Skip to content

Bacterial Surface Glycan & Biosynthetic Locus Curation

This presentation documents the multi-database streaming, chemical graph canonicalization, taxonomy normalization, and 4-tier NCBI/PubMed SeroGen biosynthetic locus pairing pipeline in kbank.io.sources and kbank.db.curation, producing the unified Structure-of-Arrays GlycanRecords archive (src/kbank/data/glycans.npz).

Key Takeaways

  • 5 Integrated Surface-Antigen Databases: Consolidates CSDB, KEGG Glycan, ECODAB, CERMAV EPS-DB, and EK3D with GlyTouCan accession cross-references into 1,530 non-redundant bacterial surface polysaccharide graphs (7,788 residue/modifier nodes and 7,660 glycosidic linkages across 154 species).
  • De Novo SeroGen Locus Pairing: Resolves NCBI Genome Assemblies and nucleotide BGC accessions via cassetter-cached Entrez queries, segments candidate polysaccharide loci with LocusPredictor, and aligns 7,940 classified BGC genes across 565 paired records via BiosyntheticAssigner.
  • Zero-Pickle SoA Archive: Compiles all graphs, loci, and bipartite GeneGlycanAssignments into src/kbank/data/glycans.npz (allow_pickle=False).

Run Locally

# Present as interactive slideshow
uv run --with marimo --with altair marimo run scripts/curate_glycans.py

# Edit notebook cells
uv run --with marimo --with altair marimo edit scripts/curate_glycans.py

Comments