Multi-ancestry gene expression models amplify transcriptome-wide association study discovery and validation.

Area of Science:

• Genetics
• Psychiatry
• Bioinformatics

Background:

• Understanding genetic influences on gene expression across diverse populations is crucial but limited.
• Existing gene expression models often lack representation of multiple ancestral backgrounds, potentially biasing results.

Purpose of the Study:

• To investigate the impact of ancestral diversity on genetically determined expression in psychiatric conditions.
• To compare the efficacy of admixed versus predominantly European ancestry gene expression models in transcriptome-wide association studies (TWAS).

Main Methods:

• Performed transcriptome-wide association studies (TWAS) for 6 psychiatric conditions using gene expression models trained on admixed ancestral cohorts (African, European, Indigenous American).
• Repeated TWAS using a model trained on predominantly European ancestry individuals for comparative analysis.
• Assessed gene-level associations using false discovery rate (FDR)-adjusted p-values < 0.05.

Main Results:

• Identified 1416 statistically significant gene-level associations across 6 psychiatric diagnoses.
• 62% of these associations were uniquely detected by the admixed gene expression models.
• Observed high correlation (>0.92) in gene-level disease risk effects across ancestries, even for results significant in only one population.
• Genes identified by admixed models implicated more neurophysiological features associated with diagnostic symptoms.

Conclusions:

• Admixed gene expression models substantially increase the yield and validation of TWAS findings.
• These models enable more precise mapping of genetic effects to pathophysiological mechanisms in psychiatric disorders.
• The findings highlight potential new avenues for therapeutic development by considering diverse genetic ancestries.