NegBioDB / tests /test_ppi_db.py
jang1563's picture
NegBioDB final: 4 domains, fully audited
6d1bbc7
"""Tests for NegBioDB PPI domain database layer."""
import sqlite3
from pathlib import Path
import pytest
from negbiodb_ppi.ppi_db import (
create_ppi_database,
get_connection,
refresh_all_ppi_pairs,
run_ppi_migrations,
)
MIGRATIONS_DIR = Path(__file__).resolve().parent.parent / "migrations_ppi"
@pytest.fixture
def ppi_db(tmp_path):
"""Create a temporary PPI database with all migrations applied."""
db_path = tmp_path / "test_ppi.db"
run_ppi_migrations(db_path, MIGRATIONS_DIR)
return db_path
class TestPPIMigrations:
"""Test PPI schema creation and migrations."""
def test_create_ppi_database(self, tmp_path):
db_path = tmp_path / "test.db"
result = create_ppi_database(db_path, MIGRATIONS_DIR)
assert result == db_path
assert db_path.exists()
def test_migration_001_applied(self, ppi_db):
conn = get_connection(ppi_db)
try:
versions = conn.execute(
"SELECT version FROM schema_migrations"
).fetchall()
assert ("001",) in versions
finally:
conn.close()
def test_idempotent_migrations(self, ppi_db):
"""Running migrations twice should not fail."""
applied = run_ppi_migrations(ppi_db, MIGRATIONS_DIR)
assert applied == []
def test_all_tables_exist(self, ppi_db):
conn = get_connection(ppi_db)
try:
tables = {
row[0]
for row in conn.execute(
"SELECT name FROM sqlite_master WHERE type='table'"
).fetchall()
}
expected = {
"schema_migrations",
"dataset_versions",
"proteins",
"ppi_experiments",
"ppi_negative_results",
"protein_protein_pairs",
"ppi_split_definitions",
"ppi_split_assignments",
}
assert expected.issubset(tables), f"Missing: {expected - tables}"
finally:
conn.close()
def test_all_indices_exist(self, ppi_db):
conn = get_connection(ppi_db)
try:
indices = {
row[0]
for row in conn.execute(
"SELECT name FROM sqlite_master WHERE type='index'"
).fetchall()
}
expected = {
"idx_proteins_uniprot",
"idx_proteins_gene",
"idx_ppi_exp_source",
"idx_ppi_nr_protein1",
"idx_ppi_nr_protein2",
"idx_ppi_nr_pair",
"idx_ppi_nr_tier",
"idx_ppi_nr_source",
"idx_ppi_nr_unique_source",
"idx_ppp_protein1",
"idx_ppp_protein2",
"idx_ppp_confidence",
"idx_ppi_splits_fold",
}
assert expected.issubset(indices), f"Missing: {expected - indices}"
finally:
conn.close()
class TestPPISchema:
"""Test PPI schema constraints and foreign keys."""
def test_protein_unique_constraint(self, ppi_db):
conn = get_connection(ppi_db)
try:
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('P12345')"
)
conn.commit()
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('P12345')"
)
finally:
conn.close()
def test_evidence_type_check(self, ppi_db):
conn = get_connection(ppi_db)
try:
# Insert required proteins
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('A00001')"
)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('B00002')"
)
conn.commit()
# Valid evidence type
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (1, 2, 'experimental_non_interaction', 'gold', "
"'intact', 'R001', 'database_direct')"
)
conn.commit()
# Invalid evidence type
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (1, 2, 'invalid_type', 'gold', "
"'intact', 'R002', 'database_direct')"
)
finally:
conn.close()
def test_confidence_tier_check(self, ppi_db):
conn = get_connection(ppi_db)
try:
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('A00001')"
)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('B00002')"
)
conn.commit()
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (1, 2, 'experimental_non_interaction', 'platinum', "
"'intact', 'R001', 'database_direct')"
)
finally:
conn.close()
def test_canonical_ordering_enforced(self, ppi_db):
"""protein1_id must be < protein2_id."""
