fpenny |
Please help by correcting and extending the Wiki pages.
There is, however, a price to be paid for the certainty that one has found all members of the set of most parsimonious trees. The problem of finding these has been shown (Graham and Foulds, 1982; Day, 1983) to be NP-complete, which is equivalent to saying that there is no fast algorithm that is guaranteed to solve the problem in all cases (for a discussion of NP-completeness, see the Scientific American article by Lewis and Papadimitriou, 1978). The result is that this program, despite its algorithmic sophistication, is VERY SLOW.
The program should be slower than the other tree-building programs in the package, but useable up to about ten species. Above this it will bog down rapidly, but exactly when depends on the data and on how much computer time you have (it may be more effective in the hands of someone who can let a microcomputer grind all night than for someone who has the "benefit" of paying for time on the campus mainframe computer). IT IS VERY IMPORTANT FOR YOU TO GET A FEEL FOR HOW LONG THE PROGRAM WILL TAKE ON YOUR DATA. This can be done by running it on subsets of the species, increasing the number of species in the run until you either are able to treat the full data set or know that the program will take unacceptably long on it. (Making a plot of the logarithm of run time against species number may help to project run times).
The search strategy used by PENNY starts by making a tree consisting of the first two species (the first three if the tree is to be unrooted). Then it tries to add the next species in all possible places (there are three of these). For each of the resulting trees it evaluates the number of steps. It adds the next species to each of these, again in all possible spaces. If this process would continue it would simply generate all possible trees, of which there are a very large number even when the number of species is moderate (34,459,425 with 10 species). Actually it does not do this, because the trees are generated in a particular order and some of them are never generated.
Actually the order in which trees are generated is not quite as implied above, but is a "depth-first search". This means that first one adds the third species in the first possible place, then the fourth species in its first possible place, then the fifth and so on until the first possible tree has been produced. Its number of steps is evaluated. Then one "backtracks" by trying the alternative placements of the last species. When these are exhausted one tries the next placement of the next-to-last species. The order of placement in a depth-first search is like this for a four-species case (parentheses enclose monophyletic groups):
Make tree of first two species (A,B) Add C in first place ((A,B),C) Add D in first place (((A,D),B),C) Add D in second place ((A,(B,D)),C) Add D in third place (((A,B),D),C) Add D in fourth place ((A,B),(C,D)) Add D in fifth place (((A,B),C),D) Add C in second place: ((A,C),B) Add D in first place (((A,D),C),B) Add D in second place ((A,(C,D)),B) Add D in third place (((A,C),D),B) Add D in fourth place ((A,C),(B,D)) Add D in fifth place (((A,C),B),D) Add C in third place (A,(B,C)) Add D in first place ((A,D),(B,C)) Add D in second place (A,((B,D),C)) Add D in third place (A,(B,(C,D))) Add D in fourth place (A,((B,C),D)) Add D in fifth place ((A,(B,C)),D)
Among these fifteen trees you will find all of the four-species rooted bifurcating trees, each exactly once (the parentheses each enclose a monophyletic group). As displayed above, the backtracking depth-first search algorithm is just another way of producing all possible trees one at a time. The branch and bound algorithm consists of this with one change. As each tree is constructed, including the partial trees such as (A,(B,C)), its number of steps is evaluated. In addition a prediction is made as to how many steps will be added, at a minimum, as further species are added.
This is done by counting how many binary characters which are invariant in the data up the species most recently added will ultimately show variation when further species are added. Thus if 20 characters vary among species A, B, and C and their root, and if tree ((A,C),B) requires 24 steps, then if there are 8 more characters which will be seen to vary when species D is added, we can immediately say that no matter how we add D, the resulting tree can have no less than 24 + 8 = 32 steps. The point of all this is that if a previously-found tree such as ((A,B),(C,D)) required only 30 steps, then we know that there is no point in even trying to add D to ((A,C),B). We have computed the bound that enables us to cut off a whole line of inquiry (in this case five trees) and avoid going down that particular branch any farther.
The branch-and-bound algorithm thus allows us to find all most parsimonious trees without generating all possible trees. How much of a saving this is depends strongly on the data. For very clean (nearly "Hennigian") data, it saves much time, but on very messy data it will still take a very long time.
