Imputation of missing genotypes: an empirical evaluation of IMPUTE
2008

Evaluating the Accuracy of Genotype Imputation

Sample size: 270 publication Evidence: moderate

Author Information

Author(s): Zhao Zhenming, Timofeev Nadia, Hartley Stephen W, Chui David HK, Fucharoen Supan, Perls Thomas T, Steinberg Martin H, Baldwin Clinton T, Sebastiani Paola

Primary Institution: Boston University School of Public Health

Hypothesis

How do ethnic background, subject ascertainment, and missing data affect the accuracy of genotype imputation?

Conclusion

IMPUTE is very accurate for Caucasian samples, slightly less accurate for Asian samples, and significantly less accurate for admixed backgrounds like African Americans.

Supporting Evidence

  • The program IMPUTE showed 97% median accuracy in Caucasian subjects with less than 10% missing SNPs.
  • Accuracy increased to 99% when requiring a minimum posterior probability of 0.95 for imputed genotypes.
  • Accuracy decreased to 86% or 94% for African American and Asian subjects, respectively.

Takeaway

This study looks at how well a computer program can guess missing genetic information, finding it works best for white people and not as well for mixed backgrounds.

Methodology

The study used publicly available genotype data from various ethnic groups and assessed the accuracy of the IMPUTE program by comparing observed and imputed genotypes.

Potential Biases

The choice of reference populations may introduce bias in the accuracy of imputation.

Limitations

The study did not consider the differences in SNP selection methods between Affymetrix and Illumina platforms, which may affect accuracy.

Participant Demographics

Participants included Caucasians, African Americans, and Asians, with varying sample sizes.

Digital Object Identifier (DOI)

10.1186/1471-2156-9-85

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication