Biologically-informed interaction filtering
IGEM uses a curated biomedical knowledge graph to reduce the GxG and GxE search space before multi-test correction — surfacing interaction hypotheses with biological grounding rather than testing every possible pair.
From combinatorial chaos to grounded hypotheses
Pairwise interaction screens collapse under their own combinatorics. IGEM uses a curated knowledge graph as a pre-correction filter — testing fewer, smarter hypotheses instead of diluting power across billions of pairs.
One pip install. Zero infrastructure to manage.
IGEM ships as a single Python package — no database to provision, no ETL to run, no graph to keep up to date. Biological knowledge lives on a managed remote server (or, for HPC, an offline Parquet snapshot via DuckDB), and your cohort data stays on your machine. From loading PLINK genotypes to producing biologically annotated Manhattan plots, the entire pipeline runs in a single Python session.
Six capability modules cover the end-to-end workflow, with the same code path running against a remote server, an offline snapshot, or inside a container.
Built on a decade of methods research
IGEM is the third generation of an integrated software line developed by the Hall Lab — extending the knowledge-driven filtering established by Biofilter for GxG to GxE and ExE, on top of the analytical surface of PLATO and CLARITE.
See publications and how to citeReady in five minutes
Install with pip, point the client at the public IGEM server, and run your first knowledge-graph query.
pip install igem