We can confirm the validity of the run detection algorithm by examining loss of heterozygosity. In fact, runs of homozygosity (calculated on genotypes) can serve as a proxy for loss of heterozygosity (calculated on B allele frequencies).
The B Allele Frequency spreadsheet in the project represents the B allele frequencies calculated for each marker by the Affymetrix B Allele Frequency Calculation script. This script first combines quantile normalized SNP A and B probe intensities for each marker into a theta value. It then calculates B allele frequencies for each marker. If we locate a run of homozygosity in the heat map containing all ROHs, we can confirm that the algorithm worked correctly by plotting the B allele frequency for the same sample that contained the run.
Now we can add a plot of the B allele frequencies for this sample.
If you zoomed in to select the sample, zoom back out to chromosome one by double clicking on it in the Full Domain View. Your plot should look like Figure 18.
B Allele frequencies (y-axis) close to 1 or 0 correspond to homozygous SNPs whereas a frequency closer to .5 suggests heterozygous SNPs. Notice over the same section as the long run from the heat map (essentially the whole q arm of Chromosome 1), the B allele frequency plot contains relatively few markers in the middle. Alternatively, where there are no runs of homozygosity for this given sample (p arm of Chromosome 1), the data are reasonably distributed between 0, .5 and 1.