Home » Computational Evolutionary Biology

Introduction

Consider an evolutionary tree of a particular shape for a set of entities, each with given character states. An 'optimal' tree can be defined as one in which the internal nodes are such that it has the minimum number of changes in the states of all the characters summed every pair of connected nodes in the tree -- the 'parsimony' criterion. This demonstration enables you to experiment with such trees containing small numbers of entities.

Notation

• Binary trees are described in the 'linear bracketted' style, but using '[' and ']' rather than '(' and ')'. Thus [['1','2'],['4',['5','2']]] is a valid binary tree. Only the leaves of the tree can be specified.

• Each character state is described by a SINGLE letter or digit.

  • The letters 'A'-'Z' represent discrete character states.
  • All other letters and digits represent ordered character states, in which distances are defined by the ASCII code of the character (e.g. 'a' is 3 away from 'd').

  Characters must be consistently coded, i.e. either all the states of a character must be discrete or all the states must be ordered.

• Each entity is represented by a string of character states. Strings MUST be enclosed in ' ' (even if the character states are represented by digits). Every entity must have the same number of characters (i.e. all the strings must be the same length). Thus 'ACaX3' and 'CTeB4' are two consistent character state strings.

On pressing "Run", Fitch Parsimony will be applied to the discrete characters and Wagner Parsimony to the ordered characters. ('Fitch' and 'Wagner' are used in the sense of Ronquist & Beerli, http://people.sc.fsu.edu/~beerli/ISC5317/Lectures/02pars.pdf. However, the algorithm for the Wagner up-pass is not as given by their 2007 lecture notes: see instead Programming the Wagner Uppass Algorithm.)

Note: there is no validation of the input, so it needs to be in the right format!