Figure 1.
A comparison of DNA copy number genotypes showing Ballele (top) and logR (bottom) simulated data for a) tumour cell line showing 100% changes in each case, and b) tumour biopsy with 80% tumour and 20% stromal (normal) DNA.
In each case all common DNA copy number changes are represented (separated by vertical lines) and discussed sequentially within the text.
Figure 2.
Simulated Ballele (top) and logR (bottom) plots showing serial dilution of the a) loss of heterozygousity, and b) amplification (3n) tumour genotypes in the presence of increasing levels of stromal (normal).
In each case stromal levels of 0% (pure tumour) to 80% are represented in steps of 10% (separated by vertical lines), along with the normal 2n genotype.
Figure 3.
Estimating AMP & LOH levels in a) real data using b) SiDCoN.
Comparison to simulated data indicates that ∼60% of cells are AMP or LOH for the indicated regions (thus 40% of cells are normal 2n in each case). These data suggest that this tumour biopsy contains 40% stroma, although more copy number changes across the genome are needed to confirm this.
Figure 4.
Some examples of observed vs simulated chromosomal excerpts showing mixed populations of DNA copy number changes (top) and the manually adjusted simulations of these changes (bottom).
a) a melanoma cell line with changes including HD and mix of LOH & HD. b) an EAC tumour biopsy profile which includes LOH, N-LOH and a mix of HD and LOH. The simulator is particularly useful for explaining LOH/HD combinations in the presence of stroma as seen here. c) another EAC tumour biopsy with changes on a higher background of stroma/normal cells. Manually adjusting the simulator is useful for determining the level of tumour cell involvement in each change.
Figure 5.
Screen grabs from SiDCoN showing a) the main “datasheet” interface with space for three DNA copy number genotypes and a stromal component for each SNP, b) the lookup sheet containing information needed for calculations dependent on the copy number genotypes entered and c) the Ballele and logR output, implementing randomised values to visually simulate the look of actual data.