Alton Chad Sorensen and John Kauwe, Physiology and Developmental Biology
I was interested in trying to validate the findings of the publication by Medway et al. concerning the SNP rs199768005 using data from the Cache Country data set. Unfortunately, of the 4703 people from our data set, only 10 were carriers. Using R statistical analysis, I found that the p-value for this SNP with an association to protect for Alzheimer’s Disease (AD) was not significant. It was similarly not significant in its association with Any AD. I also ran other models to include for factors such as education or age in conjunction with this SNP and its association with AD or Any AD. The SNP was not significant in any of the models, regardless of other factors included, in its association with Alzheimer’s Disease.
To further analyze our findings, I performed a power analysis using a power calculator created by Dartmouth1. The original study by Medway et al. had an OR value of 0.1. Assuming similar conditions, the power of our study was determined to be .116, a very low power coefficient. The combination of the extreme rarity of this particular mutation and the small OR value are probably factors for this low value.
After doing some additional reading I did online about this particular SNP, I read that it occurs in only .12% of the APOE gene. Thus to reliably predict whether this SNP has protective properties, we would probably need a very large sample size to draw from in order to determine whether or not there is an association between this SNP and reduced risk of AD and if that association is significant. Unless we have access to a much larger sampling pool that could provide us with additional data about the SNP rs199768005, it may not be in our best interest to continue investigating this SNP. Based on this recommendation, we have discontinued research of this SNP for the time being.
- http://www.dartmouth.edu/~eugened/power-samplesize.php