Tips 1. Input READ the ATRIUM_Input DOCUMENT CAREFULLY. The program will stop if errors are detected in the formats of any of the input files. Please read the ATRIUM_Input document carefully and make sure that the input files are in the correct format and have concordant information. 2. Accuracy of p-values when sample size is small or allele frequency is extreme The chi-square approximation to the null distribution of the ATRIUM test statistic gives appropriate type I error when the sample size is sufficiently large. However, if the minor allele frequency of the untyped SNP is extremely small relative to the sample size, then the chi-square approximation may be inaccurate. When the expected number of copies of an allele of the untyped SNP is below 5 in either the case or control sample, the program will issue a warning about possible inaccuracy in the p-value calculated from the chi-square approximation. In that case, empirical p-values obtained via simulation (e.g. parametric bootstrap) could in principle be used, though this is not currently implemented in the software. 3. Parsimony in set of haplotypes considered in the analysis The ATRIUM method is a form of haplotype analysis of the tag SNPs of each untyped SNP. It is common that not every possible haplotype actually occurs in the data. Whether or not a haplotype occurs in the data is often not directly observable when only unphased genotype data are available on the sample. Inclusion, in the ATRIUM analysis, of haplotypes with sample frequencies close to 0 can cause numerical issues (e.g. in the inversion of the covariance matrix). We resolve this problem as follows: we first use the IQLS method (Wang and McPeek 2009 JASA 104: 1251-1260) to estimate the haplotype frequency distribution in the entire case-control sample (including any individuals of unknown phenotype). These estimates are provided in the output. If the estimated frequency for a haplotype of tag SNPs is below some threshold (currently set to <.00005), then this haplotype is dropped and is not considered further in the analysis.