MUSET: set of utilities for constructing abundance unitig matrices from sequencing data

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• K-mer matrices are widely used for analyzing large sequencing datasets but can be inefficient.
• Existing tools often represent k-mer presence/absence, losing valuable abundance information.
• There is a need for methods that efficiently represent sequence variation and abundance in large-scale genomic studies.

Purpose of the Study:

• To introduce MUSET, a novel software utility for constructing abundance unitig matrices.
• To overcome limitations of existing k-mer based approaches by integrating k-mer counting and unitig extraction.
• To provide a method for preserving sample variations while reducing computational resource demands.

Main Methods:

• MUSET integrates k-mer counting and unitig extraction to generate unitig matrices.
• Overlapping k-mers are merged to form unitigs, preserving sequence information.
• Abundance values are recorded for unitigs, unlike presence-absence in traditional k-mer matrices.

Main Results:

• MUSET successfully generated a filtered unitig matrix from a large ancient oral sequencing dataset (618 GB).
• The process was completed in under 10 hours using only 20 GB of memory.
• The resulting unitig matrix preserved sample variations while significantly reducing disk space and row count compared to k-mer matrices.

Conclusions:

• MUSET is an efficient tool for constructing abundance unitig matrices from sequencing data.
• Unitig matrices offer a space-saving and informative alternative to k-mer matrices for large-scale genomic analysis.
• MUSET demonstrates practical applicability for analyzing massive sequencing datasets, such as ancient DNA.