Bioinformatic algorithms have been developed to use SNP array information to identify genomic aberrations such as DNA copy number changes and loss-of–heterozygosity (LOH), i.e. stretches of DNA with exclusively homozygous markers 
. However, one major drawback of these methods is that genetic heterogeneity in tumor samples, caused by the mixture of cancer and stromal cells, is often not taken into account. As a consequence aberrations are often not detected in samples with a large proportion of genetically normal cells. This may partly explain why, despite the accumulation of large amounts of genomic data, the clinical impact of such analyses for diagnostic purposes is still small. Tumor tissue represents a mixture of tumor and non-tumor cells, i.e. inflammatory cells, stromal fibroblasts and cells of blood- and lymph vessels 
. The fraction of normal cells often exceeds the fraction of tumor cells in patient samples stored in biobanks (). This sample heterogeneity severely affects copy number analysis. To the best of our knowledge there are no estimates on how the sensitivity of detection of genomic aberrations depends on the proportion of normal cells in clinical tumor samples. One reason may be the difficulty to estimate the tumor vs. normal cell ratio histologically by microscopy in heterogeneous tumor samples with varying proportions of normal cells in different parts of the sample. Moreover, there is a lack of consensus on how tumor cell content in a solid cancer should be assessed and annotated. Thus, the performance of the current tools for detection of genomic aberrations in clinical tumor samples is often uncertain.
A recently developed tool takes sample heterogeneity into account for identification of copy number states 
. It is designed for studies with paired samples (tumor and normal). In practice, however, paired samples are often not available for larger patient cohorts.
In another study Nancarrow et al visualize the expected pattern of allele frequencies depending on varying proportions of normal cells in the tumor sample using simulations 
Another promising analytical tool, AsCNAR, is able to identify LOH even when one of two mixed cell lines is present only in a proportion of about 20% 
. Recently Assie et al described an algorithm that take tumor heterogeneity into account in identifying genomic aberrations in samples with 40–75% of tumor cells 
Studies suggest that copy number neutral LOH can be a mechanism for inactivation of tumor suppressor genes 
. Several studies and our own data suggest that CNNLOH is more common than previously thought 
. Taken together this suggests that CNNLOH may be important in determining certain cancer phenotypes. To analyze CNNLOH on a genome-wide scale in the tumor cells in heterogeneous samples we focused on 1) developing an algorithm to quantify the proportion of normal cells in the sample and 2) to quantify CNNLOH throughout the genome in the tumor cells. Such quantitative analysis has the potential to become an important tool for molecular cancer diagnostics.