When working on SNP detection by whole genome (or targeted) re-sequencing, the false positive rate is comparatively easy to check, namely by validating a reasonable number of predicted variants and estimating the false positive rate from there. For false negatives, in the absence of a complete knowledge of the "true" sequence, there does not seem to be an equally obvious way to estimate the miss (type-II) error rate.

So my question (or actually, two of them) is: What is the best method to estimate the false negative rate in a re-sequencing experiment, and, what is the most rational way to set an acceptable limit ?