fix: complete PPI and KaKs reference handoff

2026-08-10 23:13:05 +08:00
parent 39a2c969de
commit 7b7e334213
37 changed files with 4770 additions and 561 deletions
@@ -0,0 +1,93 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+from openpyxl import Workbook
+
+from apps.web.vue3 import main
+
+
+def _write_config(input_dir: Path, pep: Path, cds: Path) -> None:
+    workbook = Workbook()
+    sheet = workbook.active
+    sheet.title = "config"
+    sheet.append(
+        [
+            "Species_Code",
+            "Species_Role",
+            "File_Type",
+            "Original_File_Name",
+            "Ref",
+        ]
+    )
+    sheet.append(["At", "reference", "pep", pep.name, True])
+    sheet.append(["At", "reference", "cds", cds.name, True])
+    workbook.save(input_dir / "config.xlsx")
+
+
+def test_execute_validation_accepts_config_resolved_reference_pep_and_cds(
+    tmp_path: Path,
+    monkeypatch,
+) -> None:
+    input_root = tmp_path / "input"
+    job_input = input_root / "job_at"
+    external = tmp_path / "external"
+    job_input.mkdir(parents=True)
+    external.mkdir()
+    pep = external / "Arabidopsis_thaliana.TAIR10.pep.all.fa"
+    cds = external / "Arabidopsis_thaliana.TAIR10.cds.all.fa"
+    pep.write_text(">AT1G01010\nMA\n", encoding="utf-8")
+    cds.write_text(">AT1G01010\nATGGCT\n", encoding="utf-8")
+    _write_config(job_input, pep, cds)
+    monkeypatch.setattr(main, "INPUT_ROOT", input_root)
+    monkeypatch.setenv("PLANTAI_REF_DATA_ROOT", str(external))
+
+    available, source = main._reference_bundle_species_available(
+        "job_at",
+        "At",
+        require_cds=True,
+    )
+
+    assert available is True
+    assert source == str(job_input)
+
+
+def test_execute_validation_keeps_accepting_downloaded_reference_bundle(
+    tmp_path: Path,
+    monkeypatch,
+) -> None:
+    input_root = tmp_path / "input"
+    species_dir = input_root / "job_spinach" / "string_taxon_3562"
+    species_dir.mkdir(parents=True)
+    (species_dir / "string_taxon_3562.pep.fasta").write_text(">S1\nMA\n", encoding="utf-8")
+    (species_dir / "string_taxon_3562.cds.fasta").write_text(">S1\nATGGCT\n", encoding="utf-8")
+    monkeypatch.setattr(main, "INPUT_ROOT", input_root)
+
+    available, source = main._reference_bundle_species_available(
+        "job_spinach",
+        "string_taxon_3562",
+        require_cds=True,
+    )
+
+    assert available is True
+    assert source == str(species_dir)
+
+
+def test_execute_validation_still_rejects_a_truly_incomplete_reference(
+    tmp_path: Path,
+    monkeypatch,
+) -> None:
+    input_root = tmp_path / "input"
+    species_dir = input_root / "job_incomplete" / "At"
+    species_dir.mkdir(parents=True)
+    (species_dir / "At.pep.fasta").write_text(">AT1G01010\nMA\n", encoding="utf-8")
+    monkeypatch.setattr(main, "INPUT_ROOT", input_root)
+
+    available, reason = main._reference_bundle_species_available(
+        "job_incomplete",
+        "At",
+        require_cds=True,
+    )
+
+    assert available is False
+    assert "missing CDS" in reason