Bioconductor annotation without the packages

Bioconductor distributes annotation as software: one versioned R package per organism per release. bioc-on-ice publishes the same knowledge as open tables that R, Python, DuckDB, a browser or an AI assistant can query directly — gene annotation for every species at once (NCBI Gene across 51,796 taxa, Ensembl across 276 species), identifier mappings, GO and other ontologies, orthologs, interactions, pathways, GWAS associations and literature links. Each row names the source that asserts it, so providers sit side by side instead of being reconciled into one view.

Large data stays where it lives: the resource tables catalog datasets in CELLxGENE, ENCODE, BEDbase and the eQTL Catalogue and point at the original files. Everything is published as frozen, dated releases — one Parquet file per table plus a DuckDB database of views over them — served over plain HTTPS. No package to install, no account, no token.

Query it

From DuckDB (any client: CLI, R, Python):

ATTACH 'https://data.bioconice.cancerdatasci.org/latest/bioconice.duckdb' AS bioc (READ_ONLY);

Then SHOW ALL TABLES;. Every table is also a plain Parquet file; see Releases for the dated versions, sizes, row counts and download links.

Other ways in