These CNVs may be so small that they are entirely contained within a gene. Larger CNVs may cover whole genes and can be up to several megabases long. CNVs have been associated with many genomic disorders such as Charcot–Marie–Tooth disease, as well as diseases such as hypertension and schizophrenia and may also be associated with susceptibilities to certain cancers. It has been presumed that the expression of CNV genes would correlate with copy number,Ponatinib so that for example, a gene that has one allele duplicated would have 1.5 times the level of expression of the wild type copy number of 2. In other words an expression ratio of 1.5. Stranger et al. showed that SNPs are responsible for 83.6% of the detectable variation in expression among individuals compared is due to CNVs despite the fact that CNVs cover a greater proportion of an individuals genome than is covered by SNPs. Some of this variation in expression is due to CNVs that lie upstream of a gene disrupting the regulatory regions. Stranger agreed with recent work in this lab by Schuster-Bockler et al. that there is a significant difference in the expression ratios between duplicated and deleted genes, and that this difference is much smaller than expected. A similar result has also been found in mouse. However,AG-013736 there were a number of caveats that may have led to these results, including the fact that the precise CNV breakpoints were unknown so that it was possible that genes were being included that were not truly within the CNV, and vice versa, which would bias the mean expression levels. Not knowing the precise breakpoints might also lead to the inclusion of genes which partially overlap the CNV. In this case only a part of the gene is in multiple copy, so that a supposed copy number of 3 may be only one functional allele and two disrupted alleles. This would also bias average expression ratios for amplified genes. However, recent work by Conrad et al. has produced a new dataset of CNVs with breakpoints that are claimed to be accurate to within 60 nucleotides. This improvement in the accuracy of break point prediction allows us to say with a much greater degree of certainty that a gene is fully between the breakpoints of a CNV or not.