Crate cuttlefish_rs 

External-memory compacted de Bruijn graph construction.

This crate implements the Rust Cuttlefish 3 pipeline. The production path intentionally follows the phase ordering of the C++ implementation:

1. partition parses sequences and emits weak super-k-mers into atlas buckets.
2. subgraph constructs and contracts each local de Bruijn subgraph.
3. discontinuity contracts and expands the external discontinuity graph, then collates local unitigs into maximal unitigs.
4. color stores deduplicated source sets and positional color runs.

Intermediate formats are private to this implementation. Their compact layouts, ordering, and bucket fanouts are performance-sensitive; changing them requires both topology tests and matched scale benchmarks.

Modules

bgzf

Parallel decompression of block-structured (BGZF) gzip input.

buckets

External weak-super-k-mer bucket storage.

color

Positional colors and the external color repository.

colored

End-to-end colored graph construction.

discontinuity

External discontinuity-graph contraction, expansion, and unitig collation.

dna

hash

input

FASTA/FASTQ input discovery and zero-copy fragment parsing.

kmer

Packed canonical k-mer primitives.

minimizer

params

Build configuration and phase-local resource planning.

partition

Weak-super-k-mer partitioning and bucket emission.

state

Compact vertex state and colored-coordinate representations.

subgraph

Local subgraph construction and contraction.

uncolored

End-to-end uncolored graph construction.

Enums

GraphInput

Semantic class of sequence input.

Side

Side of a canonical k-mer vertex.

Constants

DEFAULT_ATLAS_COUNT

Number of atlases in the C++-compatible default partition layout.

DEFAULT_CUTOFF_READS

Default edge-frequency cutoff for sequencing-read input.

DEFAULT_CUTOFF_REFS

Default edge-frequency cutoff for reference-sequence input.

DEFAULT_GMTIG_BUCKETS

Default number of maximal-unitig coordinate buckets.

DEFAULT_GRAPHS_PER_ATLAS

Number of subgraphs buffered by each atlas.

DEFAULT_K

Default k-mer length used by the CLI.

DEFAULT_LMTIG_BUCKETS

Default number of local-unitig buckets.

DEFAULT_MINIMIZER_LEN

Default minimizer length used to partition k-mers.

DEFAULT_SUBGRAPH_COUNT

Total number of local subgraphs in the default partition layout.

DEFAULT_VERTEX_PARTITIONS

Number of vertex partitions in the external discontinuity graph.

MAX_K

Largest k-mer length supported by the packed two-word representation.

MAX_MINIMIZER_LEN

Largest supported minimizer length.

Functions

configure_global_parallelism

Configures the process-wide Rayon pool for code paths that use it.

default_threads

Returns the default worker count, matching Cuttlefish’s quarter-machine policy.

default_work_dir

Returns the platform temporary directory used for intermediate files.