We aimed to prioritise genes associated with AN and implicate potential mechanisms related to expression and/or splicing. Firstly, we conducted a comprehensive brain and blood-based transcriptome-wide association study (TWAS) (Gamazon et al., 2015; Gusev et al., 2016; Reay & Cairns, 2021). TWAS requires expression data which can be imputed from independent SNP-mRNA expression weights from multivariate models of cis-acting genetically regulated expression (GReX). TWAS then compares imputed expression with SNP-AN effect sizes to test the association between predicted expression and the odds of AN. Genes uncovered from this approach that survived multiple-testing correction were then further probed to refine candidate causal genes through conditional analysis and probabilistic finemapping (Gusev et al., 2016; Mancuso et al., 2019). Whilst mRNA expression is arguably the most well studied cellular readout, genes may operate more specifically in the pathogenesis of AN through dysregulation of other factors like protein expression and alternative splicing. As a result, we leveraged SNP weights, where available, for protein expression and splicing to also perform a proteome-wide association study (PWAS) and an alternative splicing based test (spliceWAS). Notably, PWAS and
