Usage¶

Install genoABO and infer ABO blood group from a genotype file. For pipeline internals see Workflow; for haplotype rules see Inference methods.

Install¶

pip install genoabo

Quick start¶

genoabo cohort.pgen --out abo
# writes: abo.abo.tsv, abo.qc.tsv, abo.summary.txt, abo.report.html

If --out is omitted, the prefix is the input path with its last suffix stripped (cohort.pgen → cohort; cohort.vcf.gz → cohort.vcf).

Command-line reference¶

genoabo DATA [options]

--method auto picks the first method whose required markers are present: three-marker (rs8176719, rs8176746, rs8176747) → common (rs8176719, rs8176746) → ukb (rs8176746, rs505922; proxy mode, confidence capped). extended is never auto-selected. --strict-method does not change that order and does not substitute a smaller method when you pass --method NAME; it only errors on missing min_markers (or on auto with no match). Without --strict-method, --method auto with no match still proceeds as common (samples then get MISSING_MARKER). See Inference methods.

Default --assume-bo applies the common-allele rule that the O deletion arose on an A background, so an unphased double heterozygote is taken as B/O. --no-assume-bo leaves those samples ambiguous. The flag has no effect when phase is used. On phased 1000 Genomes 30x GRCh38, unphased double-hets were B/O in AMR/EAS/EUR and 105/113 in AFR (8/113 were A/O). For unphased African-ancestry data, --no-assume-bo is recommended. See 1000 Genomes --assume-bo check.

--geno-mode auto uses GP, then DS, then hardcall GT (GP/DS replace leftover phased GT). --phase auto is cohort-wide: if any fetched genotype is phased, every sample is run in phased mode. --build auto uses marker coordinates and VCF ##reference= / similar header tokens when present; conflicting evidence is an error and requires explicit --build. Chromosome-length matching is implemented in detect_build but is not wired from the caller.

Inspect and method metadata¶

genoabo cohort.pgen --inspect
genoabo --methods
genoabo --method-info three-marker

--inspect reports which ABO tags were found, call rates, and the recommended method. --methods and --method-info do not require an input file.

Input formats¶

GRCh37/hg19, GRCh38/hg38, and T2T-CHM13v2 are supported.

Outputs¶

A successful call writes four files:

.abo.tsv columns:

See Workflow for how genotypes move from file load to these files.

.qc.tsv is a metric/value table. Besides sample counts and marker call rates it includes group counts (n_A, n_B, n_AB, n_O), n_unknown (no phenotype: TSV NA or unknown), n_ambiguous (ABO_genotype = ambiguous; those rows are also in n_unknown), n_low_confidence, and percentages.

Python API¶

ABOCaller constructor kwargs match the CLI. call() returns a CallerResult; iterate result.calls for per-sample fields.

from genoabo import ABOCaller

caller = ABOCaller(
    method="auto",
    build="auto",
    genotype_mode="auto",
    phase="auto",
    format="auto",
    dosage_threshold=0.1,
    min_genotype_prob=0.9,
    strict_method=False,
    assume_bo=True,
)
result = caller.call("ukb.pgen")
for call in result.calls:
    print(call.sample.iid, call.abo_group, call.abo_genotype)

Inspect without writing calls:

result = caller.call("ukb.pgen", inspect_only=True)
print(result.build, result.method, result.inspect)

See also¶

• Workflow — load-to-output pipeline
• Inference methods — haplotype and phenotype rules
• Markers — coordinates and 1000 Genomes LD r²
• Assumptions — limitations
• Examples — synthetic and 1000 Genomes demos