Gerbil: A Fast and Memory-Efficient $k$-mer Counter with GPU-Support

TL;DR

Gerbil is an open-source, GPU-accelerated $k$-mer counting software optimized for large $k$, offering efficient performance and low memory usage in genome analysis, especially for long-read sequencing data.

Contribution

Gerbil introduces a two-disk approach combined with optional GPU support to efficiently count large $k$-mers, outperforming existing tools in speed and memory usage.

Findings

Outperforms state-of-the-art $k$-mer counters for large genomes

Supports large $k$ (≥32) efficiently with minimal performance loss

Uses a two-disk approach and GPU acceleration for improved scalability

Abstract

A basic task in bioinformatics is the counting of $k$-mers in genome strings. The $k$-mer counting problem is to build a histogram of all substrings of length $k$ in a given genome sequence. We present the open source $k$-mer counting software Gerbil that has been designed for the efficient counting of $k$-mers for $k\geq32$. Given the technology trend towards long reads of next-generation sequencers, support for large $k$ becomes increasingly important. While existing $k$-mer counting tools suffer from excessive memory resource consumption or degrading performance for large $k$, Gerbil is able to efficiently support large $k$ without much loss of performance. Our software implements a two-disk approach. In the first step, DNA reads are loaded from disk and distributed to temporary files that are stored at a working disk. In a second step, the temporary files are read again, split into $k$-mers and counted via a hash table approach. In addition, Gerbil can optionally use GPUs to accelerate the counting step. For large $k$, we outperform state-of-the-art open source $k$-mer counting tools for large genome data sets.