Overview of ATRIUM

ATRIUM is a program, written in C++, for case-control association testing 
between a binary trait and an untyped SNP, based on genotype data from 
multiple typed SNPs that are in linkage disequilibrium with the untyped 
SNP, where information on the joint distribution of typed and untyped 
SNPs is obtained from an external reference panel, and where the 
individuals in the sample can be related.

The program allows for testing association of the trait with any number 
of untyped SNPs (e.g. it can be used for GWAS), where separate tests are 
performed at each untyped SNP.  The program is applicable to association 
studies with completely general combinations of related and unrelated 
outbred individuals, where the relationships among the sampled 
individuals are assumed to be known.

The main reference for this program is Wang Z. and McPeek M. S. (2009) 
"ATRIUM: Testing Untyped SNPs in Case-Control Association Studies with 
Related Individuals". American Journal of Human Genetics 85:667-678.

The ATRIUM program can be considered an extension, to testing untyped
SNPs, of the IQLS program for haplotype testing.  The original IQLS 
method is described in Wang Z. and McPeek M. S. (2009) "An 
Incomplete-Data Quasi-likelihood Approach to Haplotype-Based Genetic 
Association Studies on Related Individuals", Journal of the American 
Statistical Association 104: 2151-1260.

For each untyped SNP, the ATRIUM program computes a test statistic for 
association.  The ATRIUM test is a 1-df, quasi-likelihood score test that 
allows for missing phase information and/or missing marker genotype data 
in the tag SNPs of an untyped SNP.  ATRIUM uses information from an 
external reference panel, such as HapMap, to select an optimal direction 
for testing association with an untyped SNP, based on genotype data from 
typed SNPs.  For more detail, see Wang and McPeek (2009) Am J Hum Genet.  
For each test, a p-value is calculated based on the chi-square asymptotic 
null distribution.  

To calculate the ATRIUM statistic, an estimate of the population prevalence 
of the trait must be specified by the user.  As described in Thornton and 
McPeek (2007), we emphasize that the test will be valid regardless of the 
input value.  We recommend using an estimate from previous studies or 
registry data from the population.

Additional features of the ATRIUM tests include:

(1) The ATRIUM analysis is conditional on the external LD information.  
    This results in the property that the ATRIUM tests are valid regardless 
    of the choice of the reference panel, i.e., ATRIUM maintains the nominal 
    type I error rate even when the external reference panel is not 
    well-matched to the case-control sample.

(2) The ATRIUM test improves power by using the external LD information to 
    select the direction in which to perform the 1-df test of association 
    with an untyped SNP.  It also improves power by taking advantage of the 
    principle that there is enrichment for predisposing variants in affected 
    individuals with affected relatives.

(3) The ATRIUM test takes into account available parental genotype 
    information when considering an individual's haplotype.