RAD: A Read-structure Agnostic Demultiplexer for Single-Cell Long-Read Sequencing and Analysis.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell long-read sequencing (LRS) offers full transcript analysis but faces computational hurdles due to high error rates and complex read structures.
• Existing methods often require custom workflows for accurate cell barcode extraction and error correction.

Purpose of the Study:

• To introduce RAD (Read-structure Agnostic Demultiplexer), an error-robust computational tool designed to overcome LRS data challenges.
• To enable efficient and accurate demultiplexing and barcode correction for diverse LRS data, including complex layouts.

Main Methods:

• RAD allows users to define read structures, including adapter sequences and barcode positions, for flexible element extraction.
• It incorporates efficient barcode correction strategies, supporting scenarios with known or unknown barcode whitelists and integration with short-read data.

Main Results:

• RAD demonstrates superior speed and significantly higher sensitivity compared to existing pipelines in both synthetic and real-world benchmarks, with comparable precision.
• The tool successfully applied to high-definition long-read spatial transcriptomic data.

Conclusions:

• RAD provides a robust and efficient solution for demultiplexing and error correction in single-cell long-read sequencing.
• Its application to spatial transcriptomics enables advanced single-cell and spatial analyses, such as B cell isotype and secretion state characterization.