conn = get_connection(ppi_db)
try:
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('A00001')"
)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('B00002')"
)
conn.commit()
# protein1_id=2, protein2_id=1 violates CHECK
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (2, 1, 'experimental_non_interaction', 'gold', "
"'intact', 'R001', 'database_direct')"
)
finally:
conn.close()
def test_canonical_ordering_equal_ids_rejected(self, ppi_db):
"""protein1_id == protein2_id violates CHECK (strict less-than)."""
conn = get_connection(ppi_db)
try:
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('A00001')"
)
conn.commit()
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (1, 1, 'experimental_non_interaction', 'gold', "
"'intact', 'R001', 'database_direct')"
)
finally:
conn.close()
def test_extraction_method_check(self, ppi_db):
conn = get_connection(ppi_db)
try:
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('A00001')"
)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('B00002')"
)
conn.commit()
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (1, 2, 'experimental_non_interaction', 'gold', "
"'intact', 'R001', 'invalid_method')"
)
finally:
conn.close()
def test_foreign_key_enforcement(self, ppi_db):
"""FK violations should raise errors."""
conn = get_connection(ppi_db)
try:
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (999, 1000, 'experimental_non_interaction', 'gold', "
"'intact', 'R001', 'database_direct')"
)
finally:
conn.close()
def test_dedup_unique_index(self, ppi_db):
"""Same (pair, experiment, source, record) should fail."""
conn = get_connection(ppi_db)
try:
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('A00001')"
)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('B00002')"
)
conn.commit()
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (1, 2, 'experimental_non_interaction', 'gold', "
"'intact', 'R001', 'database_direct')"
)
conn.commit()
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (1, 2, 'ml_predicted_negative', 'silver', "
"'intact', 'R001', 'database_direct')"
)
finally:
conn.close()
def test_dedup_allows_different_source_records(self, ppi_db):
"""Same pair but different source_record_id is OK."""
conn = get_connection(ppi_db)
try:
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('A00001')"
)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('B00002')"
)
conn.commit()
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (1, 2, 'experimental_non_interaction', 'gold', "
"'intact', 'R001', 'database_direct')"
)
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, confidence_tier, "
" source_db, source_record_id, extraction_method) "
"VALUES (1, 2, 'experimental_non_interaction', 'gold', "
"'intact', 'R002', 'database_direct')"
)
conn.commit()
count = conn.execute(
"SELECT COUNT(*) FROM ppi_negative_results"
).fetchone()[0]
assert count == 2
finally:
conn.close()
def test_experiment_source_db_check(self, ppi_db):
conn = get_connection(ppi_db)
try:
# Valid source_db
conn.execute(
"INSERT INTO ppi_experiments "
"(source_db, source_experiment_id) "
"VALUES ('intact', 'EXP001')"
)
conn.commit()
# Invalid source_db
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO ppi_experiments "
"(source_db, source_experiment_id) "
"VALUES ('invalid_db', 'EXP002')"
)
finally:
conn.close()
def test_experiment_huri_source_db(self, ppi_db):
"""HuRI is a valid source_db."""
conn = get_connection(ppi_db)
try:
conn.execute(
"INSERT INTO ppi_experiments "
"(source_db, source_experiment_id) "
"VALUES ('huri', 'HI-III-20')"
)
conn.commit()
count = conn.execute(
"SELECT COUNT(*) FROM ppi_experiments WHERE source_db = 'huri'"
).fetchone()[0]
assert count == 1
finally:
conn.close()
def test_split_strategy_check(self, ppi_db):
conn = get_connection(ppi_db)
try:
# Valid strategy
conn.execute(
"INSERT INTO ppi_split_definitions "
"(split_name, split_strategy) "
"VALUES ('random_v1', 'random')"
)
conn.commit()
# Invalid strategy
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO ppi_split_definitions "
"(split_name, split_strategy) "
"VALUES ('bad_v1', 'invalid_strategy')"
)
finally:
conn.close()
def test_pairs_canonical_ordering(self, ppi_db):
"""protein_protein_pairs also enforces protein1 < protein2."""
