Purchase this article with an account.
Michelle Grunin, Karen Y He, Robert Igo, Yeunjoo E Song, Mathias Gorski, Iris M Heid, Xiaofeng Zhu, Jonathan L Haines; Dense SNP IAMDGC Dataset Developed through Imputation. Invest. Ophthalmol. Vis. Sci. 2020;61(7):3517.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The International Age-Related Macular Degeneration Genomics Consortium (IAMDGC) published a full genome-wide association study (GWAS) on 16,144 AMD individuals and 17,832 controls of European descent (Fritsche et al., 2016). However, the IAMDGC was a multi-ethnic cohort, and many of the participants could not have been imputed correctly with the imputation reference panels available at that time.
We set out to re-impute the IAMDGC data using the Trans-Omics for Precision Medicine (TOPMed) Program imputation panel and reanalyze the data including multi-ethnic participants to not only improve the imputation quality and diversity of the study, but also to increase the power of the study. Variant filtering was utilized in the same way as the original IAMDGC analysis, including noting those samples with whole genome amplification (WGA) and the variants that differ between WGA and blood samples, and including quality control applied to males and females separately.
52,189 individuals (30,886 AMD individuals, 21,303 controls) were collected via the IAMDGC, including those classified as non-European and related individuals. Of the 569,645 variants on the genotyped SNP chip, 85,591 variants were used to investigate the ancestry of the individuals via principal components analysis and comparison with 2,504 individuals of 27 ancestries from the 1000 Genomes Project (1KG) Phase 3 using PLINK. These 1KG samples were added to the dataset for quality control, for a total of 54,693 individuals to be imputed. We also imputed the X chromosome using the XWAS software suite. We performed phasing with Eagle, and imputation using the Michigan Imputation Server (MIS) and Minimac4 using the TOPMed Freeze 5 reference panel, including imputation of the X chromosome.
We estimate an improvement in imputation, according to the latest TOPMed data, of 50 percent as compared to the original imputation with the 1KG panel) of the variants with minor allele frequencies less than 0.5%. Using this methodology, we can investigate rare variants that were previously unable to be imputed, as well as have a more accurately imputed multi-ethnic cohort from the original IAMDGC study for further GWAS.
This is a 2020 ARVO Annual Meeting abstract.
This PDF is available to Subscribers Only