Jackknife analysis is similar to bootstrap analysis in that it is a way of testing the reliability of the dataset. It creates subsets of the original dataset by resampling contiguous sites of which their total number is smaller than that of the number of sites that make up the original dataset. This method allows you to analyse your dataset for a bias, which may be the result of the presence of, for instance, different domain stuctures in a single protein. Imagine that a protein has two similar domains that have been created by an event of gene duplication followed by the incorporation of a largely different number of mutations in the two domains due to adaptation to a new function or condition. One domain may have evolved into a regulatory domain, while the other domain kept its catalytic properties. Another example could be the case of multiple related genes such as the members of a gene family. Homologous recombination affecting only certain parts of such genes may lead in largely different apparent rates of evolution affecting different parts of such genes. Such differences in mutation rate may show up in a jackknife analysis when the subset size is set to 50% or less of the total size.
Thus Jackknife analysis:
NB: If the entire dataset is compatible and has not been biased by stochastic effects, all bootstarp trees should in principle have the same topology!
However, if the original dataset is biased, a cluster may be regarded as statistically significant, even if it is a wrong one !
created by :Fred Opperdoes
Last updated: 8 August 1997.