Every microbetag database, that of our server (GTDB) and one of your own genomes, has the same tables
(schema/microbetagDB.sql), and is built the same way: the pre-calculations of a
genome catalogue (the pipeline's precalc workflow), exported as one file per table with microbetag-export-db,
then loaded by MySQL. The server reads, for each network, the rows of its genomes, and computes their pathway and
seed complementarities; nothing is stored per pair of genomes.
| Table | One row per | Content |
|---|---|---|
genomeTaxonomy |
genome | its lineage; with TAXON_MATCHING=lineage, taxa are matched to the genomes sharing their deepest rank |
genome2taxNcbiId |
genome | its NCBI Taxonomy id, for taxa given as NCBI ids (the GTDB database); empty otherwise |
patricId2genomeId |
genome | the id of its GEM (PATRIC ids for the GTDB database; the genome itself otherwise) |
phenDB |
genome | phenotrex predictions: presence and score of each trait |
uniqueComplements |
alternative | a way to complete a KEGG module: the module, its missing KOs, the KOs of its path |
genomeAlternatives |
genome | the alternatives of its incomplete modules |
genomeModuleKos |
genome and module | the KOs of the module it has (for when it is the donor) |
seedSets |
GEM | its seeds, with their confidence |
nonSeedSets |
GEM | its non-seeds |
Run the precalc workflow of the pipeline (pipeline/) on your genome catalogue, then export its output, with
the taxonomy of each genome (e.g. GTDB-Tk's gtdbtk.*.summary.tsv, or a tab-separated genome and lineage):
microbetag-export-db --precalc <precalc output folder> --genome-taxonomy genome_taxonomy.tsv --outdir db_filesMySQL runs schema/microbetagDB.sql at its first start, from /docker-entrypoint-initdb.d/, with the files in
/var/lib/mysql-files/:
docker run -d --name microbetag-db \
-e MYSQL_ROOT_PASSWORD=<password> -e MYSQL_DATABASE=microbetagDB \
-e MYSQL_USER=<user> -e MYSQL_PASSWORD=<password> \
-v microbetag-db-data:/var/lib/mysql \
-v $PWD/schema/microbetagDB.sql:/docker-entrypoint-initdb.d/microbetagDB.sql:ro \
-v $PWD/db_files:/var/lib/mysql-files \
mysql:8.4or, with the server, docker compose up from server/ (services/db/data is the folder of the files, see
docker-compose.yml). The /var/lib/mysql-files mount needs to be writable, as MySQL
changes its owner at start, and on a Linux file system (not NTFS or FAT); on Fedora and other SELinux systems,
add :Z to the mounts.
MySQL loads the files only when its data folder (/var/lib/mysql, the microbetag-db-data / mysql-data volume)
is empty. To load new files, remove that volume (docker compose down -v, or docker volume rm) and start
again.
Built the same way, from the GTDB r207 representative genomes with a ModelSEED GEM (33,756 GEMs; KEGG annotations
for 16,098 of them), with the scripts of gtdb/:
run_seed_sets.py: the seed sets of the GEMs (Zenodo record 23368237);build_gtdb_precalc.py: the inputs in the layout ofprecalc's output, from the seed sets, the KEGG annotations (Zenodo record 8316356), the phenotrex predictions and the genomes' NCBI and PATRIC ids of the former database, and GTDB r207's metadata;microbetag-export-dbwith--gem-ids(the PATRIC GEM of each genome) and--ncbi-tax-ids.
The exported files are about 550 MB; loaded, about 0.5 GB on disk, and MySQL uses about 0.6 GB of memory.