Reference Data

Reference Data

The classifier uses the full 562-genome SAR11_165 reference alignment and tree, comprising 542 SAR11 genomes and 20 non-SAR11 outgroups:

• data/classifier/msa/sar11.concatenated_trimmed.faa;
• data/classifier/sar11.concatenated_trimmed.faa.raxml.bestTree;
• data/classifier/sar11.concatenated_trimmed.faa.raxml.bestModel;
• data/classifier/marker_order.tsv;
• data/classifier/sar11.partitions.txt;
• data/classifier/trusted_nodes.tsv;
• data/classifier/reference_taxonomy.tsv;
• data/classifier/reference_taxonomy_metadata.json;
• data/classifier/reference_genomes.tsv;
• data/classifier/placement_neighbors.tsv.

Runtime archives are published in the SGAtools datasets Zenodo record:

• SAR11_Orthogroups_4577.hmm.tar.gz for mapper OG HMMs;
• SAR11_165_marker_hmms.tar.gz for the 165 marker HMM set;
• fna_files.tar.gz for fastANI reference genomes.

The 20 non-SAR11 outgroups remain in the full tree and alignment so that placement can distinguish SAR11, Clade IV-related sequences, and non-SAR11 sequences.

reference_taxonomy_metadata.json records the Atlas table source, version, retrieval date, record count, and source CSV hash. sga-db validate and classifier startup verify that reference_taxonomy.tsv and reference_genomes.tsv contain matching genome IDs and shared taxonomy.

Downloaded archives and extracted runtime assets are cached under:

~/.cache/sga-tools/

Set SGA_CACHE_DIR to relocate the cache. Existing files are reused and verified; downloads occur only when an asset is absent. To prefetch assets:

sga-db download
sga-db download-marker-hmm

If no local archive is available, SGAtools uses the configured Zenodo URL.