Shepherd

Shepherd performs error correction of DNA barcodes to enable accurate lineage tracking by distinguishing true barcode sequences from sequencing-induced variants in next-generation sequencing (NGS) data.

Key Features:

• K-mer Indexing System: Uses k-mers (substrings of length k) derived from barcode sequences to create an index that supports grouping of similar reads for clustering.
• Bayesian Statistical Test: Applies a Bayesian test that incorporates a substitution error rate to differentiate true barcode sequences from those generated by sequencing substitutions.
• Clustering Formulation: Frames barcode error correction as a clustering problem to aggregate reads into consensus barcode sequences.
• Benchmarking Performance: In synthetic-data evaluations, produces 10 to 150 times fewer spurious lineages and yields more accurate barcode count estimates compared to state-of-the-art methods.

Scientific Applications:

• Lineage Tracking: Enables higher-resolution lineage tracking and more accurate barcode count estimates for studying evolutionary dynamics in microbial populations and monitoring disease progression such as breast cancer.
• Detection of Small-Effect Mutations: Improves sensitivity for detecting small-effect mutations by reducing spurious lineages and refining count accuracy.

Methodology:

Frames error correction as a clustering problem, constructs a k-mer index of barcode sequences, and applies a Bayesian statistical test that incorporates a substitution error rate to distinguish true barcodes from sequencing errors.