0/0 1/1 1/2 ?

http://www.ncbi.nlm.nih.gov/pubmed/16100312

J Epidemiol Community Health. 2005 Sep;59(9):749-54.

Measuring paternal discrepancy and its public health consequences. Bellis MA,

Hughes K, Hughes S, Ashton JR. Source

:-)

The problem is many of these mendelian errors are in reality sequencing/variant calling errors, so the number depends on the technology and analysis applied. Considering that de novo mutation rate in humans is really low (1.20 × 10−8 per nucleotide per generation, reference) you can expect a similar number.