Overview

Cuttlefish is a fast, parallel, and very low-memory tool for constructing the compacted de Bruijn graph from sequencing reads or reference sequences. It is highly scalable in the size of the input data.

This section documents the C++ implementation — Cuttlefish 1 and Cuttlefish 2, which share one cuttlefish build executable. For the Rust implementation, see Cuttlefish 3.

Cuttlefish 1 and Cuttlefish 2

Both live in the same binary, and which one runs is selected by the arguments:

• Passing --read or --ref invokes Cuttlefish 2.
• Passing neither invokes Cuttlefish 1.

They differ in scope:

	Cuttlefish 1	Cuttlefish 2
Reference sequences	yes	yes
Sequencing reads	no	yes
FASTA output	yes	yes
GFA 1.0 / GFA 2.0 output	yes	not yet
Reduced GFA (tilings) output	yes	not yet

See Differences between 1 & 2 for the full comparison.

The papers

The work is described in two papers:

• Cuttlefish (original)
• Cuttlefish 2

Please cite the appropriate one if Cuttlefish contributed to your work.

Dependencies

If you install from Bioconda, nothing else is required. Building from source needs:

• GCC or Clang, for C++17 and C11
• CMake version 3.14 or newer
• zlib
• bzip2

These are usually already present, and otherwise available from your package manager:

# Linux
sudo apt-get install build-essential cmake zlib1g-dev libbz2-dev

# macOS
brew install --with-toolchain llvm
brew install cmake zlib bzip2