Datasets:
Formats:
parquet
Languages:
English
Size:
10M - 100M
Tags:
biology
chemistry
drug-discovery
clinical-trials
protein-protein-interaction
gene-essentiality
License:
File size: 12,017 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 | """STRING v12.0 negative PPI ETL — zero-score pairs between well-studied proteins.
Derives Bronze-tier negatives from STRING: protein pairs where BOTH proteins
have degree >= min_degree but no interaction evidence across all STRING channels.
Since the STRING links file only contains pairs with score > 0, we must:
1. Build the set of well-studied proteins (degree >= threshold)
2. Restrict to a protein universe (e.g., proteins in IntAct/hu.MAP)
3. Compute Cartesian product minus STRING-linked pairs
4. Cap at max_pairs to prevent DB bloat
License: CC BY 4.0
"""
import gzip
import itertools
import random
from pathlib import Path
from negbiodb_ppi.protein_mapper import canonical_pair, get_or_insert_protein, validate_uniprot
def load_string_mapping(mapping_path: str | Path) -> dict[str, str]:
"""Load ENSP → UniProt mapping from STRING mapping file.
File format: 5 columns, tab-separated, NO header
Col 1: Species taxon ID (e.g., '9606')
Col 2: 'P31946|1433B_HUMAN' → split on '|' → take [0]
Col 3: '9606.ENSP00000361930'
Col 4: Identity %
Col 5: Bit score
Returns:
Dict mapping STRING ID (e.g., '9606.ENSP00000361930') to UniProt accession.
"""
mapping = {}
path = Path(mapping_path)
opener = gzip.open if path.suffix == ".gz" else open
with opener(path, "rt") as f:
for line in f:
line = line.strip()
if not line or line.startswith("#"):
continue
parts = line.split("\t")
if len(parts) < 3:
continue
# Col 2: 'P31946|1433B_HUMAN'
uniprot_raw = parts[1].split("|")[0].strip()
acc = validate_uniprot(uniprot_raw)
if not acc:
continue
# Col 3: STRING ID
string_id = parts[2].strip()
if string_id:
mapping[string_id] = acc
return mapping
def compute_protein_degrees(
links_path: str | Path,
ensp_to_uniprot: dict[str, str] | None = None,
) -> dict[str, int]:
"""Compute degree for each protein from STRING links file.
File format: space-separated, header: 'protein1 protein2 combined_score'
IDs: '9606.ENSP00000361930'
Args:
links_path: Path to STRING links file (may be .gz).
ensp_to_uniprot: If provided, returns degrees keyed by UniProt accession.
Returns:
Dict mapping protein ID (ENSP or UniProt) to degree.
"""
degrees: dict[str, int] = {}
path = Path(links_path)
opener = gzip.open if path.suffix == ".gz" else open
with opener(path, "rt") as f:
for line in f:
line = line.strip()
if not line or line.startswith("protein"):
continue
parts = line.split()
if len(parts) < 3:
continue
p1, p2 = parts[0], parts[1]
if ensp_to_uniprot is not None:
p1 = ensp_to_uniprot.get(p1, p1)
p2 = ensp_to_uniprot.get(p2, p2)
if p1 == p2:
continue # skip isoform self-links
degrees[p1] = degrees.get(p1, 0) + 1
degrees[p2] = degrees.get(p2, 0) + 1
return degrees
def load_linked_pairs(
links_path: str | Path,
ensp_to_uniprot: dict[str, str],
) -> set[tuple[str, str]]:
"""Load all STRING-linked UniProt pairs (score > 0).
Returns:
Set of canonical (uniprot_a, uniprot_b) pairs.
"""
pairs = set()
path = Path(links_path)
opener = gzip.open if path.suffix == ".gz" else open
with opener(path, "rt") as f:
for line in f:
line = line.strip()
if not line or line.startswith("protein"):
continue
parts = line.split()
if len(parts) < 3:
continue
acc_a = ensp_to_uniprot.get(parts[0])
acc_b = ensp_to_uniprot.get(parts[1])
if acc_a and acc_b and acc_a != acc_b:
pairs.add(canonical_pair(acc_a, acc_b))
return pairs
def extract_zero_score_pairs(
linked_pairs: set[tuple[str, str]],
protein_degrees: dict[str, int],
min_degree: int = 5,
max_pairs: int = 500_000,
protein_universe: set[str] | None = None,
random_seed: int = 42,
) -> list[tuple[str, str]]:
"""Extract protein pairs with no STRING evidence.
Strategy:
1. Collect all proteins with degree >= min_degree
2. If protein_universe provided, intersect with it
3. Compute Cartesian product of qualifying proteins
4. Subtract all pairs present in STRING links
5. Cap at max_pairs (random sample if exceeding)
Args:
linked_pairs: Set of canonical pairs from STRING links.
protein_degrees: Dict mapping UniProt to degree.
min_degree: Minimum STRING degree to consider "well-studied".
max_pairs: Maximum number of negative pairs.
protein_universe: Optional set of proteins to restrict to.
random_seed: Seed for reproducible sampling.
Returns:
Sorted list of (uniprot_a, uniprot_b) canonical pairs.
