5 Methods for Association Studies
101
decision-making to achieve the highest quality and most relevant results. Here we
discuss some of the most common tools utilized to measure genetic information and
some considerations to contemplate in deciding on methods.
5.4.1 Common Variants
SNP genotyping arrays are currently the most pervasive tool utilized for evaluating
genetic information. GWAS in particular typically employ microarray-based tag
SNP genotyping techniques that capture common variation in the human genome.
The arrays for early GWAS generally contained between 100,000 and 500,000
variants identified in databases such as HapMap. More current chips include
approximately one million or more variants. No matter the set of variants, the array
is then typed in a specific set of individuals. The arrays have generally been designed
to measure variants at or above a minor allele frequency of 5%, though they may
even miss some common variation (Jorgenson and Witte 2006). More recently,
however, microarrays have been designed to detect variants down to a minor allele
frequency of 1% (Hoffmann et al. 2011a, b). The platforms most often come from
one of two companies: Illumina (San Diego, CA) or Affymetrix (Santa Clara, CA;
now owned by Thermo Fisher Scientific) (Hindorff et al. 2009).
One consideration in designing a chip for assessment is determining the set of
SNPs required to capture common variation in the population of interest. Consider,
for example, the number of SNPs required to capture variation across the genome
for African versus European populations. When the latter emigrated from Africa,
they experienced a bottleneck that reduced the population size and resulting genetic
variation. They thus have more LD than the former (see Chap. 8 on human
demographic history). As a result, the chip used for a study of an African population
requires more SNPs to obtain the same overall genomic coverage.
5.4.2 Rare Variants
The recent considerable expansion of the human population and negative selection
of deleterious alleles over time have resulted in low allele frequencies for many
disease-causing variants. Consequently, rare variants with substantial effects may
remain untyped by standard genotyping assays. In addition, whole-genome sequencing is not generally cost-effective for the evaluation of rare variants, because the
sample sizes required for association studies are normally much too large. More
effective methods to identify rare disease-causing variants involve utilizing exome
sequencing or exome genotyping arrays to investigate coding variation, even though
these approaches ignore potentially important parts of the genome.
Sequencing and capture technologies are now able to accurately determine the
sequence of nearly all protein-coding variants in humans (Choi et al. 2009; Gnirke
et al. 2009; Ng et al. 2009; Teer and Mullikin 2010). They allow researchers to
detect and genotype variants found in particular individuals without requiring that
101
decision-making to achieve the highest quality and most relevant results. Here we
discuss some of the most common tools utilized to measure genetic information and
some considerations to contemplate in deciding on methods.
5.4.1 Common Variants
SNP genotyping arrays are currently the most pervasive tool utilized for evaluating
genetic information. GWAS in particular typically employ microarray-based tag
SNP genotyping techniques that capture common variation in the human genome.
The arrays for early GWAS generally contained between 100,000 and 500,000
variants identified in databases such as HapMap. More current chips include
approximately one million or more variants. No matter the set of variants, the array
is then typed in a specific set of individuals. The arrays have generally been designed
to measure variants at or above a minor allele frequency of 5%, though they may
even miss some common variation (Jorgenson and Witte 2006). More recently,
however, microarrays have been designed to detect variants down to a minor allele
frequency of 1% (Hoffmann et al. 2011a, b). The platforms most often come from
one of two companies: Illumina (San Diego, CA) or Affymetrix (Santa Clara, CA;
now owned by Thermo Fisher Scientific) (Hindorff et al. 2009).
One consideration in designing a chip for assessment is determining the set of
SNPs required to capture common variation in the population of interest. Consider,
for example, the number of SNPs required to capture variation across the genome
for African versus European populations. When the latter emigrated from Africa,
they experienced a bottleneck that reduced the population size and resulting genetic
variation. They thus have more LD than the former (see Chap. 8 on human
demographic history). As a result, the chip used for a study of an African population
requires more SNPs to obtain the same overall genomic coverage.
5.4.2 Rare Variants
The recent considerable expansion of the human population and negative selection
of deleterious alleles over time have resulted in low allele frequencies for many
disease-causing variants. Consequently, rare variants with substantial effects may
remain untyped by standard genotyping assays. In addition, whole-genome sequencing is not generally cost-effective for the evaluation of rare variants, because the
sample sizes required for association studies are normally much too large. More
effective methods to identify rare disease-causing variants involve utilizing exome
sequencing or exome genotyping arrays to investigate coding variation, even though
these approaches ignore potentially important parts of the genome.
Sequencing and capture technologies are now able to accurately determine the
sequence of nearly all protein-coding variants in humans (Choi et al. 2009; Gnirke
et al. 2009; Ng et al. 2009; Teer and Mullikin 2010). They allow researchers to
detect and genotype variants found in particular individuals without requiring that