Using results of error estimation

[jcaccavo]

We would like to compare two different sets of genotype likelihoods (GLs) - one from a set of poor quality, degraded samples, and one from a set of better quality samples. We thought that we could use the error estimation feature in ANGSD as a metric to compare the two GL sets. We were thinking of using the Kim 2011 method (from polymoprhic sites), but would also consider using an error free individual (we have some samples in the better quality sample set with >15x coverage that could be used for this purpose).

Does anyone have any experience presenting this kind of analysis or using the error estimation feature for this purpose? I dug around in the literature and couldn't find any papers in which the error estimation was presented (and looking up papers citing Kim 2011 mostly talk about using GLs generally, not error estimation specifically).

Thanks!