Population Stratification Analysis in Genome-Wide Association Studies - Mathematical Approaches to Polymer Sequence Analysis and Related Problems - page 180

Biomedical Technologies, University of Milan. For the genome-wide association
study, approximately 750 ng of genomic DNA was used to genotype each sub-
ject for 317503 Phase I Hap Map tagging SNPs on the Infinium HumanHap300
BeadArrays (Illumina, San Diego, USA). Samples were processed according to
the Illumina Infinium 2 assay. Briefly, each sample was whole-genome amplified,
fragmented, precipitated and hybridized overnight for a minimum of 16 h at 48 ı C
to locus-specific probes on the BeadArray. Non-specifically hybridized fragments
were removed by washing while the remaining specifically hybridized DNA frag-
ments were processed for the single base extension reaction, stained and imaged on
an Illumina BeadArray Reader. Normalized bead intensity data obtained for each
sample were analysed with Illumina Beadstudio 2.0 software, which generated SNP
genotypes from fluorescent intensities using the manufacturer default cluster set-
tings [ 29 , 30 ]. After removal of SNPs with no calls and those with a minor allele
frequency less than 0.01, we were left with 297197 SNPs with an average call fre-
quency rate of 98.9%.

9.2.1.2 CATIE-NIMH Sample

The CATIE-NIMH sample contains 741 schizophrenics of the CATIE project
matched with 751 controls collected from the NIMH Genetics repository 25, whose
genotype and phenotype data are available to the scientific community ( www.
nimhgenetics.org ). GWAS genotyping was conducted by Perlegen Sciences using
the Affymetrix 500K “A” chipset (Nsp and Sty) and Perlegen custom 164K chip:
each subject was genotyped for 495172 SNPs. In terms of ethnicity, 56.17% of
subjects are Europeans, 29.62% Africans and 14.21% are selected as other or more
than one racial category (American-Indian/Alaska Native, Asian, Black/African-
American, Native Hawaiian/other Pacific Islander, White and Hispanic/Latino).

9.2.2 Stratification Approaches

To detect the possible PS, we applied different methods listed in Table 9.1 where
for each method, the number of markers used, the statistical method on which are

Tabl e 9. 1 Methods for population homogeneity test

Software # markers Statistical method Detection/correction

STRUCTURE Limited Bayesian approach Detection

FST Limited/GW F statistics Detection

Genomic control Limited/GW T statistics Both

PLINK Genome-wide Cochran-Mantel-Haenszel Both

EIGENSTRAT Genome-wide Armitage trend test Both

GW genome-wide