Distribution of P
of sequences of different lengths. For a candidate sequence with the given length in nucleotides n (50 to 160) and a composition of 25% of each nucleotide (AUGC), the MFE of 1000 randomized sequences was calculated. The distribution was computed and plotted (green) using the distribution density function in R. The average mean and standard deviation of the resulting MFE sequence set was used to define the normal distribution function (red). The good correspondence between the two distributions shows that the normal distribution-based probability PN is a good approximation for the empirical probability PE.