Using synonymous vs. non-synonymous ALT for SNP data validation

Question

[rightmirem]

I'm looking again at validating SNP calls (whole genome) and stumbled across this article

The effect of strand bias in Illumina short-read sequencing data

...in which the author(s) mention ...

...the number of novel non-synonymous SNPs can also be a very good indicator of the false positive rate. A study[12] have shown that only 200–300 novel nonsynonymous SNPs should be identified per person by exome sequencing; a higher number would likely indicate a higher false-positive rate.

Reference "12" is Here ... however I didn't actually see that statistic (200-300) discussed.

I'm curious if anyone has used this method for validating SNPs? If so, is there a tool/protocol I can follow?

Thanks!