Datasets:
Formats:
parquet
Languages:
English
Size:
10M - 100M
Tags:
biology
chemistry
drug-discovery
clinical-trials
protein-protein-interaction
gene-essentiality
License:
File size: 15,601 Bytes
6d1bbc7 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 | """Tests for GE ML export module.
Tests split generation, conflict resolution, and control negatives
using a small synthetic database.
"""
from pathlib import Path
import numpy as np
import pandas as pd
import pytest
from negbiodb_depmap.depmap_db import get_connection, run_ge_migrations, refresh_all_ge_pairs
from negbiodb_depmap.export import (
apply_split_to_dataset,
build_ge_m1,
build_ge_m2,
export_ge_negatives,
generate_cold_both_split,
generate_cold_cell_line_split,
generate_cold_gene_split,
generate_degree_balanced_split,
generate_random_split,
generate_uniform_random_negatives,
)
MIGRATIONS_DIR = Path(__file__).parent.parent / "migrations_depmap"
@pytest.fixture
def tmp_db(tmp_path):
db_path = tmp_path / "test_ge.db"
run_ge_migrations(db_path, MIGRATIONS_DIR)
return db_path
@pytest.fixture
def seeded_db(tmp_db):
"""DB with 3 genes × 3 cell lines = 6 negative pairs."""
conn = get_connection(tmp_db)
# Genes
conn.execute("INSERT INTO genes (gene_id, entrez_id, gene_symbol, is_reference_nonessential) VALUES (1, 1001, 'GeneA', 1)")
conn.execute("INSERT INTO genes (gene_id, entrez_id, gene_symbol) VALUES (2, 1002, 'GeneB')")
conn.execute("INSERT INTO genes (gene_id, entrez_id, gene_symbol, is_common_essential) VALUES (3, 1003, 'GeneC', 1)")
# Cell lines
conn.execute("INSERT INTO cell_lines (cell_line_id, model_id, lineage) VALUES (1, 'ACH-001', 'Lung')")
conn.execute("INSERT INTO cell_lines (cell_line_id, model_id, lineage) VALUES (2, 'ACH-002', 'Breast')")
conn.execute("INSERT INTO cell_lines (cell_line_id, model_id, lineage) VALUES (3, 'ACH-003', 'Skin')")
# Screen
conn.execute(
"""INSERT INTO ge_screens (screen_id, source_db, depmap_release, screen_type, algorithm)
VALUES (1, 'depmap', 'test', 'crispr', 'Chronos')"""
)
# 6 negative results (GeneA in 3 CLs, GeneB in 3 CLs)
for gene_id in [1, 2]:
for cl_id in [1, 2, 3]:
effect = 0.05 if gene_id == 1 else -0.4
conn.execute(
"""INSERT INTO ge_negative_results
(gene_id, cell_line_id, screen_id, gene_effect_score, dependency_probability,
evidence_type, confidence_tier, source_db, source_record_id, extraction_method)
VALUES (?, ?, 1, ?, 0.1, 'crispr_nonessential', 'silver',
'depmap', ?, 'score_threshold')""",
(gene_id, cl_id, effect, f"r{gene_id}_{cl_id}"),
)
conn.commit()
refresh_all_ge_pairs(conn)
conn.commit()
conn.close()
return tmp_db
# ── Split tests ───────────────────────────────────────────────────────────
class TestRandomSplit:
def test_all_pairs_assigned(self, seeded_db):
conn = get_connection(seeded_db)
try:
result = generate_random_split(conn, seed=42)
total = sum(result["counts"].values())
assert total == 6
finally:
conn.close()
def test_fold_coverage(self, seeded_db):
conn = get_connection(seeded_db)
try:
result = generate_random_split(conn, seed=42)
assert set(result["counts"].keys()).issubset({"train", "val", "test"})
finally:
conn.close()
class TestColdGeneSplit:
def test_all_pairs_assigned(self, seeded_db):
conn = get_connection(seeded_db)
try:
result = generate_cold_gene_split(conn, seed=42)
total = sum(result["counts"].values())
assert total == 6
finally:
conn.close()
def test_no_gene_leakage(self, seeded_db):
"""Test genes in train should not appear in test."""
