Bartender: a fast and accurate clustering algorithm to count barcode reads.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Barcode sequencing (bar-seq) is a powerful, cost-effective, high-throughput method for analyzing cell lineages and genotypes.
• Current bar-seq computational pipelines struggle with clustering accuracy, leading to under- or over-clustering artifacts.
• Applications of bar-seq are expanding, necessitating improved analytical tools for diverse biological studies.

Purpose of the Study:

• To develop an accurate and robust computational algorithm for clustering barcodes from next-generation sequencing data.
• To address the limitations of existing bar-seq clustering methods, specifically under- and over-clustering artifacts.
• To provide a user-friendly tool for precise barcode detection and abundance estimation.

Main Methods:

• Developed Bartender, a novel clustering algorithm for bar-seq data.
• Employed a modified two-sample proportion test incorporating cluster size, unlike methods relying solely on sequence similarity.
• Integrated unique molecular identifier (UMI) handling and a 'multiple time point' mode for time-course data analysis.

Main Results:

• Bartender demonstrates higher accuracy in barcode clustering compared to existing methods.
• The algorithm significantly reduces under- and over-clustering artifacts.
• Bartender offers efficient processing via simple command-line tools, suitable for laptop execution.

Conclusions:

• Bartender provides a significant advancement in bar-seq data analysis, enhancing accuracy and reducing artifacts.
• The tool's features, including UMI handling and time-course support, make it versatile for various bar-seq applications.
• Bartender is freely available, promoting wider adoption and facilitating research in genomics and evolutionary biology.