CheckAMG database

Databases required to run CheckAMG. The annotation module ("checkamg annotate") and the de-novo module ("checkamg denovo") use separate databases, distributed as two independent archives, so the larger de-novo database is only downloaded when needed.

• CheckAMG_annotate_db_v1.1_20260316.tar.gz: profile HMM databases and their associated score cutoff files required by "checkamg annotate". Contains:
  • *.hmm, *.h3m, *.h3i, *.h3f, and *.h3p files for each of KEGG, FOAM, Pfam-A, PHROGs, dbCAN (dbCAN_HMMdb_v14), METABOLIC_custom, and CAMPER
  • KEGG_cutoffs.tsv: score cutoffs for KEGG KOfam HMMs
  • FOAM_cutoffs.tsv: score cutoffs for FOAM HMMs
  • CAMPER_cutoffs.tsv: score cutoffs for CAMPER HMMs
  • METABOLIC_cutoffs.tsv: score cutoffs for METABOLIC_custom HMMs
  • README.txt: dependent database versions and citations
• CheckAMG_denovo_db_v1_20260605.tar.gz: trained Protein Set Transformer (PST) model and precomputed reference data required by "checkamg denovo". Contains:
  • checkAMG-PST_TL-P__large_4.20260605.ckpt: trained CheckAMG-PST model checkpoint
  • checkAMG-PST_TL-P__large_4.20260605.PST-EMBED.h5: reference protein PST embeddings (with labels)
  • checkAMG-PST_TL-P__large_4.20260605.PST-EMBED.index.faiss: FAISS index over the reference embeddings for nearest-neighbor search
  • checkAMG-PST_TL-P__large_4.20260605.PST-EMBED.labels.h5: reference protein labels
  • README.txt: model/database version and citation