Below, we demonstrate the plotting capabilities of SNPchip via some case studies. Displaying SNP chip data in general is shown in Figure 1. The zoom feature is shown in Figure 2. Examples of a deletion and an amplification are shown in Figure 3. The data displayed in Figure 4 highlight the dramatic genomic aberrations that can occur in breast cancer cell lines. Other features illustrated include uniparental disomy (Figure 5), mosaicism (Figure 6), and sample mislabeling (Figure 7). The R code for generating comparable figures from an abbreviated 50K Hind and Xba SNP chip can be obtained on the example code page after installing the SNPchip package from Bioconductor (see the installation how to page). Please also visit the help files and run the examples for the respective functions.
[ Figure 1 | Figure 2 | Figure 3 | Figure 4 | Figure 5 | Figure 6 | Figure 7 ]
Click on any Figure to enlarge.
Figure 1: Genomic data data of a normal female (top) and a normal male (bottom). The x-axis denotes the loci along chromosomes 1-22 and the X chromosome; the y-axis denotes the estimates of the respective copy numbers. A data point is colored in blue if it is called homozygous (AA or BB), red (plotted on top of the blue) if it is called homozygous (AB), and green if no call is made. In normal individuals, the average copy number for the autosomes is 2, and homozygous (~2/3) and heterozygous (~1/3) genotypes somewhat alternate. Since the normal male has only one X chromosome, the average copy number for the male X chromosome is one, and virtually all genotype calls are homozygous. The normal female has two X chromosomes, and so the average copy number and genotype for the X chromosome are comparable to the autosomes. The chromosomes are indicated by horizontal dark bars (with the size in Mb indicated), the centromeres are indicated by lighter vertical bars.
Figure 2: Magnification of chromosomes 19-22 and chromosome X of the above samples (click to enlarge).
Figure 3: Zooming in on chromosomes 11 and 12 of a sample (plot 1, left). We see a large region with loss of heterozygosity (LOH) and decreased copy number on the q-arm of chromosome 1, indicating a deletion. Plot 2 (right) shows a large region of amplification on the q arm of chromosome 7.
Figure 4: Genomic data of three breast cancer cell lines highlight the dramatic chromosomal changes that can be encountered when analyzing SNP chip data. Many deletions and duplications are clearly visible even when the data of the entire genome is plotted.
Figure 5: On the q arm of chromosome 1 is a huge stretch of almost exclusively homozygous calls. However, the average copy number is still approximately 2. The phenomenon in this example is called uniparental disomy, and occurs when an individual inherits two copies of a chromosome from one parent, and no copy from the other parent. Here, the subject inherited two copies of the same parental chromosome (isodisomy). This differs from the case when both parental chromosomes are inherited (heterodisomy), which would not give rise to a pattern as shown in this figure.
Figure 6: The genomic data on chromosome 2 are shown. At first glance, it seems that a deletion has occurred on the q arm of chromosome 2, as the copy number is clearly lower on a stretch of that arm. However, there are still a fair number of heterozygous calls made on that stretch (red dots). If a deletion had occurred, all of the genotype calls should be homozygous. The solution to this puzzle is that a somatic mutation has occurred, and a mixture of cells with different chromosomal makeup had been run in the sample. Some of the cells contained a chromosome 2 with a deletion, other cells had intact chromosomes. Therefore the average copy number was down, a higher proportion of SNP calls were homozygous, but other still showed up as heterozygous. Also noticeable is the larger fraction of no-calls, i.e. SNPs for which the fluorescence pattern on the SNP chip was ambiguous. FISH (fluorescence in situ hybridization) confirmed this phenomenon (usually called mosaicism) experimentally. Also plotted here is the cytoband for chromosome 2.
Figure 7: Plot 1 on the left shows 100K SNP chip data of chromosomes 19-22 and chromosome X for two subjects in the trios data set made public by Affymetrix. The apparent mosaicism on chromosome X for both samples in plot 1 is a result of merging Xba and Hind SNP chips, one of which had mislabeled sample names. To see this, we plot the Xba (plot 2) and Hind (plot 3) chips separately.
