Skip to content

Protein FASTA

protein-fasta provides the shared protein-FASTA and reusable database boundary for APB and fasta_gen.

It provides:

  • streaming parsing of plain, gzip, and bzip2 FASTA files;
  • minimal normalized Python records and compositional diagnostics;
  • streaming aggregate diagnostic summaries;
  • independent, configurable decoy, contaminant, entrapment, and sentinel labels;
  • an exact three-column base Polars frame;
  • config-driven homogeneous UniProtKB and RefSeq enrichment;
  • versioned fast sequence/peptide hashes and exact-file checksums;
  • UniProt catalog synchronization and reproducible proteome FASTA acquisition;
  • separate source preparation, biological assembly, and subsequent decoy generation;
  • canonical peptide inventories, mappings, FASTA products, and exact comparisons;
  • SQLite and DuckDB registry indexing, pair metrics, comparison fingerprints, and clustering; and
  • optional table, build, registry, and analytical commands through Cyclopts.

The package deliberately excludes AnnData/MuData persistence, protein inference, application installation and authorization, and the site-curated contaminant/QC catalog.

For a first run, use the concise direct CLI. Request-driven commands write validated authored JSON before computation; replay that record explicitly with --request.

Start with the build workflows, then continue with the API reference, artifact contracts, rule configuration, CLI guide and executable walkthrough, architecture, or benchmarks.