AutoCAT

AutoCAT predicts cancer-associated T-cell receptors (TCRs) from targeted TCR sequencing (TCR-seq) data to identify tumor-associated receptors as biomarkers of cancer immune response.

Key Features:

• Input data: Uses targeted TCR sequencing (TCR-seq) data as the primary input for model training and prediction.
• Prediction: Predicts cancer-associated T-cell receptors (TCRs) to enable identification of tumor-associated clones.
• Model benchmarking: Benchmarks a new predictive model that utilizes TCR-seq data to identify tumor-associated TCRs.
• Comparison to DeepCAT: Maintains predictive accuracy relative to DeepCAT, which was trained using tumor bulk RNA-seq samples.
• Sample classification: Differentiates healthy and cancer patient samples based on T-cell receptor profiles.
• Biomarker identification: Facilitates discovery of TCRs specifically associated with cancerous conditions as biomarkers of immune response.

Scientific Applications:

• TCR biomarker discovery: Identification of tumor-associated TCRs as biomarkers for cancer immune response studies.
• Immunotherapy research: Supports development and evaluation of cancer immunotherapy strategies by identifying tumor-associated TCRs.
• Diagnostic profiling: Enables differentiation between healthy and cancer patient samples through T-cell repertoire analysis.
• Computational immunology: Provides a benchmark for predictive models using TCR-seq data in computational immunology research.

Methodology:

Benchmarks a predictive model trained on targeted TCR sequencing (TCR-seq) data and compares performance to models trained on tumor bulk RNA-seq (such as DeepCAT).