| name | spacedust |
| description | Spacedust — de novo discovery of conserved gene clusters in microbial genomes using Foldseek structure comparison plus clustering/order-conservation statistics. Use when: (1) Finding conserved neighborhoods / modules in MAG collections, (2) Annotating dark genes via genomic context, (3) Recovering defense or other systems as clusters. Upstream: https://github.com/soedinglab/Spacedust. Route via microbial-mining / metagenomics-llm; complement defensefinder / antismash for known systems.
|
| license | MIT |
| category | analysis-tools |
| tags | ["Spacedust","gene-cluster","Foldseek","annotation","structure"] |
| upstream | https://github.com/soedinglab/Spacedust |
| stage | mining |
Spacedust
Upstream: soedinglab/Spacedust
Citation
Zhang, R., Mirdita, M. & Söding, J. De novo discovery of conserved gene clusters
in microbial genomes with Spacedust. Nat. Methods (2025).
https://doi.org/10.1038/s41592-025-02816-x
See also docs/references.md.
Analytical thinking
In prokaryotes/viruses, gene neighborhood conservation signals functional
association. Spacedust finds conserved clusters de novo with sensitive
Foldseek structure comparison and order-conservation statistics — useful
when sequence annotation covers only a fraction of genes.
Paper-scale demonstration: all-vs-all on >1k bacterial genomes recovered large
fractions of genes in conserved clusters and most PADLOC-annotated antiviral
defense clusters — still validate priority systems with specialized tools
(defensefinder, padloc, antismash, experiments).
| vs | Notes |
|---|
antismash | Known BGC rules vs de novo neighborhoods |
defensefinder | Known defense HMMs vs de novo clusters |
eggnog-mapper | Per-gene orthology vs multi-gene context |
How to run
Decision tree
Dark genes / modules in MAG sets?
├─ Conserved neighborhoods de novo → spacedust
├─ Known BGCs → antismash
├─ Known defense → defensefinder
└─ Per-gene orthology → eggnog-mapper
Related skills
foldseek · folddisco · mmseqs2 · metagenomics-llm · defensefinder ·
antismash · eggnog-mapper · deepblast · microbial-mining · tool-selection
