candidate_summary_card validator accepts lowercase sequences but its test requires rejection (code-vs-test conflict) #842

Summary

tests/evidence/test_candidate_summary_card.py::test_lowercase_sequence_fails fails on main. It asserts a lowercase sequence must fail validation (assert not result.passed), but validate_candidate_summary_card passes it.

This was hidden until now because CI's pytest step aborted at collection (fixed in #841). This test fails on unmodified main.

Root cause

src/openamp_foundry/evidence/candidate_summary_card.py:79:

invalid_chars = set(entry.sequence.upper()) - VALID_AMINO_ACIDS

The validator uppercases the sequence before checking it against VALID_AMINO_ACIDS, so a lowercase sequence like "kwklfkkiek" normalizes to valid characters and passes. The test expects the opposite.

Why this needs a decision (not a blind fix)

There is a genuine conflict about intended behavior:

• The validator's own code (.upper()) is written to accept/normalize lowercase.
• Its own test demands rejection.
• src/openamp_foundry/qc/presynth_check.py:208 also does seq.upper().strip() — i.e. normalizes (accepts) lowercase elsewhere in the codebase.
• docs/evidence/GUIDE_SEQUENCE_VALIDATION.md:9 states "Sequence is uppercase."

So either:

• (a) the validator should reject non-uppercase input (add an explicit entry.sequence == entry.sequence.upper() check) — matches the test and the guide, but diverges from presynth_check's normalize-and-accept behavior; or
• (b) the test is stale and lowercase should be accepted/normalized — matches the code and presynth_check, but contradicts the guide.

Both are defensible. Picking one changes validation semantics, so it needs an owner's call, not an agent guess.

Repro

pip install -e ".[dev]"
python -m pytest tests/evidence/test_candidate_summary_card.py::test_lowercase_sequence_fails

Introduced in #820 (Phase L L3 candidate summary card).