We have developed and implemented one such initial approach to test for such associations. For sites showing both allelic and copy number variation, we regress the phenotype (either a quantitative trait or disease status) on both the sum and the difference of the number of copies of each allele. A significant regression coefficient for the sum represents an association with overall copy number, whereas a significant coefficient for the difference represents an association with variation tagged by a SNP. If there is either no CNV or no SNP at a site, we fit a reduced model by removing the appropriate term (the sum or the difference, respectively), effectively giving a standard one-dimensional regression of phenotype.
