Genotype analysis 85!

The SNPs included in this analysis were genotyped at two different times using two different genotyping platforms. Most SNPs were evaluated with a Custom GoldenGate

Genotyping assay from Illumina (Illumina, Inc., San Diego, CA). A few SNPs that failed Illumina genotyping analysis were reassessed using a Taqman panel. The Taqman analysis also included a number of GWAS hits identified in the time since the Illumina panel was performed.

These analyses are used to determine whether an individual has two copies of the common allele, two copies of the variant allele, or one copy of each. These three possible genotypes are also described as homozygous for the major allele, homozygous for the minor allele, or heterozygous. In some cases, the allele that is more common in whites was the rare allele in African Americans.

3.1.5.1 Illumina assays

Initially, 1762 SNPs were submitted to Illumina for review and preliminary quality control assessment. Illumina validated each SNP according to its dbSNP identification number [278, 279], and provided designability scores to indicate the probability that each SNP would be genotyped successfully on their platform. 1365 SNPs passed this initial review process. To replace the failed SNPs, CBCS investigators selected the best possible substitutes and resubmitted the SNP panel to Illumina until a complete set of 1536 SNPs with passable design scores was identified.

Once the custom solid-phase bead assay was received from Illumina, the 3857 CBCS participants with available blood samples were genotyped for all 1536 SNPs in UNC’s Mammalian Genotyping Core lab. At UNC, 2 µg of participant DNA were extracted, then activated and combined with assay oligonucleotides, hybridization buffer, and paramagnetic particles [280]. Each assay contained three locus-specific oligonucleotides for each SNP, two of which bonded directly to the allele sites, and a third that hybridized downstream from the

other two. The third oligonucleotide contained a unique address sequence that corresponded to a particular bead type.

After several wash steps and the addition of DNA ligase, the oligonucleotide-tagged sequences formed double-stranded DNA fragments. These fragments were amplified with dyed polymerase chain reaction (PCR) primers such that each allele of a given SNP was uniquely labeled. Lastly, the finalized, single-stranded, dye-labeled DNA products were hybridized to the bead type specified by the address sequence contained in the third oligonucleotide. In this final form, the fluorescence signal of each bead, and thus the genotype of each SNP, could be measured using Illumina’s BeadArray Reader.

3.1.5.2 Illumina Quality Control

CBCS investigators used several quality control techniques to measure and improve overall data accuracy. This included the use of blind duplicate samples, lab controls, the examination of individual call rates, and careful inspection of assay intensity data and genotype clustering images.

Roughly four percent (169 of 3857) of samples underwent duplicate, blinded genotyping. Of more then 200,000 possible discrepancies, 11 genotype miscalls were identified in these samples. DNA samples from the HapMap CEU population (provided by the Coriell Institute for Medical Research) were also repeatedly genotyped as lab controls, with only 2 discrepancies in 184 replications. As these error rates were well-within

reasonable limits, no SNPs were excluded based on these results. However, closer inspection of assay intensity data and genotype clustering images did lead to the exclusion of 163 of the 1536 genotyped SNPs (11%). Here, SNPs were excluded from further analyses if they showed low signal intensity or if genotype clusters were overlapping, as both are indications

of genotyping error.

Individuals were excluded from further analyses if they had call rates of less than 95% for the 1383 remaining SNPs. 103 participants met this exclusion criteria. Six other subjects were omitted, including 5 genotyped as male and 1 with conflicting duplicate samples. Case, race, age, and stage distributions were similar for excluded and included subjects [273]. After exclusions, genotyping information was available for 3748 participants (1972 cases, 1776 controls) and 1373 SNPs, including 57 of the 83 candidate SNPs selected for this analysis and 144 AIMs.

3.1.5.3 Taqman assays

The remaining 26 candidate SNPs were genotyped using Applied Biosystems’ (Foster City, CA) fluorogenic 5’ nuclease assay (‘TaqMan®’) and PRISM® 7900HT Sequence Detection System in UNC’s Mammalian Genotyping Core lab. As with the Illumina assays, the TaqMan assays were customized to target specific candidate SNPs. However, instead of oligonucleotides, TaqMan assays contain site-specific primers and fluorescently labeled probes for each selected SNP. TaqMan is more appropriate for genotyping a small number of SNPs in up to several thousand samples.

The genotyping process begins when activated DNA is combined with the custom- designed SNP genotyping assay and the Taqman Universal PCR Master Mix. During the PCR process, which consists of 10 minutes at 95˚ C, followed by 40 cycles of 15 seconds at 92˚ C and 1 minute at 60˚ C, the AmpliTaq Gold® DNA polymerase contained in the master mix amplifies genomic regions where the primers and probes have annealed [281]. As the probe is digested by the polymerase, it releases a unique fluorescence signal which is read using the PRISM® 7900HT Sequence Detection System.

3.1.5.4 Taqman Quality Control

Approximately 10% of the CBCS samples were run as blind duplicates, all of which were in agreement. Although several SNPs failed the initial design phase, all of the

genotyped SNPs met quality controls standards. No additional participants were removed due to low call rates.

3.1.5.5 SNP exclusions

I excluded six SNPs with MAFs less than 1% in white participants and ten SNPs with MAF less than 1% in African Americans, leaving a total of 77 evaluable SNPs in whites and 73 in African Americans. 79 SNPs with MAF greater than 1% in the non-racially stratified CBCS population were evaluated in subtype analyses.