DOLMOVE -- Interactive Dollo and Polymorphism Parsimony
(c) Copyright 1986-1993 by Joseph Felsenstein and 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.
DOLMOVE is an interactive parsimony program which uses the Dollo and
Polymorphism parsimony criteria. It is inspired on Wayne Maddison and David
Maddison's marvellous program MacClade, which is written for Apple MacIntosh
computers. DOLMOVE reads in a data set which is prepared in almost the same
format as one for the Dollo and polymorhism parsimony program DOLLOP. It
allows the user to choose an initial tree, and displays this tree on the
screen. The user can look at different characters and the way their states are
distributed on that tree, given the most parsimonious reconstruction of state
changes for that particular tree. The user then can specify how the tree is to
be rearraranged, rerooted or written out to a file. By looking at different
rearrangements of the tree the user can manually search for the most
parsimonious tree, and can get a feel for how different characters are affected
by changes in the tree topology.
This program is compatible with fewer computer systems than the other
programs in PHYLIP. It can be adapted to PCDOS systems or to any system whose
screen or terminals emulate DEC VT52 or VT100 terminals (such as, for example,
Zenith Z19, Z29, and Z49 terminals Telnet programs for logging in to remote
computers over a TCP/IP network, VT100-compatible windows in the X windowing
system, and any terminal compatible with ANSI standard terminals). For any
other screen types, there is a generic option which does not make use of screen
graphics characters to display the character states. This will be less
effective, as the states will be less easy to see when displayed.
The input data file is set up almost identically to the data files for
DOLLOP. The sole exception is that the user trees are not contained in the
input file, but in the tree file, which is used both for input of the starting
tree (if desired) and for output of the final tree. Note that this means that
the user tree supplied on input will possibly be overwritten. The W (Weights),
A (Ancestors), and F (Factors) are possible options specified in the input file
(some must also be chosen in the menu).
The user interaction starts with the program presenting a menu. The menu
looks like this:
Interactive Dollo or polymorphism parsimony, version 3.5c
Settings for this run:
P Parsimony method? Dollo
T Use Threshold parsimony? No, use ordinary parsimony
A Use ancestral states in input file? No
U Initial tree (arbitrary, user, specify)? Arbitrary
0 Graphics type (IBM PC, VT52, ANSI)? ANSI
L Number of lines on screen? 24
Are these settings correct? (type Y or the letter for one to change)
The P (Parsimony Method) option is the one that toggles between polymorphism
parsimony and Dollo parsimony. The program defaults to Dollo parsimony.
The T (Threshold) and 0 (Graphics type) options are the usual ones and are
described in the main documentation file. The L option allows the program to
take advantage of larger screens if available. The X (Mixed Methods option is
not available in DOLMOVE. The U (initial tree) option allows the user to
choose whether the initial tree is to be arbitrary, interactively specified by
the user, or read from a tree file. Typing U causes the program to change
among the three possibilities in turn. I would recommend that for a first run,
you allow the tree to be set up arbitrarily (the default), as the "specify"
choice is difficult to use and the "user tree" choice requires that you have
available a tree file with the tree topology of the initial tree. If you wish
to set up some particular tree you can also do that by the rearrangement
commands specified below. The T (threshold) option allows a continuum of
methods between parsimony and compatibility. Thresholds less than or equal to
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!
After the initial menu is displayed and the choices are made, the program
then sets up an initial tree and displays it. Below it will be a one-line menu
of possible commands, which looks like this:
NEXT? (Options: R # + - S . T U W O F C H ? X Q) (H or ? for Help)
If you type H or ? you will get a single screen showing a description of each
of these commands in a few words. Here are slightly more detailed
R ("Rearrange"). This command asks for the number of a node which is to be
removed from the tree. It and everything to the right of it on the tree is
to be removed (by breaking the branch immediately below it). The command
also asks for the number of a node below which that group is to be
inserted. If an impossible number is given, the program refuses to carry
out the rearrangement and asks for a new command. The rearranged tree is
displayed: it will often have a different number of steps than the
original. If you wish to undo a rearrangement, use the Undo command, for
which see below.
# This command, and the +, - and S commands described below, determine which
character has its states displayed on the branches of the trees. The
initial tree displayed by the program does not show states of sites. When
# is typed, the program does not ask the user which character is to be
shown but automatically shows the states of the next binary character that
is not compatible with the tree (the next character that does not perfectly
fit the current tree). The search for this character "wraps around" so
that if it reaches the last character without finding one that is not
compatible with the tree, the search continues at the first character; if
no incompatible character is found the current character is shown, and if
no current character is shown then the first character is shown. If the
last character has been reached, using + again causes the first character
to be shown. The display takes the form of different symbols or textures
on the branches of the tree. The state of each branch is actually the
state of the node above it. A key of the symbols or shadings used for
states 0, 1 and ? are shown next to the tree. State ? means that either
state 0 or state 1 could exist at that point on the tree, and that the user
may want to consider the different possibilities, which are usually
apparent by inspection.
