR–epitope interactions.
• Single-Cell RNA and TCR Sequencing Integration: Quantifies epitope-specific TCRs and identifies epitope-specific immune cells (e.g., HBV-epitope specific T cells) and their transcriptomic states in contexts such as hepatocellular carcinoma.

Methodology:

Applies a Gaussian process model to amino acid sequences from CDRs of TCRα and TCRβ, processes single-cell scRNA and TCRαβ sequencing data, and is evaluated on epitope-specific TCR sequencing datasets using AUROC metrics.