A novel statistical method for decontaminating T-cell receptor sequencing data.

Area of Science:

• Immunology
• Bioinformatics
• Computational Biology

Background:

• The T-cell receptor (TCR) repertoire is crucial for immune responses and is profiled using TCR sequencing (TCR-seq).
• High-throughput TCR-seq experiments are susceptible to contamination during sample collection, preparation, and sequencing.
• Existing methods often fail to address data contamination, leading to inaccurate immunological insights.

Purpose of the Study:

• To develop a novel statistical model for systematic detection and removal of contamination in TCR-seq data.
• To provide tools for assessing contamination severity and implementing strategies for data correction.
• To enable accurate downstream analysis of T-cell repertoire data without repeating experiments.

Main Methods:

• Development of a statistical model to identify pairwise and cross-cohort contamination sources.
• Utilizing prior information from 14 existing TCR-seq datasets.
• Implementation of a Bayesian model for statistically identifying contaminated samples.
• Strategies for removing impacted sequences for downstream analysis.

Main Results:

• The proposed model effectively detects and quantifies contamination in TCR-seq data.
• Visualizations and summary statistics aid in assessing contamination severity.
• The model demonstrates robust performance in simulation studies compared to existing methods.
• Successful application of the method to two local TCR-seq datasets.

Conclusions:

• The novel statistical model provides a robust solution for detecting and removing contamination in TCR-seq data.
• This approach enhances the reliability of T-cell repertoire analysis.
• The method offers a cost-effective way to salvage potentially compromised datasets, avoiding costly re-experiments.