Printer-friendly version

In some studies, SNPs or haplotypes are expected to be the causative agents in the phenotype of interest.  For these types of studies, it is important to include all the SNPs (in the genomic region of interest).  Once SNPs are found that are associated with the phenotype, more detailed analysis of the gene and protein networks are done to understand the processes involved.

However, in many studies, SNPs are used as genetic markers - they are not necessarily expected to be causal in the phenotype, but SNPs that are associated with the phenotype are expected to be in LD with causal genomic regions.  When using SNPs as genetic markers, we do not need to have all the SNPs - we need a set of SNPs that are well distributed across the genome.

With quantitative phenotypes such as height, weight, or a measure of insulin resistance,  the analysis attempts to find the region between two markers that is highly associated with the phenotype.  Such a region is called a Quantitative Trait Locus.  

When we use a dense set of markers such as SNPs and we use all of the SNPs available, the analysis is called a Genome-Wide Association Study (GWAS). In a GWAS in humans we usually have at least a half a million features that we are looking at, and we do a test of each feature.

Another analysis that has become popular links gene expression and SNP analysis into eQTL(expression QTL) analysis. In eQTL analysis the phenotype is the expression level of one of more genes.  Potentially we could have half a million markers and 30,000 gene expression values for each sample. We are talking here about very big data sets! There is not much difference between eQTL's and other QTL's in terms of the basic statistical testing. The main differences are how the data are handled and how the data are filtered.  Often for each eQTL, only SNPs which are physically close on the chromosome to the expressing gene, or which are close to genes in a network involving the expressing gene are used for the QTL analysis, which greatly reduces the numbers of tests.