The algorithm in the program differs from the one outlined here in some essential details: it investigates possibilities in the order of their apparent promise. This applies to the order of addition of species, and to the places where they are added to the tree. After the first two-species tree is constructed, the program tries adding each of the remaining species in turn, each in the best possible place it can find. Whichever of those species adds (at a minimum) the most additional steps is taken to be the one to be added next to the tree. When it is added, it is added in turn to places which cause the fewest additional steps to be added. This sounds a bit complex, but it is done with the intention of eliminating regions of the search of all possible trees as soon as possible, and lowering the bound on tree length as quickly as possible.
The program keeps a list of all the most parsimonious trees found so far. Whenever it finds one that has fewer steps than these, it clears out the list and restarts the list with that tree. In the process the bound tightens and fewer possibilities need be investigated. At the end the list contains all the shortest trees. These are then printed out. It should be mentioned that the program CLIQUE for finding all largest cliques also works by branch-and-bound. Both problems are NP-complete but for some reason CLIQUE runs far faster. Although their worst-case behavior is bad for both programs, those worst cases occur far more frequently in parsimony problems than in compatibility problems.
Among the quantities available to be set at the beginning of a run of PENNY, two (howoften and howmany) are of particular importance. As PENNY goes along it will keep count of how many trees it has examined. Suppose that howoften is 100 and howmany is 1000, the default settings. Every time 100 trees have been examined, PENNY will print out a line saying how many multiples of 100 trees have now been examined, how many steps the most parsimonious tree found so far has, how many trees of with that number of steps have been found, and a very rough estimate of what fraction of all trees have been looked at so far.
When the number of these multiples printed out reaches the number howmany (say 1000), the whole algorithm aborts and prints out that it has not found all most parsimonious trees, but prints out what is has got so far anyway. These trees need not be any of the most parsimonious trees: they are simply the most parsimonious ones found so far. By setting the product (howoften times howmany) large you can make the algorithm less likely to abort, but then you risk getting bogged down in a gigantic computation. You should adjust these constants so that the program cannot go beyond examining the number of trees you are reasonably willing to wait for. In their initial setting the program will abort after looking at 100,000 trees. Obviously you may want to adjust howoften in order to get more or fewer lines of intermediate notice of how many trees have been looked at so far. Of course, in small cases you may never even reach the first multiple of howoften and nothing will be printed out except some headings and then the final trees.
The indication of the approximate percentage of trees searched so far will be helpful in judging how much farther you would have to go to get the full search. Actually, since that fraction is the fraction of the set of all possible trees searched or ruled out so far, and since the search becomes progressively more efficient, the approximate fraction printed out will usually be an underestimate of how far along the program is, sometimes a serious underestimate.
A constant that affects the result is "maxtrees", which controls the maximum number of trees that can be stored. Thus if "maxtrees" is 25, and 32 most parsimonious trees are found, only the first 25 of these are stored and printed out. If "maxtrees" is increased, the program does not run any slower but requires a little more intermediate storage space. I recommend that "maxtrees" be kept as large as you can, provided you are willing to look at an output with that many trees on it! Initially, "maxtrees" is set to 100 in the distribution copy.
The counting of the length of trees is done by an algorithm nearly identical to the corresponding algorithms in MIX, and thus the remainder of this document will be nearly identical to the MIX document. MIX is a general parsimony program which carries out the Wagner and Camin-Sokal parsimony methods in mixture, where each character can have its method specified. The program defaults to carrying out Wagner parsimony.
The Camin-Sokal parsimony method explains the data by assuming that changes 0 --> 1 are allowed but not changes 1 --> 0. Wagner parsimony allows both kinds of changes. (This under the assumption that 0 is the ancestral state, though the program allows reassignment of the ancestral state, in which case we must reverse the state numbers 0 and 1 throughout this discussion). The criterion is to find the tree which requires the minimum number of changes. The Camin-Sokal method is due to Camin and Sokal (1965) and the Wagner method to Eck and Dayhoff (1966) and to Kluge and Farris (1969).
Here are the assumptions of these two methods:
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).
