Assumptions and limitations¶

Research use only. genoSecretor infers FUT2 secretor status from a small set of tag SNPs. It is not a serological or clinical test.

What is modelled¶

• Recessive FUT2: two loss-of-function / weak haplotypes are required for non-secretor or weak-secretor phenotypes.
• se428 (rs601338 W143X) as a complete null.
• se385 (rs1047781 I129F) as a reduced-activity (weak) allele. Under default --phenotype-coding biochemical, homozygous or compound-with-se genotypes are weak-secretor. --phenotype-coding gwas recodes those calls as non-secretor.
• rs602662 as an LD proxy for se428 only.

What is not modelled¶

• Rare FUT2 nulls, fusions, and regulatory alleles.
• FUT3 / Lewis (Le^a / Le^b) phenotypes.
• Salivary ABH serology or norovirus binding assays.
• Calibrated probabilities (confidence is ordinal: high / medium / low / none).

How weak is Se^w¶

se385 (rs1047781 I129F) is not a complete FUT2 null. Published COS-cell assays put residual Se^w function well below wild-type Se, but the number depends on the readout. Complete-null se428 (W143X) is at or near zero in the same experiments.

Figure. Percent of wild-type Se in the same assay. Lysate α(1,2)fucosyltransferase activity is ~2–5% of Se (Kudo et al. 1996; Koda et al. 2001). Remaining Vmax is ~22% with a similar Km; the I129F protein is less stable (Henry et al. 1996; Vmax 230 vs 1030 pmol h⁻¹ mg⁻¹). Cell-surface H-antigen-positive cells are ~19% of Se and ~7× se428 (Soejima and Koda 2021; 4.86 ± 0.76% vs 26.22 ± 5.06% vs 0.68 ± 0.19%). Not a saliva titre.

In vivo, Se^w/Se^w and Se^w/se are partial secretors / Le(a+b+): salivary ABH is reduced but often still detectable (Yu et al. 1995; Henry et al. 1996; Henry 2014). There is no community-standard saliva ABH fold-change for Se^w/Se^w versus Se/Se. Milk 2'-FL papers in East Asian cohorts often recode TT as non-secretor, or (incorrectly) label AT heterozygotes as weak-secretor; those medians are not a Se^w haplotype residual and are not plotted.

Weak-secretor vs GWAS binary coding¶

Because residual activity is low, many GWAS papers treat homozygous se385 (rs1047781 T/T) as non-secretor (Smyth et al. 2011; He et al. 2023 BBJ peptic-ulcer coding). Blood-group biochemistry still distinguishes weak-secretor from se/se.

• --phenotype-coding biochemical (default): Se^w/Se^w and Se^w/se → weak-secretor; se/se → non-secretor.
• --phenotype-coding gwas: those Se^w diplotypes are recoded as non-secretor. Heterozygotes with a functional Se copy (Se/Se^w, Se/se) stay secretor.

fut2_genotype always keeps Se^w. Python: SecretorCaller(phenotype_coding="gwas").

--assume-trans¶

Unphased heterozygotes at both functional SNPs are compatible with trans Se^w/se (weak-secretor) or cis Se plus a double-mutant haplotype (secretor). The default assigns trans. Pass --no-assume-trans to leave those samples ambiguous.

In 1000 Genomes 30x hg38, that unphased pattern is rare because se428 and se385 almost never co-occur (zero double-hets in AFR/AMR/EUR; one trans double-het in EAS). See 1000 Genomes --assume-trans. That check uses phase, not salivary serology.

Proxy LD¶

rs602662 is strongly correlated with rs601338 in European cohorts and is a poorer tag in populations where se385 (not se428) is the common reduced-activity allele. Proxy calls are always low confidence.

Ancestry and auto-selection¶

--ancestry auto infers East Asian–like vs European-like status from FUT2 derived-allele frequency (se428 / se385), not from PCA or self-reported ancestry. It only fires when a single functional SNP is present and n ≥ 50. East Asian secretor typing still requires rs1047781 in the genotype file; common and proxy remain inappropriate as the sole method in East Asian cohorts. Pass --ancestry eas (or eur / afr / amr) to skip the AF heuristic. See Inference methods.

Validation¶

genoSecretor has not been validated against salivary ABH typing across ancestries.