+ This command is the same as # except that it goes forward one character,
showing the states of the next character. If no character has been shown,
using + will cause the first character to be shown. Once the last
character has been reached, using + again will show the first character.
- This command is the same as + except that it goes backwards, showing the
states of the previous character. If no character has been shown, using -
will cause the last character to be shown. Once character number 1 has
been reached, using - again will show the last character.
S ("Show"). This command is the same as + and - except that it causes the
program to ask you for the number of a character. That character is the
one whose states will be displayed. If you give the character number as 0,
the program will go back to not showing the states of the characters.
. This command simply causes the current tree to be redisplayed. It is of
use when the tree has partly disappeared off of the top of the screen owing
to too many responses to commands being printed out at the bottom of the
T ("Try rearrangements"). This command asks for the name of a node. The
part of the tree at and above that node is removed from the tree. The
program tries to re-insert it in each possible location on the tree (this
may take some time, and the program reminds you to wait). Then it prints
out a summary. For each possible location the program prints out the
number of the node to the right of the place of insertion and the number of
steps required in each case. These are divided into those that are better,
tied, or worse than the current tree. Once this summary is printed out,
the group that was removed is inserted into its original position. It is
up to you to use the R command to actually carry out any the arrangements
that have been tried.
U ("Undo"). This command reverses the effect of the most recent
rearrangement, outgroup re-rooting, or flipping of branches. It returns to
the previous tree topology. It will be of great use when rearranging the
tree and when a rearrangement proves worse than the preceding one -- it
permits you to abandon the new one and return to the previous one without
remembering its topology in detail.
W ("Write"). This command writes out the current tree onto a tree output
file. If the file already has been written to by this run of DOLMOVE, it
will ask you whether you want to replace the contents of the file, add the
tree to the end of the file, or not write out the tree to the file. The
tree is written in the standard format used by PHYLIP (a subset of the New
Hampshire standard). It is in the proper format to serve as the User-
Defined Tree for setting up the initial tree in a subsequent run of the
O ("Outgroup"). This asks for the number of a node which is to be the
outgroup. The tree will be redisplayed with that node as the left
descendant of the bottom fork. The number of steps required on the tree
may change on re-rooting. Note that it is possible to use this to make a
multi-species group the outgroup (i.e., you can give the number of an
interior node of the tree as the outgroup, and the program will re-root the
tree properly with that on the left of the bottom fork).
F ("Flip"). This asks for a node number and then flips the two branches at
that, so that the left-right order of branches at that node is changed.
This does not actually change the tree topology (or the number of steps on
that tree) but it does change the appearance of the tree.
C ("Clade"). When the data consist of more than 12 species (or more than
half the number of lines on the screen if this is not 24), it may be
difficult to display the tree on one screen. In that case the tree will be
squeezed down to one line per species. This is too small to see all the
interior states of the tree. The C command instructs the program to print
out only that part of the tree (the "clade") from a certain node on up.
The program will prompt you for the number of this node. Remember that
thereafter you are not looking at the whole tree. To go back to looking at
the whole tree give the C command again and enter "0" for the node number
when asked. Most users will not want to use this option unless forced to.
H ("Help"). Prints a one-screen summary of what the commands do, a few
words for each command.
? ("?"). A synonym for H. Same as Help command.
X ("Exit"). Exit from program. If the current tree has not yet been saved
into a file, the program will ask you whether it should be saved.
Q ("Quit"). A synonym for X. Same as the eXit command.
If the A option is used, then the program will infer, for any character
whose ancestral state is unknown ("?") whether the ancestral state 0 or 1 will
give the fewest changes (according to the criterion in use). If these are
tied, then it may not be possible for the program to infer the state in the
internal nodes, and many of these will be shown as "?". If the A option is not
used, then the program will assume 0 as the ancestral state.
When reconstructing the placement of forward changes and reversions under
the Dollo method, keep in mind that each polymorphic state in the input data
will require one "last minute" reversion. This is included in the counts.
Thus if we have both states 0 and 1 at a tip of the tree the program will
assume that the lineage had state 1 up to the last minute, and then state 0
arose in that population by reversion, without loss of state 1.
When DOLMOVE calculates the number of characters compatible with the tree,
it will take the F option into account and count the multistate characters as
units, counting a character as compatible with the tree only when all of the
binary characters corresponding to it are compatible with the tree.
