Wednesday, February 4, 2009
MOTIVATION: Fractal methods have been identified as a possible approach to the reconstruction of the 'Tree of Life' (eg human progress). However, a limitation of such methods is that, typically, they use just leaf-labelled phylogenetic trees to simply infer the resulting supertree without further proof.
RESULT: Unclear definition of progress or measurement of same (eg inherent doubt/flux). There are several new supertree algorithms that extend the allowable information that can be used for phylogenetic inference.
AVAILABILITY: These new algorithms are freely available.
For example, input could include information such as whether one particular event occurred before or after an
alteration juncture. This allows the inclusion of general and specific nested information in the input. In this way each life can be viewed as a variation of BUILD (one of the oldest algorithms). In addition, new algorithms for ancestral divergence dates and nested information, respectively, can be applied to all data sets.
THE BUILD APPROACH
Originally designed for other purposes, BUILD PROGRAMS (see eg; germain, swerdlow, giovannitti, others et al., 1971) is an imprecise algorithm in that it outputs lives precisely because (...simply put...) the input collection satisfies a particular compatibility criteria. In other words: A rooted tree ("T") displays a rooted subset of T with at least two (often more) vertices. See for example any number of fractals: evolution, revolution, young/old, friends, family, even exchange student, ... bottom line: a karuma sign of something very good.
It is important to note that these "tree of life" algorithms are not "all-or-nothing" devices. Each algorithm either returns a supertree with certain desirable properties relative to the norm or returns a statement indicating alternative selfdom -- these are things that are outside of our control and further understanding can only be obtained by exponential reduplication to a higher power (eg comparison of different sized infinities). In practice, this limits the utility of these algorithms. However, such important first steps are needed, indeed, that is the whole point.
Lastly, two natural questions arise: (i) how many such supertrees are there and (ii) what common information is carried by all of these supertrees? In order to address those two questions Sensitivity Analysis can be used to identify uncertainties. Typically, such analysis has several basic input variables. In other words, to realize supertree potential, one must get to the root of the matter.