Oddly enough, although Mr Self and the Christian bookseller would presumably place themselves at opposite ends of the exegetical spectrum, their responses share an ancestral root.
Sometimes, working deep in the mint thicket, I follow a runner trail to what must have been an ancestral root cluster, a truly daunting mass of subterranean plant fibers.
More recently, whole genome sequencing has resolved that Asian isolates of B. pseudomallei share an Australian ancestral root [7].
This illustrates the importance of using a second-order Markov chain at the ancestral root sequence.
A first-order Markov chain at the ancestral root sequence yields the largest increase in model fit, outperforming a second-order Markov chain at the ancestral root, which also drastically increases model fit compared to the reference model.
To assess whether the probability of observing a given nucleotide at a given site in the ancestral root sequence depends only on its immediate preceding site, we have also modelled a second-order Markov chain at the ancestral root sequence.
The first series of simulations focuses on a second-order ancestral root distribution decoupled from the context-dependent evolutionary model while the second series of simulations focuses on a first-order ancestral root distribution coupled to the context-dependent model.
For example, we have shown that an ancestral root distribution coupled to the context-dependent model does not improve model fit as much as a decoupled ancestral root distribution does.
Moreover, we have also shown that the choice of an ancestral root distribution, changes across different datasets.
The model switches between fast and slow states only when a character type switch is emitted from the ancestral root.
Even when assuming a site independent distribution at the ancestral root sequence, a drastic increase in model fit is realized through context-dependent model frequencies.
