Barcode identification for single cell genomics.

Area of Science:

• Genomics
• Bioinformatics

Background:

• Single-cell sequencing relies on DNA barcode tags to identify cellular origin.
• Grouping reads by barcode is essential but challenging due to sequencing errors.
• High rates of mismatch and deletion errors in barcodes impede accurate data analysis.

Purpose of the Study:

• To develop a robust method for identifying and correcting errors in DNA barcodes.
• To improve the accuracy of read assignment in single-cell sequencing experiments.
• To enhance the recovery of single-cell transcriptome data.

Main Methods:

• Utilizes de Bruijn graphs of circularized barcode k-mers for error identification.
• Employs circularization to generate error-free k-mers, even with short barcodes.
• Assigns reads to consensus fingerprints derived from k-mers.

Main Results:

• Demonstrates improved recovery of accurate single-cell transcriptome estimates.
• Shows enhanced performance with high per-read error rates.
• Confirms robustness across various error types (mismatch, insertion, deletion) and cell abundances.

Conclusions:

• The circularization approach significantly improves single-cell RNA-Seq data accuracy.
• The method is effective even with substantial barcode errors.
• A software package, Sircel, implementing this approach is publicly available.