Petri-DB: Cyanobacteria reference (v3)

v3 update (2026-06-03): MMseqs2 databases now bundle a precomputed prefilter k-mer index (built with mmseqs createindex) for fast search. In v2 the index was rebuilt on every search, which made MMseqs2 slower than BLAST for the same query. BLAST databases are unchanged from v2. Because of the bundled index, the MMseqs2 tarballs are now ~14–17 GB to download (~23–28 GB extracted).
Cyanobacteria phylum reference databases for BLAST and MMseqs2 homology search, derived from two complementary sources (NCBI RefSeq and GTDB r220).

v2 update (2026-05-31):

NCBI source expanded from 257 → 2,075 strain assemblies (1 best per strain taxid). 

Now includes Synechocystis sp. PCC 6803, S. elongatus PCC 7942, and many more.

All BLAST DBs are taxonomy-aware (built with `-taxid_map`).

Enables `blastp/blastn -taxids <id>` to pre-filter by NCBI taxid at search time.

All MMseqs2 DBs are taxonomy-aware (built with `mmseqs createtaxdb`).

Enables `mmseqs search --taxon-list <id>` to pre-filter by NCBI taxid.

FASTA headers use `accession__seqid` separator (changed from `|` to avoid BLAST taxid_map parsing collision).

GTDB source content unchanged from v1, rebuilt only to add taxonomy embedding.

Build parameters: BLAST 2.16.0+, MMseqs2 18.8cc5c, snapshot 2026-05-31.

Required at search time for -taxids/--taxon-list: NCBI taxdb.