NEW YORK – A research team led by investigators at the University of Michigan has developed an analytical method for boosting the insights that can be gleaned from transcriptome-wide association studies (TWAS) in individuals from multiple population backgrounds.
"[A]lmost all existing TWAS methods have been thus far focused on using expression studies collected on individuals from a single genetic ancestry, typically European ancestry," senior and corresponding author Xiang Zhou, a biostatistics researcher at the University of Michigan's School of Public Health, and his colleagues explained, adding that "[c]urrent TWAS methods are unable to take advantage of the many recent expression studies performed in multiple genetic ancestries."
As they reported in the American Journal of Human Genetics on Thursday, the investigators came up with a "multi-ancestry transcriptome-wide analysis" (METRO) approach to tap into these growing sets of gene expression data from individuals in a range of ancestry groups. In both simulated and real datasets, they demonstrated that this approach could improve the detection of informative associations, bumping up TWAS statistical power.
The method "is capable of inferring the contribution of expression prediction models in different genetic ancestries toward explaining and informing the gene-trait association, allowing us to interrogate the ancestry-dependent transcriptomic mechanisms underlying gene-trait association," the authors reported, noting that "[w]e illustrate the benefits of METRO in both simulations and applications to seven complex traits and diseases obtained from four GWAS."
Using the METRO method, the team combined multi-ancestry gene expression profiles from large-scale efforts such as the "Genetic Epidemiology Network of Arteriopathy" (GENOA) study with data from four GWAS for complex traits or conditions. For two of the GWAS, most of the 42,752 cases and 23,827 controls had African American ancestry. European ancestry predominated in the two other GWAS, which involved 188,577 cases and 339,226 controls.
Together with new gene expression and genotyping data, these datasets led to gene-trait ties that were not found with other TWAS methods, the researchers reported. They found that the enhanced ability to detect gene-trait associations with METRO was especially pronounced in the studies involving individuals with African ancestry.
"The benefits of METRO are most prominent in applications to GWAS of African ancestry where sample size is typically small," the authors explained, "and thus a powerful TWAS method is crucial for identifying gene-trait associations."
When it came to complex traits or conditions — including several blood lipid profiles and traits such as body mass index or type 2 diabetes — the team unearthed previously unappreciated pathways and regulatory features by digging into TWAS associations identified by METRO.
The authors cautioned that the current study "focused on demonstrating the benefits of METRO in simulations and real data applications with only two ancestries, due to the limited data availability of other ancestries."
"While METRO is a general modeling framework that can be directly applied to expression studies with more than two ancestries," they wrote, "we note that examining the performance of METRO for TWAS analysis in more than two ancestries remains an important direction in the future."