"""Pre-warm the gnomAD SQLite cache for a list of variants. The gnomAD GraphQL API is open but slow under sustained load. For batch classification runs (eval set, day-1 lab variant pull), pre-fetching every variant ID into the local SQLite cache (`data/gnomad_cache.db`) keeps the critical-path classification under 30s per variant. Input ----- A text file with one gnomAD variant ID per line (format: `chr-pos-ref-alt`, no `chr` prefix). Lines starting with `#` are skipped. Usage ----- python -m scripts.warm_gnomad_cache variant_ids.txt """ from __future__ import annotations import argparse import asyncio import logging import sys from pathlib import Path from backend.app.services.gnomad import GnomADClient logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(message)s") logger = logging.getLogger("warm_gnomad") async def warm(ids: list[str], concurrency: int) -> None: client = GnomADClient() sem = asyncio.Semaphore(concurrency) async def one(vid: str) -> None: async with sem: try: freq = await client.lookup(vid) logger.info( "%s — AF=%s, hom=%s", vid, f"{freq.overall_af:.2e}" if freq.overall_af else "—", freq.homozygote_count or 0, ) except Exception as e: logger.warning("%s failed: %s", vid, e) await asyncio.gather(*(one(v) for v in ids)) def main() -> int: parser = argparse.ArgumentParser() parser.add_argument("ids_file", type=Path, help="One variant ID per line (chr-pos-ref-alt)") parser.add_argument("--concurrency", type=int, default=4, help="Parallel requests (gnomAD rate-limits)") args = parser.parse_args() if not args.ids_file.exists(): logger.error("file not found: %s", args.ids_file) return 1 raw = args.ids_file.read_text().splitlines() ids = [ln.strip() for ln in raw if ln.strip() and not ln.startswith("#")] logger.info("warming %d variants with concurrency=%d", len(ids), args.concurrency) asyncio.run(warm(ids, args.concurrency)) return 0 if __name__ == "__main__": sys.exit(main())