I have a method that divides the genome up into regions based on genomic structure. The regions can vary somewhat in size, i.e. are not uniform.
I want to see whether these regions are in linkage disequilibrium, i.e. these regions are generally inherited as a block. I thought that one way to do so would be to compare data to 1000 genomes data, i.e. calculate LD within these regions according to how often the SNVs in these regions are inherited together, using some metric. This could then be compared to a random model by dividing the chromosomes up into regions of the same size as in the method results, but at random points.
So, for example imagine that my method says the genome is divided at: 1-3, 4-9, 10-12 and 13-19. I would calculate LD using this divisions.
I would compare these results to dividing the genome randomly e.g. 1-6,7-9,10-16,17-19 and calculating the LD scores. This would be repeated many times.
Does anyone have any experience doing anything like this? More generally,
1) Is there an easy way to calculate LD for genome regions using 1000 genome data 2) Does this seem like a reasonable way to calculate the significance of LD?
