Most GWAS have been conducted in populations of European ancestries, but these results do not reflect the global population or replicate well in non-European populations. Additionally, investigating traits at the proteome level may provide more insight to biological mechanisms than at the genome level. Using data from the TOPMed consortium, we have built protein models to perform PWAS using S-PrediXcan in published multiethnic GWAS data from the PAGE study (Wojcik et al 2019). This output reveals significant associations between genes and a variety of complex traits in non-European populations.