Skip to content

CRISPRme+ v2.7.2 — #46 Tier-1 co-occurrence AF fix

Choose a tag to compare

@lucapinello lucapinello released this 08 Oct 02:15
· 810 commits to main since this release

Fixed

  • #46 (co-occurrence path): Tier-1 genotype axis was not filtered to VCF-genotyped samples.
    The phantom-sample filter (genotyped_samples) was wired on the Tier-0 registry
    (compile_from_dict) but not on the Tier-1 genotype store (compile_genotypes_from_dict), so an
    over-listing samplesID inflated the Tier-1 sample axis / AN and deflated the per-sample
    CONFIRMED-cis joint-AF denominator
    . The same genotyped_samples set is now threaded through to
    tier0_compile.build_sample_meta (already filter-capable); no-op when None (batteries install /
    source VCF absent). Build-path only — correctly-built shipped indices are unaffected (e.g.
    HGDP1kGP's header-derived 4,091-sample panel has no phantoms). Added
    test_genotyped_samples_filter_drops_phantom_from_axis.