Inference methods¶

genoSecretor maps FUT2 tag SNPs to haplotype alleles Se (functional), se (complete LoF, se428 / W143X), and Se^w (reduced activity, se385 / I129F), then to a recessive phenotype.

Research use only. These methods type common alleles, not the full FUT2 allelic series. FUT3/Lewis typing is out of scope.

Haplotype rules¶

For each chromosome copy:

1. If rs601338 = LOF → haplotype = se (epistatic: truncated FUT2 is null even if rs1047781 is also WEAK)
2. Else if rs1047781 = WEAK → haplotype = Se^w
3. Else if either marker is FUNCTIONAL → haplotype = Se
4. Else if a leftover (unmodelled) state is present → haplotype = Se
5. Both markers missing on that copy → no haplotype

Figure. One chromosome copy: LOF yields se; otherwise WEAK yields Se^w and FUNCTIONAL (or any leftover unmodelled state) yields Se. Both markers missing on that copy is not a haplotype. Under proxy, rs601338 / rs602662 copy-count disagreement is CONFLICT, not a forced haplotype.

Phenotype table¶

--phenotype-coding biochemical (default) uses this table. --phenotype-coding gwas recodes Se^w/Se^w and Se^w/se as non-secretor (Smyth 2011 / BBJ-style binary trait). fut2_genotype still keeps Se^w.

Figure. Sorted haplotypes (Se < Se^w < se) map to phenotype: any Se copy → secretor; Se^w/Se^w and Se^w/se → weak-secretor; se/se → non-secretor.

How much residual FUT2 activity Se^w retains versus Se is assay-dependent (~2–5% lysate activity, ~22% remaining Vmax, ~19% of wild-type H-antigen-positive cells, and still above complete-null se428). See How weak is Se^w.

two-marker (default when both SNPs are present)¶

Uses rs601338 and rs1047781.

Unphased double heterozygotes (het at both SNPs) are assigned Se^w/se under --assume-trans (default), because se428 and se385 typically sit on different continental haplotypes. --no-assume-trans leaves those samples ambiguous (Se^w/se vs cis Se/se).

Figure. Unphased LOF/FUNCTIONAL × FUNCTIONAL/WEAK is compatible with trans Se^w/se or cis Se/se (the double-mutant copy is still se). Default --assume-trans assigns Se^w/se; --no-assume-trans leaves the call ambiguous. Phased GT does not use this assumption.

common¶

rs601338 only. Homozygous LOF → se/se (non-secretor); otherwise secretor. Never emits weak-secretor. Poor choice as the sole method in East Asian cohorts.

eas¶

rs1047781 only. Homozygous WEAK → Se^w/Se^w (weak-secretor); otherwise secretor. Never emits se/se from W143X. Poor choice as the sole method in European/African cohorts.

proxy¶

If rs601338 is usable, same as common. If it is missing, rs602662 SE_PROXY (A) is mapped to an LOF-like state and confidence is low. If both markers are present and copy counts disagree, QC status is CONFLICT.

Auto-selection¶

--method auto still prefers two-marker when both functional SNPs are present. After that, allowed fallbacks depend on --ancestry:

--ancestry auto (default) infers EAS-like vs EUR-like only when a single functional SNP is present and n ≥ 50 non-missing genotypes:

• rs601338 present, rs1047781 missing, se428 AF < 0.05 → treat as EAS and refuse common
• rs1047781 present, rs601338 missing, se385 AF < 0.05 → treat as EUR-like and refuse eas

n < 50 is not inferred (behaves like unknown) unless --ancestry eas (or eur / afr / amr) is set. Explicit --method still runs and prints a warning if it contradicts ancestry.

--strict-method disables fallback when a requested method's minimum markers are missing.

List methods:

genosecretor --methods
genosecretor --method-info two-marker