>

(d) Variance of the Correction Factor

E(H_nb) is the mean of the noisy estimate H_nb. The variance (calculated exactly) can be shown to be:

$$Var(H_{nb}) = \left( \sum_{\mbox{\scriptsize {all} nb}} P_{nb}(H_{nb})^2 \right) - E(H_{nb})^2 .$$ (19)

This can be used to estimate the standard deviation of R_sequence owing to sampling error. If a site is r bases wide then the standard deviation is $\sqrt { r Var(H_{nb}) }$.