Sex differences in the genetic architecture of common diseases are rare and often driven by phenotypic artefacts

### DescriptionThis repository contains the conference poster titled **"Sex differences in the genetic architecture of common diseases are rare and often driven by phenotypic artefacts"**, presented by Samuel E. Jones, Raymond K. Walters, Reza Jabal, FinnGen, Mark Daly, and Mitja I. Kurki. #### Background & AimMany common complex diseases differ between men and men in prevalence, onset, and severity. While some recent large biobank studies have reported widespread sex-heterogeneous genetic effects, the sensitivity of these tests to quality control and analytic choices remains contested. The aim of this work is to characterize how many truly sex-heterogeneous signals remain when each disease endpoint's clinical definition, diagnostic practice, and population ascertainment are rigorously examined. #### Methods & DataUsing the FinnGen R12 dataset (500,348 individuals; 282,064 women and 218,284 men), we evaluated sex-stratified Genome-Wide Association Studies (GWAS) using REGENIE v3.3 across 559 non-overlapping disease endpoints (each with ≥1,000 cases per sex). We assessed between-sex heterogeneity using three core measures:* Heritability difference (HerDif)* Genetic correlation between sexes (rg < 1)* Variant-level heterogeneity (sdVar)Replication analyses were conducted across 25 matching endpoints within the UK Biobank. #### Key Findings* **Rare True Heterogeneity:** Only 26 of 536 (4.8%) evaluable endpoints showed initial evidence of sex-heterogeneity across 9 unique loci. After strict evaluation, only 8 candidate endpoints remained, with just 2 replicating in the UK Biobank (cholelithiasis at the UGT1A6 locus, and coxarthrosis).* **Phenotypic Artefacts:** The vast majority of apparent genetic sex-heterogeneity is driven by three distinct types of phenotypic bias rather than dimorphic biology: 1. *Endpoint composition bias:* Composite codes mixing distinct clinical conditions with unequal sex ratios (e.g., hernias). 2. *Diagnostic conflation bias:* The same ICD code covering fundamentally different diseases or causes in each sex (e.g., urinary retention). 3. *Differential ascertainment bias:* Sex-unequal diagnosis thresholds or variations in time-at-risk altering the overall case mix (e.g., depression). #### Main ConclusionGenuine sex-specific genetic effects in common complex diseases are exceptionally rare. Most apparent differences reflect the complex structural ways in which clinical cases are defined, diagnosed, and tracked in large-scale biobanks.