conn = get_connection(ppi_db)
try:
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('A00001')"
)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('B00002')"
)
conn.commit()
with pytest.raises(sqlite3.IntegrityError):
conn.execute(
"INSERT INTO protein_protein_pairs "
"(protein1_id, protein2_id, num_experiments, num_sources, "
" best_confidence) "
"VALUES (2, 1, 1, 1, 'gold')"
)
finally:
conn.close()
class TestRefreshPPIPairs:
"""Test protein_protein_pairs aggregation."""
def _insert_test_data(self, conn):
"""Insert minimal test data for aggregation testing."""
# 3 proteins: A(1), B(2), C(3)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('A00001')"
)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('B00002')"
)
conn.execute(
"INSERT INTO proteins (uniprot_accession) VALUES ('C00003')"
)
# Experiments
conn.execute(
"INSERT INTO ppi_experiments "
"(source_db, source_experiment_id) "
"VALUES ('intact', 'EXP001')"
)
conn.execute(
"INSERT INTO ppi_experiments "
"(source_db, source_experiment_id) "
"VALUES ('humap', 'HUMAP001')"
)
# A-B: 2 results (gold + silver, different sources)
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, experiment_id, evidence_type, "
" confidence_tier, source_db, source_record_id, "
" extraction_method, publication_year, interaction_score) "
"VALUES (1, 2, 1, 'experimental_non_interaction', 'gold', "
"'intact', 'R001', 'database_direct', 2020, 0.05)"
)
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, experiment_id, evidence_type, "
" confidence_tier, source_db, source_record_id, "
" extraction_method, publication_year, interaction_score) "
"VALUES (1, 2, 2, 'ml_predicted_negative', 'silver', "
"'humap', 'R002', 'ml_classifier', 2023, 0.01)"
)
# A-C: 1 result (bronze)
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, "
" confidence_tier, source_db, source_record_id, "
" extraction_method, publication_year) "
"VALUES (1, 3, 'low_score_negative', 'bronze', "
"'string', 'R003', 'score_threshold', 2022)"
)
# B-C: 1 result (copper)
conn.execute(
"INSERT INTO ppi_negative_results "
"(protein1_id, protein2_id, evidence_type, "
" confidence_tier, source_db, source_record_id, "
" extraction_method, publication_year) "
"VALUES (2, 3, 'compartment_separated', 'copper', "
"'string', 'R004', 'score_threshold', 2021)"
)
conn.commit()
def test_refresh_pair_count(self, ppi_db):
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
count = refresh_all_ppi_pairs(conn)
conn.commit()
assert count == 3 # A-B, A-C, B-C
finally:
conn.close()
def test_best_confidence_aggregation(self, ppi_db):
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
row = conn.execute(
"SELECT best_confidence, num_experiments, num_sources "
"FROM protein_protein_pairs "
"WHERE protein1_id = 1 AND protein2_id = 2"
).fetchone()
assert row[0] == "gold" # best of gold + silver
assert row[1] == 2 # 2 distinct experiments
assert row[2] == 2 # 2 distinct sources (intact, humap)
finally:
conn.close()
def test_best_evidence_type_priority(self, ppi_db):
"""experimental_non_interaction should rank above ml_predicted."""
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
row = conn.execute(
"SELECT best_evidence_type FROM protein_protein_pairs "
"WHERE protein1_id = 1 AND protein2_id = 2"
).fetchone()
assert row[0] == "experimental_non_interaction"
finally:
conn.close()
def test_score_range(self, ppi_db):
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
row = conn.execute(
"SELECT min_interaction_score, max_interaction_score "
"FROM protein_protein_pairs "
"WHERE protein1_id = 1 AND protein2_id = 2"
).fetchone()
assert row[0] == pytest.approx(0.01)
assert row[1] == pytest.approx(0.05)
finally:
conn.close()
def test_earliest_year(self, ppi_db):
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
row = conn.execute(
"SELECT earliest_year FROM protein_protein_pairs "
"WHERE protein1_id = 1 AND protein2_id = 2"
).fetchone()
assert row[0] == 2020
finally:
conn.close()
def test_degree_computation_protein1(self, ppi_db):
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
# Protein A (id=1) appears as protein1 in A-B and A-C → degree 2
row = conn.execute(
"SELECT protein1_degree FROM protein_protein_pairs "
"WHERE protein1_id = 1 AND protein2_id = 2"
).fetchone()
assert row[0] == 2
finally:
conn.close()
def test_degree_computation_protein2(self, ppi_db):
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
# Protein C (id=3) appears as protein2 in A-C and B-C → degree 2
row = conn.execute(
"SELECT protein2_degree FROM protein_protein_pairs "
"WHERE protein1_id = 1 AND protein2_id = 3"
).fetchone()
assert row[0] == 2
finally:
conn.close()
def test_degree_computation_both_sides(self, ppi_db):
"""Protein B (id=2) appears as protein2 in A-B and protein1 in B-C.
True degree = 2 (partners A and C) regardless of which side it's on."""
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
# In A-B row, protein2 is B → protein2_degree = B's full degree = 2
row = conn.execute(
"SELECT protein2_degree FROM protein_protein_pairs "
"WHERE protein1_id = 1 AND protein2_id = 2"
).fetchone()
assert row[0] == 2
# In B-C row, protein1 is B → protein1_degree = B's full degree = 2
row = conn.execute(
"SELECT protein1_degree FROM protein_protein_pairs "
"WHERE protein1_id = 2 AND protein2_id = 3"
).fetchone()
assert row[0] == 2
finally:
conn.close()
def test_refresh_is_idempotent(self, ppi_db):
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
count1 = refresh_all_ppi_pairs(conn)
conn.commit()
count2 = refresh_all_ppi_pairs(conn)
conn.commit()
assert count1 == count2
finally:
conn.close()
def test_refresh_clears_split_assignments(self, ppi_db):
"""Refresh should succeed even when split assignments exist (FK)."""
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
# Create a split and assign a pair
conn.execute(
"INSERT INTO ppi_split_definitions "
"(split_name, split_strategy) VALUES ('test_v1', 'random')"
)
conn.execute(
"INSERT INTO ppi_split_assignments (pair_id, split_id, fold) "
"VALUES (1, 1, 'train')"
)
conn.commit()
# Re-refresh should not crash despite FK to split_assignments
count = refresh_all_ppi_pairs(conn)
conn.commit()
assert count == 3
# Split assignments should be cleared
sa_count = conn.execute(
"SELECT COUNT(*) FROM ppi_split_assignments"
).fetchone()[0]
assert sa_count == 0
finally:
conn.close()
def test_null_experiment_counted(self, ppi_db):
"""Results without experiment_id should use COALESCE sentinel."""
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
# A-C has NULL experiment_id → COALESCE(-1) → counted as 1
row = conn.execute(
"SELECT num_experiments FROM protein_protein_pairs "
"WHERE protein1_id = 1 AND protein2_id = 3"
).fetchone()
assert row[0] == 1
finally:
conn.close()
def test_single_source_count(self, ppi_db):
conn = get_connection(ppi_db)
try:
self._insert_test_data(conn)
refresh_all_ppi_pairs(conn)
conn.commit()
# A-C has 1 source (string)
row = conn.execute(
"SELECT num_sources FROM protein_protein_pairs "
"WHERE protein1_id = 1 AND protein2_id = 3"
).fetchone()
assert row[0] == 1
finally:
conn.close()