Figure 6

Typical plots of the cumulative distribution functions (CDF) of the differences of sample variances. In these plots, Δ σ _Ω is a calculated difference between sample variances for any combination of n amino acids and p _Ω is the associated probability of obtaining such difference of sample variances for the property Ω just randomly choosing n amino acids, and the CDF profiles vary depending on the number of amino acids selected. Continuing with the example of Figure 2, a calculated difference Δ σ _{9,(Ω1,Ω2),6} of the sample variances between the properties Ω ₁ and Ω ₂ at position i = 9 and n = 6, has an associated probability p _{9,(Ω1,Ω2),6} that can be found using the corresponding CDF_{(Ω1,Ω2), n}. Therefore, the significance level of the null hypothesis H ₀ is p'_{i,(Ω1,Ω2),n}= 1 - p_{i,(Ω1,Ω2),n}.