Process and query NCBI SRA metadata locally
with streaming ingestion, full-text search, and vector similarity
Process 14GB+ NCBI archives with minimal memory. HTTP to Gzip to Tar to XML to SQLite in a single pass.
Bleve full-text, SQLite FTS5, and SapBERT vector similarity search with configurable hybrid ranking.
Filter by organism, platform, library strategy, date range, and sequencing metrics during ingestion and search.
HTTP API for programmatic access. Model Context Protocol support for AI assistant integration.
Export to classic SRAmetadb.sqlite format for compatibility with existing R and Python tools.
Follows XDG Base Directory Specification. All paths configurable via environment variables.