In the input file the W (Weights) option is available, as usual. It and
the A (Ancestral states) and F (Factors) options require the option to be
declared on the first line of the input file and other information to be
present in the input file. If the Ancestral States option is used the A option
must also be chosen in the menu. The F (Factors) option is available in this
program. It is used to inform the program which groups of characters are to be
counted together in computing the number of characters compatible with the
tree. Thus if three binary characters are all factors of the same multistate
character, the multistate character will be counted as compatible with the tree
only if all three factors are compatible with it.
ADAPTING THE PROGRAM TO YOUR COMPUTER AND TO YOUR TERMINAL
As we have seen, the initial menu of the program allows you to choose
among four screen types (PCDOS, Ansi, VT52 and none). If you want to avoid
having to make this choice every time, you can change some of the constants at
the beginning of the program to have it initialize itself in the proper way,
and recompile it. Among the constants at the beginning of the program you will
find three that determine which kind of screen graphics the program will use.
These constants are "ibmpc0", "vt520", and "ansi0". In the distribution
version of the programs, "ansi0" is set to true and the others to false, so
that the version will work with ANSI compatible terminals.
On the other hand if you have a terminal compatible with DEC's VT52, but
not with the ANSI terminal, you should change the constant "ansi0" to false and
"vt520" to true. If you have instead a terminal which is compatible with IBM
PC graphics, you should set the constant "ibmpc0" to true and the others to
false. If your terminal is compatible with none of these, you will have to
set the constants all false, in which case special graphics characters will not
be used to indicate character states, but only "*" for 1, "=" for 0, and "."
for ? states. This is less easy to look at.
The program should work successfully on DEC VAX systems under either the
VMS or the Unix operating systems without any other changes.
MORE ABOUT THE PARSIMONY CRITERION
DOLMOVE uses as its numerical criterion the Dollo and polymorphism
parsimony methods. The program defaults to carrying out Dollo parsimony.
The Dollo parsimony method was first suggested in print in verbal form by
Le Quesne (1974) and was first well-specified by Farris (1977). The method is
named after Louis Dollo since he was one of the first to assert that in
evolution it is harder to gain a complex feature than to lose it. The
algorithm explains the presence of the state 1 by allowing up to one forward
change 0-->1 and as many reversions 1-->0 as are necessary to explain the
pattern of states seen. The program attempts to minimize the number of 1-->0
The assumptions of this method are in effect:
1. We know which state is the ancestral one (state 0).
2. The characters are evolving independently.
3. Different lineages evolve independently.
4. The probability of a forward change (0-->1) is small over the
evolutionary times involved.
5. The probability of a reversion (1-->0) is also small, but still far
larger than the probability of a forward change, so that many reversions are
easier to envisage than even one extra forward change.
6. Retention of polymorphism for both states (0 and 1) is highly
7. The lengths of the segments of the true tree are not so unequal that
two changes in a long segment are as probable as one in a short segment.
One problem can arise when using additive binary recoding to represent a
multistate character as a series of two-state characters. Unlike the Camin-
Sokal, Wagner, and Polymorphism methods, the Dollo method can reconstruct
ancestral states which do not exist. An example is given in my 1979 paper. It
will be necessary to check the output to make sure that this has not occurred.
The polymorphism parsimony method was first used by me, and the results
published (without a clear specification of the method) by Inger (1967). The
method was published by Farris (1978a) and by me (1979). The method assumes
that we can explain the pattern of states by no more than one origination (0--
>1) of state 1, followed by retention of polymorphism along as many segments of
the tree as are necessary, followed by loss of state 0 or of state 1 where
necessary. The program tries to minimize the total number of polymorphic
characters, where each polymorphism is counted once for each segment of the
tree in which it is retained.
The assumptions of the polymorphism parsimony method are in effect:
1. The ancestral state (state 0) is known in each character.
2. The characters are evolving independently of each other.
3. Different lineages are evolving independently.
4. Forward change (0-->1) is highly improbable over the length of time
involved in the evolution of the group.
5. Retention of polymorphism is also improbable, but far more probable
that forward change, so that we can more easily envisage much polymorhism than
even one additional forward change.
6. Once state 1 is reached, reoccurrence of state 0 is very improbable,
much less probable than multiple retentions of polymorphism.
7. The lengths of segments in the true tree are not so unequal that we can
more easily envisage retention events occurring in both of two long segments
than one retention in a short segment.
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).
At the beginning of the program are a series of constants, which can be
changed to help adapt the program to different computer systems. "nmlngth" is
the length of the species names. "screenlines" specifies the number of lines
per screen, which you will normally want to leave at its default value of 24.
I have already described the constants "ibmpc0", "vt520", and "ansi0" for
specifying the terminal type.
Below is a test data set, but we cannot show the output it generates
because of the interactive nature of the program.
-------------------------------TEST DATA SET----------------------------