fix: allow reordered subfamily member records

This commit is contained in:
wtx
2026-08-06 21:32:49 +08:00
parent 4750bcc76f
commit 8c3d601072
4 changed files with 83 additions and 4 deletions
+3 -3
View File
@@ -118,7 +118,7 @@ def _atomic_promote_pair(curated_id: Path, curated_pep: Path, final_id: Path, fi
tmp_pep.write_text(curated_pep.read_text(encoding="utf-8", errors="ignore"), encoding="utf-8")
tmp_ids = _read_id_list(tmp_id)
tmp_fasta_ids = _read_fasta_ids(tmp_pep)
if tmp_ids != tmp_fasta_ids:
if sorted(tmp_ids) != sorted(tmp_fasta_ids):
tmp_id.unlink(missing_ok=True)
tmp_pep.unlink(missing_ok=True)
raise RuntimeError(f"Temporary final member validation failed for {final_id.stem}")
@@ -160,7 +160,7 @@ def _promote_curated_members_to_final(final_dir: Path, prefixes: Iterable[str],
if not curated_ids:
raise ValueError(f"Curated member ID file is empty: {curated_id}")
fasta_ids = _read_fasta_ids(curated_pep)
if curated_ids != fasta_ids:
if sorted(curated_ids) != sorted(fasta_ids):
raise ValueError(
f"Curated member ID/FASTA mismatch for {member_prefix}: "
f"{len(curated_ids)} IDs vs {len(fasta_ids)} FASTA records"
@@ -179,7 +179,7 @@ def _promote_curated_members_to_final(final_dir: Path, prefixes: Iterable[str],
promoted_ids = _read_id_list(final_id)
promoted_fasta_ids = _read_fasta_ids(final_pep)
if promoted_ids != curated_ids or promoted_fasta_ids != curated_ids:
if sorted(promoted_ids) != sorted(curated_ids) or sorted(promoted_fasta_ids) != sorted(curated_ids):
raise RuntimeError(f"Post-promotion final member validation failed for {member_prefix}")
promotions.append({
"prefix": member_prefix,
+1 -1
View File
@@ -259,7 +259,7 @@ def _build_result(
id_list = _read_ids(path_id) if path_id else []
pep_ids = _read_fasta_ids(path_pep) if path_pep else []
if path_id and path_pep and id_list != pep_ids:
if path_id and path_pep and sorted(id_list) != sorted(pep_ids):
raise MemberResolutionError(f"Member ID/FASTA mismatch: {path_id} vs {path_pep}")
member_count = len(id_list or pep_ids)
warnings = []
@@ -116,6 +116,49 @@ def test_pre_confirm_prefers_curated_before_identified(tmp_path: Path):
assert result.member_count == 2
def test_pre_confirm_accepts_same_curated_members_in_different_order(tmp_path: Path):
ws = tmp_path / "workspace" / "job1"
final_dir = ws / "final"
input_dir = tmp_path / "input" / "job1"
prefix = "Target.Sl.GRAS"
_write(final_dir / f"{prefix}.curated_members.id", "gene1\ngene2\n")
_write(final_dir / f"{prefix}.curated_members.pep.fa", _pep(["gene2", "gene1"]))
result = resolve_effective_members(
workspace=ws,
final_dir=final_dir,
input_dir=input_dir,
prefix=prefix,
phase="pre_confirm",
require="pair",
)
assert result.member_count == 2
assert result.path_id == final_dir / f"{prefix}.curated_members.id"
assert result.path_pep == final_dir / f"{prefix}.curated_members.pep.fa"
def test_pre_confirm_rejects_different_curated_members_with_equal_counts(tmp_path: Path):
ws = tmp_path / "workspace" / "job1"
final_dir = ws / "final"
input_dir = tmp_path / "input" / "job1"
prefix = "Target.Sl.GRAS"
_write(final_dir / f"{prefix}.curated_members.id", "gene1\ngene2\n")
_write(final_dir / f"{prefix}.curated_members.pep.fa", _pep(["gene1", "gene3"]))
with pytest.raises(MemberResolutionError, match="ID/FASTA mismatch"):
resolve_effective_members(
workspace=ws,
final_dir=final_dir,
input_dir=input_dir,
prefix=prefix,
phase="pre_confirm",
require="pair",
)
def test_subfamily_evidence_required_rejects_unreviewed_identified_members(tmp_path: Path):
ws = tmp_path / "workspace" / "job1"
final_dir = ws / "final"
@@ -1,5 +1,7 @@
from pathlib import Path
import pytest
from apps.cli.run_subfamily_confirm import _promote_curated_members_to_final
@@ -25,3 +27,37 @@ def test_promote_curated_members_to_final_records_checksums(tmp_path: Path):
assert promotion["source_curated_id_checksum"]
assert (final_dir / f"{prefix}.final_members.id").read_text(encoding="utf-8").splitlines() == curated_ids
assert (final_dir / f"{prefix}.final_members.pep.fa").exists()
def test_promote_curated_members_accepts_same_members_in_different_order(tmp_path: Path):
final_dir = tmp_path / "final"
final_dir.mkdir()
prefix = "Target.Sl.GRAS"
(final_dir / f"{prefix}.curated_members.id").write_text("gene1\ngene2\n", encoding="utf-8")
curated_pep = ">gene2\nMT\n>gene1\nMA\n"
(final_dir / f"{prefix}.curated_members.pep.fa").write_text(curated_pep, encoding="utf-8")
labels = {
"gene1": {"confirmed_subfamily": "GRAS-a"},
"gene2": {"confirmed_subfamily": "GRAS-b"},
}
promotions = _promote_curated_members_to_final(final_dir, [prefix], labels)
assert promotions[0]["count"] == 2
assert (final_dir / f"{prefix}.final_members.id").read_text(encoding="utf-8") == "gene1\ngene2\n"
assert (final_dir / f"{prefix}.final_members.pep.fa").read_text(encoding="utf-8") == curated_pep
def test_promote_curated_members_rejects_different_members_with_equal_counts(tmp_path: Path):
final_dir = tmp_path / "final"
final_dir.mkdir()
prefix = "Target.Sl.GRAS"
(final_dir / f"{prefix}.curated_members.id").write_text("gene1\ngene2\n", encoding="utf-8")
(final_dir / f"{prefix}.curated_members.pep.fa").write_text(">gene1\nMA\n>gene3\nMT\n", encoding="utf-8")
labels = {
"gene1": {"confirmed_subfamily": "GRAS-a"},
"gene2": {"confirmed_subfamily": "GRAS-b"},
}
with pytest.raises(ValueError, match="ID/FASTA mismatch"):
_promote_curated_members_to_final(final_dir, [prefix], labels)