Genomic Claims You Can Trace

 BGPT Odds of True

The hypothesis claims LOEUF shifts between HBG1/HBG2 reflect version-dependent alignment/callability artifacts. Supplied sources confirm extreme HBG1/HBG2 paralog homology and variant/assignment ambiguity, making artifacts plausible; but no source verifies the LOEUF values or gnomAD release metadata, so confidence is moderate.

 Hypothesis Novelty

Paralog-alignment artifacts affecting constraint metrics are a known methodological concern, but applying this critique specifically to the HBG1/HBG2 LOEUF divergence in gnomAD v4.1 is a specific, underexplored methodological question.

 Quick Analysis Plan Copied

 Long Analysis Plan

Evaluating the paralog-artifact hypothesis for HBG2/HBG1 LOEUF divergence

Verification gap: The specific LOEUF values (HBG2 0.83→0.38; HBG1 1.13→1.64) are stated in the hypothesis but are not present in any supplied source, so they cannot be independently confirmed here. The analysis below evaluates the mechanistic plausibility using the supplied records.

Support for the artifact interpretation: HBG1 and HBG2 encode nearly identical γ-globin proteins and sit in a duplicated, recombination-prone locus; the supplied records document this homology burden concretely. CRISPR guide design at HBG1/HBG2 shows pervasive variant-driven ambiguity — most guides are impacted by population variants, and haplotype-resolved analysis is required to even assign guides between paralogs Variant-aware HBG1/HBG2 guide ambiguity. Similarly, even a carefully designed HBG1/2 sgRNA in human embryos produced off-target editing at 11 of 14 homologous sites, demonstrating that sequence-level assignment between these paralogs is intrinsically error-prone HBG1/2 sgRNA off-target homology editing. Short-read exome alignment between such paralogs produces read-misassignment that changes with reference build, alignment pipeline, and callable-region definitions across gnomAD releases — exactly the kind of technical change that can move LOEUF in opposite directions for two near-identical paralogs without any change in true constraint.

Biology argues against a true constraint change: γ-globin genes are not under strong germline LoF constraint in the classic sense — HBG1/HBG2 are developmentally silenced after birth, and no mechanism would plausibly alter HBG2 constraint (not HBG1's) between consecutive gnomAD versions. Functional HBG2 studies instead show promoter-level regulatory roles (e.g., a single-base distal-promoter insertion reactivating γ-globin without off-target effects HBG2 promoter therapeutic insertion study), not protein-level intolerances that could shift LOEUF.

Counterpoints and blindspots: An alternative is that v4.1's larger, more diverse cohort genuinely improved variant discovery, and because HBG2 (Aγ vs Gγ expression dynamics differ) tolerates rare pLoF variants differently, real constraint differences could exist. Also, the HBG2 distal promoter is functionally important, so variant annotation changes at promoter-proximal exonic boundaries could be genuine HBG2 promoter functional regulation. The decisive test would be direct comparison of callable exonic length, observed/expected pLoF counts, and alignment mappability scores for HBG1 vs HBG2 exons between the two gnomAD releases — data not supplied here.

Verdict (moderate confidence): The bidirectional divergence of LOEUF between two >99%-identical paralogs in adjacent releases is a classic signature of alignment/callability artifacts rather than biology; the artifact interpretation is plausible but unproven without the gnomAD release-level metadata.