CNVEM is a Bayesian Expectation-Maximization algorithm that infers carrier status of CNVs in large samples from SNP genotyping data, such as are available in genome-wide association studies.  Using Bayesian computations the program calculates the posterior probability for carrier status of known CNV in each individual of a sample by jointly analyzing genotype information and hybridization intensity.  Signal intensity is modeled as a mixture of normal distributions, allowing for locus-specific and allele-specific distributions.  Using an expectation maximization algorithm, these distributions are estimated and then used to infer the carrier status of each individual the boundaries of the CNV.  The program is presently implemented only for deletions; a version analyzing duplications is in development.

The algorithm is implemented in a program controlled by a command line interface.

The current version is 0.710.

If you use CNVEM please e-mail or fill out the registration form. The C source code and a short manual can be downloaded here.  To unpack the file use tar -xvf CNVEM.tar.gz, then follow the instructions in the manual.