PARS - Discrete character parsimony
© Copyright 1986-2000 by the University of Washington. Written by Joseph Felsenstein. Permission is granted to copy this document provided that no fee is charged for it and that this copyright notice is not removed.
PARS is a general parsimony program which carries out the Wagner parsimony method with multiple states. Wagner parsimony allows changes among all states. The criterion is to find the tree which requires the minimum number of changes. The Wagner method was originated by Eck and Dayhoff (1966) and by Kluge and Farris (1969). Here are its assumptions:
That these are the assumptions of parsimony methods has been documented in a series of papers of mine: (1973a, 1978b, 1979, 1981b, 1983b, 1988b). For an opposing view arguing that the parsimony methods make no substantive assumptions such as these, see the papers by Farris (1983) and Sober (1983a, 1983b), but also read the exchange between Felsenstein and Sober (1986).
The input for PARS is the standard input for discrete characters programs, described above in the documentation file for the discrete-characters programs, except that multiple states (up to 9 of them) are allowed. Any characters other than "?" are allowed as states, up to a maximum of 9 states. In fact, one can use different symbols in different columns of the data matrix, although it is rather unlikely that you would want to do that. The symbols you can use are:
PARS can handle both bifurcating and multifurcating trees. In doing its search for most parsimonious trees, it adds species not only by creating new forks in the middle of existing branches, but it also tries putting them at the end of new branches which are added to existing forks. Thus it searches among both bifurcating and multifurcating trees. If a branch in a tree does not have any characters which might change in that branch in the most parsimonious tree, it does not save that tree. Thus in any tree that results, a branch exists only if some character has a most parsimonious reconstruction that would involve change in that branch.
It also saves a number of trees tied for best (you can alter the number it saves using the V option in the menu). When rearranging trees, it tries rearrangements of all of the saved trees. This makes the algorithm slower than earlier programs such as MIX.
The options are selected using a menu:
The Weights (W) option takes the weights from a file whose default name is "weights". The weights follow the format described in the main documentation file, with integer weights from 0 to 35 allowed by using the characters 0, 1, 2, ..., 9 and A, B, ... Z.
The User tree (option U) is read from a file whose default name is intree. The trees can be multifurcating. They must be preceded in the file by a line giving the number of trees in the file.
The options J, O, T, and M are the usual Jumble, Outgroup, Threshold parsimony, and Multiple Data Sets options, described either in the main documentation file or in the Discrete Characters Programs documentation file.
The M (multiple data sets option) will ask you whether you want to use multiple sets of weights (from the weights file) or multiple data sets. The ability to use a single data set with multiple weights means that much less disk space will be used for this input data. The bootstrapping and jackknifing tool Seqboot has the ability to create a weights file with multiple weights.
The O (outgroup) option will have no effect if the U (user-defined tree) option is in effect. The T (threshold) option allows a continuum of methods between parsimony and compatibility. Thresholds less than or equal to 1.0 do not have any meaning and should not be used: they will result in a tree dependent only on the input order of species and not at all on the data!
Output is standard: if option 1 is toggled on, the data is printed out, with the convention that "." means "the same as in the first species". Then comes a list of equally parsimonious trees. Each tree has branch lengths. These are computed using an algorithm published by Hochbaum and Pathria (1997) which I first heard of from Wayne Maddison who invented it independently of them. This algorithm averages the number of reconstructed changes of state over all sites a over all possible most parsimonious placements of the changes of state among branches. Note that it does not correct in any way for multiple changes that overlay each other.
If option 2 is toggled on a table of the number of changes of state required in each character is also printed. If option 5 is toggled on, a table is printed out after each tree, showing for each branch whether there are known to be changes in the branch, and what the states are inferred to have been at the top end of the branch. This is a reconstruction of the ancestral sequences in the tree. If you choose option 5, a menu item D appears which gives you the opportunity to turn off dot-differencing so that complete ancestral sequences are shown. If the inferred state is a "?", there will be multiple equally-parsimonious assignments of states; the user must work these out for themselves by hand. If option 6 is left in its default state the trees found will be written to a tree file, so that they are available to be used in other programs.
If the U (User Tree) option is used and more than one tree is supplied, the program also performs a statistical test of each of these trees against the best tree. This test, which is a version of the test proposed by Alan Templeton (1983) and evaluated in a test case by me (1985a). It is closely parallel to a test using log likelihood differences due to Kishino and Hasegawa (1989), and uses the mean and variance of step differences between trees, taken across sites. If the mean is more than 1.96 standard deviations different then the trees are declared significantly different. The program prints out a table of the steps for each tree, the differences of each from the best one, the variance of that quantity as determined by the step differences at individual sites, and a conclusion as to whether that tree is or is not significantly worse than the best one. It is important to understand that the test assumes that all the discrete characters are evolving independently, which is unlikely to be true for many suites of morphological characters.
Option 6 in the menu controls whether the tree estimated by the program is written onto a tree file. The default name of this output tree file is "outtree". If the U option is in effect, all the user-defined trees are written to the output tree file.
TEST DATA SET
TEST SET OUTPUT (with all numerical options on)