Human genome variant interpretation — the sibling to MethylAsk (methylation), both built on bio-core.
MethylAsk answers "what does my methylation mean"; GeneAsk answers "what do my DNA variants mean" — a ClinVar clinical screen, trait genotypes with multi-source confidence, and pharmacogenomics. It ingests 23andMe raw data, VCFs, and the variant stream from an ONT modBAM (split out by bio-core).
GeneAsk holds human variant knowledge (ClinVar, PRS, pharmacogenomics, the trait framework). All organism-agnostic mechanism — the provider/report layer, evidence tiering, coordinate handling — lives in bio-core and is imported.
geneask.ingest.from_23andme/from_complete_genomics/consensus— raw genotype → reference-anchored VCF, multi-source consensusgeneask.interpret.traits— genotype lookup with multi-source confidence → bio-core Findingsgeneask.interpret.clinvar_screen— clinical screen against the bundled 157-gene ClinVar panel
geneask/data/reference/clinvar_panel_157genes.json.gz— ACMG SF v3.2 + carrier + pharmacogene panel (157 genes, gene-keyed, release 2026-06-06). Bundled so the clinical screen runs offline.
Consumer-array "pathogenic" calls are frequently false positives — in the source work, 10 of 11 were refuted by whole-genome sequencing. GeneAsk therefore reports the calling platform and multi-source confidence, and never presents a single-array pathogenic hit as robust. This is the same evidence-tiering discipline MethylAsk applies to methylation associations.
GeneAsk depends on the private GoodAncestor/bio-core repo, declared as a
git-source dependency in pyproject.toml. A plain pip install . resolves it
from GitHub, so the machine needs git access to the private repo (SSH key or a
token in the git credential helper).
# A) let pip pull bio-core from GitHub (needs private-repo git access)
pip install .
# B) develop against a local bio-core checkout instead
git clone https://github.com/GoodAncestor/bio-core.git ../bio-core
pip install -e ../bio-core
pip install -e . --no-deps