[ Contents | Search | Post | Reply | Next | Previous | Up ]
From: Confidence intervals
Date: 17 Oct 2011
Time: 15:09:10 -0500
Remote Name: 165.112.47.66
The confidence intervals on our mutation frequency estimates are large. However, such statistical arguments are poorly suited for determining the causative nature of genetic variants that are found in cases and controls. For example, the lysine mutation at amino acid position 1200 in GNPTAB was initially observed in a large Pakistani family with many individuals who stutter. The same mutation was observed in the affected members (but not in the unaffected members) of 3 other unrelated Pakistani stuttering families, as well as in a number of unrelated Pakistani stutterers. Population genetic studies of the chromosomes containing this mutation show that they have a single origin that is quite old (about 14,000 years ago, but again with large confidence intervals), making this a so-called founder mutation. This mutation does not occur in neurologically normal controls. It is difficult to explain the recurrent observation of the exact same mutation in the same gene without invoking a causative relationship, regardless of things like p-values.