Shepherd: accurate clustering for correcting DNA barcode errors

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• DNA barcodes are essential for tracking cell lineages in biological systems.
• Next-generation sequencing introduces errors, complicating accurate barcode identification.
• Existing methods struggle with the scale and error rates of barcode sequencing data.

Purpose of the Study:

• To develop a robust method for correcting errors in DNA barcode sequences.
• To improve the accuracy of lineage tracking using noisy sequencing data.
• To enable more precise estimation of biological quantities from lineage data.

Main Methods:

• Framed barcode error correction as a clustering problem.
• Developed Shepherd, a novel clustering method utilizing k-mer indexing.
• Incorporated a Bayesian statistical test with a substitution error rate.

Main Results:

• Shepherd significantly outperforms state-of-the-art methods in accuracy.
• Reduced spurious lineages by 10-150 times on synthetic data.
• Demonstrated consistent improvements on empirical data, enabling higher resolution lineage tracking.

Conclusions:

• Shepherd provides a significant advancement in DNA barcode error correction.
• The method enhances the accuracy of lineage tracking and biological quantity estimation.
• Enables detection of small effect mutations and improves understanding of evolutionary dynamics.