Imputed gene expression risk scores: a functionally informed component of polygenic risk

Area of Science:

• Genetics
• Genomics
• Biostatistics

Background:

• Polygenic risk scores (PRS) are crucial for predicting disease risk.
• Integrating functional genomic annotations can enhance PRS utility.
• Gene expression data offers a promising avenue for refining genetic risk prediction.

Purpose of the Study:

• To evaluate the predictive utility of gene expression risk scores (GeRS) derived from imputed gene expression and transcriptome-wide association study (TWAS) results.
• To compare the performance of GeRS against traditional PRS.
• To assess the added value of combining GeRS with PRS for various health outcomes.

Main Methods:

• Calculated GeRS using 53 gene expression-genotype panels (SNP-weight sets) across diverse tissues.
• Employed elastic net models to assess predictive performance of GeRS, PRS, and combined models.
• Evaluated models in UK Biobank and Twins Early Development Study cohorts for 12 outcomes.
• Estimated the proportion of heritability explained by cis-regulated gene expression.

Main Results:

• GeRS significantly predicted multiple neuropsychiatric and anthropometric outcomes.
• Combining GeRS across different SNP-weight sets improved prediction.
• Models integrating both GeRS and PRS showed improved prediction for several outcomes, notably rheumatoid arthritis (up to 4%) and height (0.3%).
• The contribution of cis-regulated expression to overall heritability was generally modest.

Conclusions:

• GeRS can be a valuable component of PRS, aiding in the functional stratification of genetic risk.
• GeRS offer substantial predictive improvements over PRS alone only in specific contexts.
• Further research integrating functional genomic data into genetic risk prediction models is warranted.