Enhanced Statistical Tests for GWAS in Admixed Populations
Author Information
Author(s): Pasaniuc Bogdan, Zaitlen Noah, Lettre Guillaume, Chen Gary K., Tandon Arti, Kao W. H. Linda, Ruczinski Ingo, Fornage Myriam, Siscovick David S., Zhu Xiaofeng, Larkin Emma, Lange Leslie A., Cupples L. Adrienne, Yang Qiong, Akylbekova Ermeg L., Musani Solomon K., Divers Jasmin, Mychaleckyj Joe, Li Mingyao, Papanicolaou George J., Millikan Robert C., Ambrosone Christine B., John Esther M., Bernstein Leslie, Zheng Wei, Hu Jennifer J., Ziegler Regina G., Nyante Sarah J., Bandera Elisa V., Ingles Sue A., Press Michael F., Chanock Stephen J., Deming Sandra L., Rodriguez-Gil Jorge L., Palmer Cameron D., Buxbaum Sarah, Ekunwe Lynette, Hirschhorn Joel N., Henderson Brian E., Myers Simon, Haiman Christopher A., Reich David, Patterson Nick, Wilson James G., Price Alkes L.
Primary Institution: Harvard School of Public Health
Hypothesis
Can combining SNP and admixture association improve statistical power in genome-wide association studies (GWAS) for admixed populations?
Conclusion
The new statistical methods significantly increase the power to identify disease risk loci in admixed populations compared to standard methods.
Supporting Evidence
- The new methods showed an 8% increase in statistical power for finding disease risk loci compared to standard methods.
- At imputed SNPs, there was an 11% increase in statistical power when using the local ancestry-aware imputation framework.
- The study utilized data from both simulated and real phenotypes to validate the new statistical framework.
- Results indicated that the combined approach is broadly applicable to GWAS in admixed populations.
Takeaway
This study shows that using a new way to analyze genetic data from mixed populations helps scientists find links between genes and diseases better than before.
Methodology
The study analyzed data from 6,209 unrelated African Americans using a new statistical framework that combines SNP and admixture association.
Potential Biases
Potential biases may arise from the imputation methods used for untyped SNPs.
Limitations
The study primarily focuses on admixed populations and may not be applicable to homogeneous populations.
Participant Demographics
Participants were unrelated African Americans from the CARe project.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website