KMC

KMC counts k-mers in large Next Generation Sequencing (NGS) datasets to produce k-mer occurrence histograms for downstream analyses such as de Bruijn graph assembly and repeat detection.

Key Features:

• K-mer counting: Counts k-mers (consecutive k nucleotide symbols) and constructs histograms of k-length substring occurrences.
• Disk-based processing: Employs a disk-based approach to handle large-scale sequencing datasets beyond main memory limits.
• Signatures instead of minimizers: Uses signatures, a carefully selected subset of all possible minimizers, replacing traditional minimizers.
• (k,x)-mers representation: Utilizes (k,x)-mers to reduce input/output operations.
• Parallel architecture: Implements a highly parallel processing architecture to achieve high throughput.
• Performance: Reported to run approximately twice as fast as Jellyfish 2 and KMC 1 while maintaining moderate memory usage (typically around 12 GB or less).
• Benchmark example: Processed 28-mers of a human reads collection with 44-fold coverage (≈106 GB compressed) in about 20 minutes on a 6-core Intel i7 with a solid-state disk.
• NGS data compatibility: Targets high-volume outputs produced by Next Generation Sequencing technologies.

Scientific Applications:

• Histogram construction: Produces k-mer frequency histograms used for sequence analysis and quality assessment.
• De Bruijn graph assembly: Provides k-mer counts foundational for constructing de Bruijn graph-based genome assemblers.
• Multiple sequence alignment: Facilitates fast multiple sequence alignment workflows that rely on k-mer information.
• Repeat detection: Enables detection and analysis of repetitive elements within genomic data.

Methodology:

Computational methods include a disk-based counting scheme, use of signatures (a subset of minimizers) in place of traditional minimizers, representation with (k,x)-mers to reduce I/O, and a highly parallel processing architecture.