"IQS Imputation Quality Score
Imputation of untyped variants in genotyped data using sequenced reference panels is an interesting and cost-effective solution for association testing.
Imputation has been extensively used in the context of common variant association tests where it has shown its utility.
However, less studies of the interest and limits of imputation have been performed in the context of rare variant association tests using gene-based tests.
Imputation of rare variants is more difficult than imputation of common variants and performances are likely to differ from one gene to another. These differences of imputation efficiency between genes have not been well documented so far.
Material and Methods
- Imputation was performed on the Sanger Imputation Server with the HRC.1 reference panel and using Eagle2 and pbwt
- Genotypes were imputed from data on a SNP-chip for invidivuals from FrEx
- 603,845 variants on the SNP-chip and 563 people passed filters and QC and were sent to the Sanger Imputation Server
- Only the imputed variants that were located in the genomic regions covered by the exome were considered
- Imputed genotypes were compared against exome genotypes that were considered the truth
- Singletons found in the exome data were excluded from the analysis as well as indel
- The Imputation Quality Score (IQS) (Lin et al. Plos One 2010) that takes into account per chance concordance as in the Cohen’s kappa and also account for posterior probabilities of genotypes was used to measure imputation accuracy