conn = get_connection(seeded_db)
try:
result = generate_cold_gene_split(conn, seed=42)
split_id = result["split_id"]
train_genes = {
r[0] for r in conn.execute(
"""SELECT DISTINCT p.gene_id
FROM ge_split_assignments sa
JOIN gene_cell_pairs p ON sa.pair_id = p.pair_id
WHERE sa.split_id = ? AND sa.fold = 'train'""",
(split_id,),
).fetchall()
}
test_genes = {
r[0] for r in conn.execute(
"""SELECT DISTINCT p.gene_id
FROM ge_split_assignments sa
JOIN gene_cell_pairs p ON sa.pair_id = p.pair_id
WHERE sa.split_id = ? AND sa.fold = 'test'""",
(split_id,),
).fetchall()
}
assert len(train_genes & test_genes) == 0
finally:
conn.close()
class TestColdCellLineSplit:
def test_all_pairs_assigned(self, seeded_db):
conn = get_connection(seeded_db)
try:
result = generate_cold_cell_line_split(conn, seed=42)
total = sum(result["counts"].values())
assert total == 6
finally:
conn.close()
def test_no_cell_line_leakage(self, seeded_db):
conn = get_connection(seeded_db)
try:
result = generate_cold_cell_line_split(conn, seed=42)
split_id = result["split_id"]
train_cls = {
r[0] for r in conn.execute(
"""SELECT DISTINCT p.cell_line_id
FROM ge_split_assignments sa
JOIN gene_cell_pairs p ON sa.pair_id = p.pair_id
WHERE sa.split_id = ? AND sa.fold = 'train'""",
(split_id,),
).fetchall()
}
test_cls = {
r[0] for r in conn.execute(
"""SELECT DISTINCT p.cell_line_id
FROM ge_split_assignments sa
JOIN gene_cell_pairs p ON sa.pair_id = p.pair_id
WHERE sa.split_id = ? AND sa.fold = 'test'""",
(split_id,),
).fetchall()
}
assert len(train_cls & test_cls) == 0
finally:
conn.close()
class TestColdBothSplit:
def test_all_pairs_assigned(self, seeded_db):
conn = get_connection(seeded_db)
try:
result = generate_cold_both_split(conn, seed=42)
total = sum(result["counts"].values())
assert total == 6
finally:
conn.close()
class TestDegreeBalancedSplit:
def test_all_pairs_assigned(self, seeded_db):
conn = get_connection(seeded_db)
try:
result = generate_degree_balanced_split(conn, seed=42)
total = sum(result["counts"].values())
assert total == 6
finally:
conn.close()
# ── Export tests ──────────────────────────────────────────────────────────
class TestExport:
def test_export_parquet(self, seeded_db, tmp_path):
conn = get_connection(seeded_db)
try:
generate_random_split(conn, seed=42)
out = tmp_path / "export" / "pairs.parquet"
count = export_ge_negatives(conn, out)
assert count == 6
assert out.exists()
df = pd.read_parquet(out)
assert "gene_symbol" in df.columns
assert "model_id" in df.columns
assert "split_random_v1" in df.columns
finally:
conn.close()
# ── M1/M2 tests ──────────────────────────────────────────────────────────
class TestBuildM1:
def test_basic_m1(self, seeded_db):
conn = get_connection(seeded_db)
try:
pos_df = pd.DataFrame({
"gene_id": [3, 3],
"cell_line_id": [1, 2],
"essentiality_type": ["common_essential", "common_essential"],
})
neg_df = pd.DataFrame({
"gene_id": [1, 1, 2],
"cell_line_id": [1, 2, 1],
})
result = build_ge_m1(conn, pos_df, neg_df, balanced=True, ratio=1.0)
assert (result["label"] == 1).sum() == 2
assert (result["label"] == 0).sum() == 2
finally:
conn.close()
def test_conflict_resolution(self, seeded_db):
conn = get_connection(seeded_db)
try:
# Gene 1 in CL 1 appears in both positive and negative
pos_df = pd.DataFrame({
"gene_id": [1, 3],
"cell_line_id": [1, 1],
"essentiality_type": ["selective_essential", "common_essential"],
})
neg_df = pd.DataFrame({
"gene_id": [1, 2],
"cell_line_id": [1, 1],
})
result = build_ge_m1(conn, pos_df, neg_df, balanced=False)
# (1, 1) conflict → removed from both
assert len(result) == 2 # gene3-cl1 pos + gene2-cl1 neg
finally:
conn.close()
class TestBuildM2:
def test_three_classes(self, seeded_db):
conn = get_connection(seeded_db)
try:
pos_df = pd.DataFrame({
"gene_id": [3, 2],
"cell_line_id": [1, 3],
"essentiality_type": ["common_essential", "selective_essential"],
})
neg_df = pd.DataFrame({
"gene_id": [1],
"cell_line_id": [1],
})
result = build_ge_m2(conn, pos_df, neg_df)
assert (result["label"] == 0).sum() == 1 # common essential
assert (result["label"] == 1).sum() == 1 # selective essential
assert (result["label"] == 2).sum() == 1 # non-essential
finally:
conn.close()
# ── apply_split_to_dataset tests ─────────────────────────────────────────
class TestApplySplitToDataset:
"""Test dataset-level split application for ML training."""
