# Imputation provides an opportunity to study filaggrin ( FLG) null mutations in large population cohorts that lack bespoke genotyping

## TL;DR

This study shows that imputed genetic data can reliably detect FLG gene mutations linked to atopic dermatitis in large UK population studies.

## Contribution

The study demonstrates that genome-wide imputed data can effectively capture FLG null mutations for population-scale genetic analysis.

## Key findings

- FLG null mutations are well imputed with 0.3% discordance using HRC data.
- Imputed FLG mutations remain strongly associated with atopic dermatitis in UK Biobank.
- Wild-type alleles impute more reliably than null alleles in imputed datasets.

## Abstract

Background: Low frequency mutations within the filaggrin (
FLG) gene are established genetic risk factors for atopic dermatitis. Studies of
FLG have typically used sequencing or bespoke genotyping. Large-scale population cohorts with genome-wide imputed data offer powerful genetic analysis opportunities, but bespoke
FLG genotyping is often not feasible in such studies. Therefore, we aimed to determine the quality of selected
FLG null genotype data extracted from genome-wide imputed sources, focussing on UK population data.

Methods: We compared the allele frequencies of three
FLG null mutations (R501X, R2447X and S3247X) in directly genotyped and genome-wide imputed data in the ALSPAC cohort. Logistic regression analysis was used to test the association of atopic dermatitis with imputed and genotyped
FLG null mutations in ALSPAC and UK Biobank to investigate the usefulness of imputed
FLG data.

Results: The three
FLG null mutations appear to be well imputed in datasets that use the Haplotype Reference Consortium (HRC) for imputation (0.3% discordance compared with directly genotyped data). However, a greater proportion of null alleles failed imputation compared to wild-type alleles. Despite the calling of
FLG mutations in imputed data being imperfect, they are still strongly associated with atopic dermatitis (p-values between 7x10
-10 and 5x10
-75 in UK Biobank).

Conclusions: HRC imputed data appears to be adequate for UK population-based genetic analysis of selected
FLG null mutations.