Catalogues¶

A catalogue set annotates fields of one kind of meaning, for every rule APB2 packages; data/catalogues.json states each set's purpose and users, and Catalog.description returns it. The JSON files under src/apb_catalog/data/sources/<set>/ are the source of truth; this page summarises them by hand.

identification_confidence¶

How likely each identification, or the signal quantified for it, is correct; used by apb-aggregate for confidence-weighted rollups and by apb-export's prolfqua target: per-run PEP and q-value layers, per-feature PEP, Sage's MS1-peak q-value (stage quantification), and DIA-NN's, Spectronaut's and MaxQuant's protein-group q-values. Scores, vendor summaries and other vendors' protein-level confidence are deliberately left out.

Concept confidence, kinds:

• pep: Posterior error probability: the probability that this single identification is wrong; lower is better.
• q_value: The smallest false discovery rate at which this identification, or quantified signal, is accepted; lower is better.

proteobench_entrapment¶

The precursor q-values ProteoBench's entrapment scoring ranks by; used by apb-proteobench's EntrapmentAnalyzer and run entrapment. Each q-value is its own kind, so the scorer can score every one a result offers. Only the vendors ProteoBench's entrapment module accepts carry entries.

Concept confidence, kinds:

• q_value: The smallest false discovery rate at which this identification, or quantified signal, is accepted; lower is better.
• library_q_value: The q-value of the identification's spectral-library entry, built across all runs; lower is better.
• global_q_value: The q-value of the identification across the whole experiment; lower is better.

miape¶

Draft mapping onto the MIAPE-AnnData Schema 0.4.0 draft of the HUPO-PSI AI Readiness Working Group. MIAPE-AnnData defines protein-group, peptide, peptidoform and site modalities; APB ion levels have no modality, so ion-only results gain MIAPE fields only after aggregation. Fields an exporter computes, such as detection_rate or length, are not catalogued. MIAPE-AnnData stores no per-feature confidence; it records FDR filtering as a provenance step.

Concept miape, kinds and their requirement level:

• protein_group: var, MUST: protein-group identifier as semicolon-separated UniProt accessions; on peptide and peptidoform levels the foreign key to the protein modality.
• gene_name: var, SHOULD: gene symbol of the protein group.
• organism: var, SHOULD: species of the protein group.
• protein_description: var, MAY: free-text protein description, typically the UniProt protein name.
• n_peptides_used: var, SHOULD: distinct peptide sequences aggregated into the protein group.
• is_contaminant: var, SHOULD: whether the protein group comes from a contaminant database.
• sequence: var, MUST: bare amino-acid sequence without modifications or charge.
• peptidoform: var, MUST: ProForma 2.0 peptidoform, charge-collapsed.
• is_unique: var, MUST: whether the peptide or peptidoform maps to exactly one protein group.
• n_psms: var, MAY: peptide-spectrum matches aggregated into the row across all samples.
• raw: layers, MUST: the unmodified engine-output quantification matrix.