5 Partner SNP Metrics

The partner SNP is chosen using the correlation between the anchor SNP and each potential partner SNP. The correlations are mapped to a score between 0 and 100, with 100 being the best. The SNP with the highest score is chosen as the partner SNP. The curves below (figure 2) show the correlation-score maps assuming disease models which are multiplicative, dominant and recessive.

Figure 2. Plots of the partner SNP correlation metrics. Left to right, assuming a multiplicative, dominant and recessive causal variant model respectively.

It is recommended that the default multiplicative causal variant model is used if the penetrances of any causal variants are unknown. The top 10 percent correlations are then in the interval [0.26, 0.36].

If it is known that the causal variant is dominant (recessive) then it should be beneficial to use the dominant (recessive) correlation metric using the -dominant (-recessive) option. (Note that the AI test still uses a multiplicative model and that this option only affects how the partner SNP is chosen.)

Again, for more details concerning the methodology, please read the accompanying manuscript Howey and Cordell (2014).