feat(annoont): AnnotationOntology.from_prd_input + fix msbuilder get_feature lookup

[jplfaria]

Two related changes that unblock the modelseed-api bulk-reconstruction endpoint (Phase 3 per the PRD).

Changes

1. New factory AnnotationOntology.from_prd_input

Mirrors from_kbase_data but for inputs whose ontology terms have NOT been pre-translated to ModelSEED reaction IDs.

@staticmethod
def from_prd_input(
    genome_id,
    annotations,        # {gene_id: {ontology_type: [{term, score}, ...]}}
    data_dir,
    translator,         # Callable[[namespaced_term: str], list[msrxn_id: str]]
    method="PRD",
    method_version="1.0",
    timestamp=None,
)

The translator callable is injected so the factory stays decoupled from any specific translation backend. The canonical impl is KBUtilLib.KBAnnotationUtils.translate_term_to_modelseed; tests use a small in-memory fake (no data files needed).

Synthesizes one AnnotationOntologyEvent per ontology type, keeps priority-list logic untouched. Unmapped terms (translator returns []) are retained with an empty msrxns set per the PRD requirement that unmapped genes never silently disappear.

2. Fix latent bug at msbuilder.py:789 in build_from_annotaton_ontology

The line called anno_ont.get_feature(gene.id) but AnnotationOntology has no such method (features are keyed in genes or cdss dicts). Any call path that reached this line would AttributeError the moment it tried to attach evidence to a built reaction. Replaced with the correct accessor:

annoont_gene = anno_ont.genes.get(gene.id) or anno_ont.cdss.get(gene.id)

Tests

9 unit tests in tests/core/test_annotationontology.py:

• 7 cases for from_prd_input: happy path, multi-gene/multi-ontology, score recording, default score, unmapped retention, namespaced-term passing, empty input
• 2 regression tests for the msbuilder fix: one positive (the accessor works), one negative lock (asserts AnnotationOntology still has no get_feature method, so if it ever gets added back the msbuilder accessor must be reconciled)

All 9 pass locally.

Context

This branch is the upstream half of Phase 3 (bulk reconstruction endpoint) in ModelSEED/modelseed-api. The endpoint's per-genome loop calls MSBuilder.build_from_annotaton_ontology directly, so the msbuilder.py:789 bug is a hard prereq for that work. from_prd_input is the entry point into AnnotationOntology for the PRD's {gene_id: {ontology_type: [{term, score}]}} input shape, replacing the KBase-shaped from_kbase_data path on this code path.

🤖 Generated with Claude Code