It occurs to me that we can put errorbars on these by assuming that they each one is binomially distributed, which should give some sense of the coverage per SNP.

Another option could be to draw partially transparent ellipses with the width and height corresponding to the estimates.
It occurs to me that we can put errorbars on these by assuming that they each one is binomially distributed, which should give some sense of the coverage per SNP.
Another option could be to draw partially transparent ellipses with the width and height corresponding to the estimates.