Has anyone generalized the concept of LD to variants that are in CNV events? Or what about between multiallelic variants?
What is the equivalent of r^2, and how would you express it between:
1) A SNV that lies within a duplication event (containing, for example, 3 total copies) and another biallelic SNV not on a CNV?
2) 2 SNVs both on CNVs?
3) A polyallelic SNV and a biallelic SNV?
4) 2 polyallelic SNVs?
5) 1 or more polyallelic SNVs that are on CNVs (e.g. a A/G/T variant lying within a duplication event with another SNV either in or not in the duplicated region)?
Last, some individuals will have a CNV, and some may not. How can these be integrated?
Has this been done? Standard Google searches turn up results that at least to me appear to be totally unrelated.
Found a few sources: http://www.genetics.org/content/117/2/331.abstract?ijkey=2582a823eaff8cdeaf91d96924302ab707c7eb14&keytype2=tf_ipsecsha from what I can tell this is a seminal paper for polyallelic SNV LD.
Another is http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3109359/ - a framework for determining SNP CNV LD that uses a custom method.
This paper describes the two in relation to haplotype http://bioinformatics.oxfordjournals.org/content/26/11/1437.full
Further additions are very much needed, please add any additional thoughts you may have, and thanks.