@pytest.fixture
def combined_df(self):
"""Combined pos+neg DataFrame with 100 rows, 5 genes × 4 cell lines."""
rng = np.random.RandomState(0)
n = 100
genes = rng.choice([1, 2, 3, 4, 5], n)
cls = rng.choice([10, 20, 30, 40], n)
return pd.DataFrame({
"gene_id": genes,
"cell_line_id": cls,
"label": rng.choice([0, 1], n),
"gene_degree": rng.randint(1, 2000, n),
})
def test_random_split_coverage(self, combined_df):
result = apply_split_to_dataset(combined_df, "random", seed=42)
assert "split" in result.columns
assert set(result["split"].unique()) == {"train", "val", "test"}
assert len(result) == len(combined_df)
def test_random_split_ratios(self, combined_df):
result = apply_split_to_dataset(combined_df, "random", seed=42)
n = len(combined_df)
assert abs(sum(result["split"] == "train") - int(n * 0.7)) <= 1
assert abs(sum(result["split"] == "val") - int(n * 0.1)) <= 1
def test_cold_gene_no_leakage(self, combined_df):
result = apply_split_to_dataset(combined_df, "cold_gene", seed=42)
train_genes = set(result.loc[result["split"] == "train", "gene_id"])
test_genes = set(result.loc[result["split"] == "test", "gene_id"])
assert len(train_genes & test_genes) == 0
def test_cold_cell_line_no_leakage(self, combined_df):
result = apply_split_to_dataset(combined_df, "cold_cell_line", seed=42)
train_cls = set(result.loc[result["split"] == "train", "cell_line_id"])
test_cls = set(result.loc[result["split"] == "test", "cell_line_id"])
assert len(train_cls & test_cls) == 0
def test_cold_both_metis_property(self):
"""In Metis-style: train pairs have both entities in train partition."""
rng = np.random.RandomState(0)
n = 500
# Use enough entities so val partition isn't empty (5% of 20 = 1)
genes = rng.choice(range(1, 21), n)
cls = rng.choice(range(100, 121), n)
df = pd.DataFrame({
"gene_id": genes,
"cell_line_id": cls,
"label": rng.choice([0, 1], n),
})
result = apply_split_to_dataset(df, "cold_both", seed=42)
assert set(result["split"].unique()) == {"train", "val", "test"}
assert len(result) == n
def test_degree_balanced_coverage(self, combined_df):
result = apply_split_to_dataset(combined_df, "degree_balanced", seed=42)
assert set(result["split"].unique()) == {"train", "val", "test"}
assert len(result) == len(combined_df)
def test_different_splits_differ(self, combined_df):
"""Different strategies should produce different assignments."""
r1 = apply_split_to_dataset(combined_df, "random", seed=42)
r2 = apply_split_to_dataset(combined_df, "cold_gene", seed=42)
# At least some rows should differ
assert not (r1["split"] == r2["split"]).all()
def test_different_seeds_differ(self, combined_df):
r1 = apply_split_to_dataset(combined_df, "random", seed=42)
r2 = apply_split_to_dataset(combined_df, "random", seed=99)
assert not (r1["split"] == r2["split"]).all()
def test_same_seed_reproducible(self, combined_df):
r1 = apply_split_to_dataset(combined_df, "cold_gene", seed=42)
r2 = apply_split_to_dataset(combined_df, "cold_gene", seed=42)
assert (r1["split"] == r2["split"]).all()
def test_unknown_strategy_raises(self, combined_df):
with pytest.raises(ValueError, match="Unknown split strategy"):
apply_split_to_dataset(combined_df, "bogus")
def test_degree_balanced_no_degree_col_falls_back(self):
"""Without gene_degree column, should fall back to random."""
df = pd.DataFrame({
"gene_id": [1, 2, 3, 4],
"cell_line_id": [10, 20, 30, 40],
"label": [0, 1, 0, 1],
})
result = apply_split_to_dataset(df, "degree_balanced", seed=42)
assert "split" in result.columns
# ── Control negatives ─────────────────────────────────────────────────────
class TestControlNegatives:
def test_uniform_random(self, seeded_db):
conn = get_connection(seeded_db)
try:
result = generate_uniform_random_negatives(conn, n_samples=5, seed=42)
assert len(result) <= 5
assert "neg_source" in result.columns
assert all(result["neg_source"] == "uniform_random")
finally:
conn.close()
def test_no_overlap_with_existing(self, seeded_db):
conn = get_connection(seeded_db)
try:
existing = set()
for row in conn.execute(
"SELECT gene_id, cell_line_id FROM gene_cell_pairs"
).fetchall():
existing.add((row[0], row[1]))
result = generate_uniform_random_negatives(conn, n_samples=5, seed=42)
for _, r in result.iterrows():
assert (r["gene_id"], r["cell_line_id"]) not in existing
finally:
conn.close()
|