ErrorX

ErrorX corrects sequencing errors in next-generation sequencing (NGS) datasets to improve the accuracy of B- and T-cell receptor sequence analyses.

Key Features:

• Automated Error Correction: Employs deep learning to identify and correct bases with a high probability of being erroneous, including errors arising from PCR during library preparation and miscalled bases on sequencing instruments.
• Deep Learning Integration: Leverages machine learning models to distinguish true biological sequence variation from sequencing errors in B- and T-cell receptor data.
• Benchmark Performance: Demonstrated reductions in overall error rate in public datasets by up to 36% while maintaining a false positive rate of 0.05% or less.
• Universal Application: Applies directly to existing antibody and T-cell receptor sequencing datasets without requiring changes to library preparation protocols.

Scientific Applications:

• Immune repertoire profiling: Improves the reliability of deep profiling analyses of B-cell and T-cell receptor repertoires by reducing sequencing-induced artifacts.
• Immunology research and related studies: Enhances the accuracy of downstream biological conclusions drawn from NGS-based receptor sequencing datasets.

Methodology:

Trains deep learning models on known datasets to recognize patterns indicative of sequencing errors and to predict and correct erroneous bases.