Discount

Discount performs k-mer counting and minimizer-ordering analysis to enable scalable, memory-efficient processing of large metagenomic sequencing datasets.

Key Features:

• Distributed k-mer counting: Implements scalable, out-of-core k-mer counting for large datasets using Apache Spark and distributed computing.
• Minimizer ordering analysis: Evaluates the impact of different minimizer orderings and addresses uneven bin sizes produced by traditional minimizer-based approaches.
• Universal Frequency Ordering: Combines frequency-sampled minimizers with universal k-mer hitting sets to produce evenly distributed, smaller bins.
• Memory efficiency: Demonstrates reduced memory requirements for distributed k-mer counting, with reported reductions in memory usage of up to 87% versus existing methods.

Scientific Applications:

• Genome size estimation: Provides k-mer frequency counts useful for estimating genome size from sequencing data.
• Genome assembly: Supplies scalable k-mer counting and balanced binning that support de novo assembly workflows.
• Taxonomic classification: Enables k-mer–based taxonomic classification analyses in metagenomic studies through efficient counting.
• Metagenomic community analysis: Facilitates large-scale analysis of microbial communities by improving throughput and memory efficiency of k-mer processing.

Methodology:

Discount uses an Apache Spark-based distributed implementation and empirically evaluates various minimizer orderings on metagenomic datasets, including a universal frequency ordering that combines frequency-sampled minimizers with universal k-mer hitting sets to assess effects on bin distribution and memory usage.