"""
# Well-studied proteins
candidates = {p for p, d in protein_degrees.items() if d >= min_degree}
# Only keep valid UniProt accessions
candidates = {p for p in candidates if validate_uniprot(p)}
if protein_universe is not None:
candidates = candidates & protein_universe
sorted_candidates = sorted(candidates)
# Reservoir sampling over lazy Cartesian product to avoid OOM.
# For ~15K proteins, set(combinations) would need ~11 GB.
rng = random.Random(random_seed)
result = []
count = 0
for pair in itertools.combinations(sorted_candidates, 2):
if pair in linked_pairs:
continue
count += 1
if len(result) < max_pairs:
result.append(pair)
else:
j = rng.randrange(count)
if j < max_pairs:
result[j] = pair
return sorted(result)
def run_string_etl(
db_path: str | Path | None = None,
data_dir: str | Path | None = None,
min_degree: int = 5,
max_pairs: int = 500_000,
protein_universe: set[str] | None = None,
mapping_path: str | Path | None = None,
links_path: str | Path | None = None,
) -> dict:
"""Orchestrator: load STRING zero-score negatives.
Args:
db_path: Path to PPI database.
data_dir: Directory containing STRING data files.
min_degree: Minimum STRING degree for "well-studied".
max_pairs: Maximum number of negative pairs.
protein_universe: Optional set of proteins to restrict to.
mapping_path: Explicit path to ENSP→UniProt mapping file.
links_path: Explicit path to STRING protein links file.
Returns:
Stats dict.
"""
import logging
logger = logging.getLogger(__name__)
from negbiodb_ppi.ppi_db import DEFAULT_PPI_DB_PATH, get_connection, run_ppi_migrations
if db_path is None:
db_path = DEFAULT_PPI_DB_PATH
if data_dir is None:
data_dir = Path(db_path).parent.parent / "data" / "ppi" / "string"
db_path = Path(db_path)
data_dir = Path(data_dir)
run_ppi_migrations(db_path)
# Resolve file paths — use explicit paths if provided, else glob with warnings
if mapping_path is None:
mapping_files = list(data_dir.glob("*.uniprot_2_string*"))
if not mapping_files:
raise FileNotFoundError(f"No STRING mapping file found in {data_dir}")
if len(mapping_files) > 1:
logger.warning(
"Multiple STRING mapping files found: %s. Using %s",
mapping_files, mapping_files[0],
)
mapping_path = mapping_files[0]
else:
mapping_path = Path(mapping_path)
if links_path is None:
links_files = list(data_dir.glob("*.protein.links*"))
if not links_files:
raise FileNotFoundError(f"No STRING links file found in {data_dir}")
if len(links_files) > 1:
logger.warning(
"Multiple STRING links files found: %s. Using %s",
links_files, links_files[0],
)
links_path = links_files[0]
else:
links_path = Path(links_path)
# Load mapping
ensp_to_uniprot = load_string_mapping(mapping_path)
# Compute degrees in UniProt space
degrees = compute_protein_degrees(links_path, ensp_to_uniprot)
# Load linked pairs
linked = load_linked_pairs(links_path, ensp_to_uniprot)
# Extract zero-score pairs
negatives = extract_zero_score_pairs(
linked, degrees, min_degree, max_pairs, protein_universe
)
# Insert into database
conn = get_connection(db_path)
try:
conn.execute(
"INSERT OR IGNORE INTO ppi_experiments "
"(source_db, source_experiment_id, experiment_type, description) "
"VALUES ('string', 'string-v12.0-zero-score', 'computational', "
"'Zero-score pairs between well-studied proteins in STRING v12.0')",
)
conn.commit()
exp_row = conn.execute(
"SELECT experiment_id FROM ppi_experiments "
"WHERE source_db = 'string' AND source_experiment_id = 'string-v12.0-zero-score'"
).fetchone()
experiment_id = exp_row[0]
batch = []
for acc_a, acc_b in negatives:
pid_a = get_or_insert_protein(conn, acc_a)
pid_b = get_or_insert_protein(conn, acc_b)
if pid_a > pid_b:
pid_a, pid_b = pid_b, pid_a
batch.append((
pid_a, pid_b, experiment_id,
"low_score_negative", "bronze",
"string", f"string-{acc_a}-{acc_b}",
"score_threshold",
))
if len(batch) >= 10000:
conn.executemany(
"INSERT OR IGNORE INTO ppi_negative_results "
"(protein1_id, protein2_id, experiment_id, "
" evidence_type, confidence_tier, source_db, "
" source_record_id, extraction_method) "
"VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
batch,
)
batch = []
conn.commit()
if batch:
conn.executemany(
"INSERT OR IGNORE INTO ppi_negative_results "
"(protein1_id, protein2_id, experiment_id, "
" evidence_type, confidence_tier, source_db, "
" source_record_id, extraction_method) "
"VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
batch,
)
conn.commit()
# Query actual inserted count (INSERT OR IGNORE may skip duplicates)
inserted = conn.execute(
"SELECT COUNT(*) FROM ppi_negative_results WHERE source_db = 'string'"
).fetchone()[0]
# Idempotent dataset_versions: delete old row before inserting
conn.execute(
"DELETE FROM dataset_versions WHERE name = 'string' AND version = 'v12.0'"
)
conn.execute(
"INSERT INTO dataset_versions (name, version, source_url, row_count, notes) "
"VALUES ('string', 'v12.0', "
"'https://stringdb-downloads.org/', ?, "
"'Zero-score pairs between well-studied human proteins')",
(inserted,),
)
conn.commit()
finally:
conn.close()
well_studied = {p for p, d in degrees.items() if d >= min_degree and validate_uniprot(p)}
if protein_universe is not None:
well_studied = well_studied & protein_universe
return {
"mapping_entries": len(ensp_to_uniprot),
"linked_pairs": len(linked),
"proteins_with_degree": len(degrees),
"well_studied_proteins": len(well_studied),
"negative_pairs_derived": len(negatives),
"negative_pairs_inserted": inserted,
}
|