The original article to which this Erratum refers was published in Human Mutation 34(4):546–556 (DOI 10.1002/humu.22273).
In the article cited above, the symbols for Mik and were inadvertently changed during the production process. The corrected text appears below.
The Publisher sincerely regrets this error.
At face value, this minimization is rather expensive because for each index pair u,v, the values of πi[u,v] need to be determined. The naïve evaluation of Eq. (7) can be replaced by a recursive scheme (Supp. Fig. S1). Consider sequence positions k < i < l and denote by Mik and Nil, the probabilities that i has a pairing partner in the interval [k,i – 1] and the interval [i + 1,l], respectively. Clearly, these auxiliary variables satisfy Mik = Mi(k − 1) + Pki and Nil = Ni(l + 1) + Pil. Obviously, for all k < i < l, we have πi[k,l] = Mik + Nil, , and . Precomputing M and N thus allows the evaluation of distances and correlation coefficients in linear time for each sequence interval.