Code Navigation¶
Bijux Proteomics Core is the scientific heart of the suite. Navigate by the scientific question, not by searching the package root: the implementation is divided into explicit domain families spanning raw evidence, identification, quantification, study design, PTMs, interpretation, review, and workflow artifacts.
Domain map¶
| Question | Start here | What the family owns |
|---|---|---|
| What constitutes a program, target, gate, or valid progression? | domain/ |
scientific program state, records, confidence, semantic identifiers, targets, and validation |
| How are proteins and peptides represented? | sequences/ and chemistry/ |
FASTA, digestion, protein identity, masses, modifications, fragments, isotopes, and chemical liabilities |
| How does instrument or tabular evidence enter? | io/ |
governed formats, tables, spectra, chromatograms, raw-signal evidence, and input integrity |
| How do search-engine results become comparable? | identification/adapters/ and identification/search_adapters/ |
engine dialects, normalization, field accounting, provenance, conformance, and loss |
| How are PSM, peptide, and protein claims controlled? | identification/contracts/, psm/, peptide/, protein/, and fdr/ |
evidence levels, target-decoy error control, grouping, parsimony, ambiguity, and coverage |
| How are abundance matrices built and tested? | quantification/ |
rollup, normalization, missingness, statistics, contributor decomposition, and provenance |
| How is experimental design represented? | study/ |
sample metadata, contrasts, batches, replicate structure, and design validity |
| How are PTM and DIA results handled? | ptm/, dia/, isotope_labeling/, and multiplex/ |
site parsing and localization, PTM quantification, precursor evidence, labeling, and channel-aware analysis |
| How are biological outputs bounded? | interpretation/, biology/, panels/, proteoforms/, and targeted/ |
enrichment, activity, networks, contextual annotations, assays, and validation planning |
| How does evidence reach a reviewer? | review/ and workflow/ |
claims, belief and evidence graphs, cards, reports, pipelines, governed exports, and benchmark studies |
Public entry routes¶
Use interfaces/python_api/ for application composition and interfaces/cli/
for commands. Both should terminate in the same owning scientific families.
interfaces/execution/ is the narrow seam for executing a validated program
through a supplied backend; it is not a second scientific implementation.
The root package contains compatibility facades as well as owned modules. When
a file says it is a compatibility facade, follow its import to the canonical
subfamily before changing behavior. Use governance/charter.py to confirm the
eight permitted ownership families and identify logic that belongs in runtime,
knowledge, intelligence, or lab instead.
Trace one result backwards¶
Begin at the rendered workflow report or table, identify its typed report, follow referenced claims and provenance into interpretation or quantification, then continue to identification and imported source rows. Inspect the study design and policy objects at each transformation. This backwards route exposes loss, ambiguity, thresholds, and rejected evidence that a top-level command alone cannot explain.