SAKEIMA

SAKEIMA approximates frequencies of the most frequent k-mers in high-throughput sequencing datasets to enable efficient k-mer-based analyses.

Key Features:

• Sampling-Based Approach: Employs a sampling strategy to approximate frequent k-mers and their frequencies without processing entire datasets.
• Rigorous Quality Guarantees: Leverages statistical learning theory and the Vapnik-Chervonenkis (VC) dimension to provide theoretical bounds on approximation quality and sample size.
• Efficiency for Large Datasets: Processes only a fraction of the data to achieve substantial speedups in analyzing large-scale sequencing data.
• Accurate Distance Estimation: Produces k-mer frequency estimates that enable precise calculation of k-mer-based distances between datasets.

Scientific Applications:

• Metagenomics: Facilitates computation of k-mer-based distances among metagenomic datasets for comparative analyses of microbial communities.
• Genome Assembly and Annotation: Identifies frequent k-mers that can inform genome assembly steps and support functional annotation tasks.
• Comparative Genomics: Enables comparison of genomic sequences across species or strains by providing approximate k-mer frequency profiles.

Methodology:

Uses a sampling-based algorithm informed by statistical learning theory and the Vapnik-Chervonenkis (VC) dimension to determine practical sample size bounds and approximate frequent k-mer frequencies.