It seems to be taken for granted that "Log R Ratio" and "B Allele Frequencies" data grow on trees. In reality, they are dependent of and are calculated from the detail of the genotype calling algorithm (and measures systematic deviation from the genotype clusters).
So this is what cnvPartition/CauliFlower does - identify regions of "Log R Ratio" and "B Allele Frequencies" where the Log R Ratio is rather different from zero (above for duplication, below for deletion) and B Allele Frequencies different from the biallelic value of 0,0.5,1 (simple deletion cannot take the 0.5 value, simple 3 copies takes 0, 0.33, 0.67,1 ).
However, as explained above, cnvPartition/CauliFlower needs to read and understand the genotype calling outcome of GenTrain2/SpringOnion; and downstream, a way of verification and visualization (the currently in-progress Carrot R functionlity in snpMatrix). So here are some pretty graphs of deletions and duplications on chromosome 16p13.11 recently found to be associated with ADHD and other mental illnesses:
Hin-Tak Leung, last updated 2010-10-24