% fpenny Penny algorithm, branch-and-bound Phylip character discrete states file: penny.dat Phylip penny program output file [penny.fpenny]: How many trees looked Approximate at so far Length of How many percentage (multiples shortest tree trees this long searched of 100): found so far found so far so far ---------- ------------ ------------ ------------ 1 8.00000 1 6.67 2 8.00000 3 20.00 3 8.00000 3 53.33 4 8.00000 3 93.33 Output written to file "penny.fpenny" Trees also written onto file "penny.treefile" |
Go to the input files for this example
Go to the output files for this example
Penny algorithm, branch-and-bound Version: EMBOSS:6.4.0.0 Standard (Mandatory) qualifiers: [-infile] discretestates File containing one or more data sets [-outfile] outfile [*.fpenny] Phylip penny program output file Additional (Optional) qualifiers (* if not always prompted): -weights properties Phylip weights file (optional) -ancfile properties Phylip ancestral states file (optional) -mixfile properties Phylip mix output file (optional) -method menu [Wagner] Choose the method to use (Values: Wag (Wagner); Cam (Camin-Sokal); Mix (Mixed)) -outgrno integer [0] Species number to use as outgroup (Integer 0 or more) -howmany integer [1000] How many groups of trees (Any integer value) -howoften integer [100] How often to report, in trees (Any integer value) -simple boolean Branch and bound is simple -threshold float [$(infile.discretesize)] Threshold value (Number 1.000 or more) -[no]trout toggle [Y] Write out trees to tree file * -outtreefile outfile [*.fpenny] Phylip tree output file (optional) -printdata boolean [N] Print data at start of run -[no]progress boolean [Y] Print indications of progress of run -[no]treeprint boolean [Y] Print out tree -stepbox boolean [N] Print out steps in each site -ancseq boolean [N] Print states at all nodes of tree Advanced (Unprompted) qualifiers: (none) Associated qualifiers: "-outfile" associated qualifiers -odirectory2 string Output directory "-outtreefile" associated qualifiers -odirectory string Output directory General qualifiers: -auto boolean Turn off prompts -stdout boolean Write first file to standard output -filter boolean Read first file from standard input, write first file to standard output -options boolean Prompt for standard and additional values -debug boolean Write debug output to program.dbg -verbose boolean Report some/full command line options -help boolean Report command line options and exit. More information on associated and general qualifiers can be found with -help -verbose -warning boolean Report warnings -error boolean Report errors -fatal boolean Report fatal errors -die boolean Report dying program messages -version boolean Report version number and exit |
Qualifier | Type | Description | Allowed values | Default | ||||||
---|---|---|---|---|---|---|---|---|---|---|
Standard (Mandatory) qualifiers | ||||||||||
[-infile] (Parameter 1) |
discretestates | File containing one or more data sets | Discrete states file | |||||||
[-outfile] (Parameter 2) |
outfile | Phylip penny program output file | Output file | <*>.fpenny | ||||||
Additional (Optional) qualifiers | ||||||||||
-weights | properties | Phylip weights file (optional) | Property value(s) | |||||||
-ancfile | properties | Phylip ancestral states file (optional) | Property value(s) | |||||||
-mixfile | properties | Phylip mix output file (optional) | Property value(s) | |||||||
-method | list | Choose the method to use |
|
Wagner | ||||||
-outgrno | integer | Species number to use as outgroup | Integer 0 or more | 0 | ||||||
-howmany | integer | How many groups of trees | Any integer value | 1000 | ||||||
-howoften | integer | How often to report, in trees | Any integer value | 100 | ||||||
-simple | boolean | Branch and bound is simple | Boolean value Yes/No | No | ||||||
-threshold | float | Threshold value | Number 1.000 or more | $(infile.discretesize) | ||||||
-[no]trout | toggle | Write out trees to tree file | Toggle value Yes/No | Yes | ||||||
-outtreefile | outfile | Phylip tree output file (optional) | Output file | <*>.fpenny | ||||||
-printdata | boolean | Print data at start of run | Boolean value Yes/No | No | ||||||
-[no]progress | boolean | Print indications of progress of run | Boolean value Yes/No | Yes | ||||||
-[no]treeprint | boolean | Print out tree | Boolean value Yes/No | Yes | ||||||
-stepbox | boolean | Print out steps in each site | Boolean value Yes/No | No | ||||||
-ancseq | boolean | Print states at all nodes of tree | Boolean value Yes/No | No | ||||||
Advanced (Unprompted) qualifiers | ||||||||||
(none) | ||||||||||
Associated qualifiers | ||||||||||
"-outfile" associated outfile qualifiers | ||||||||||
-odirectory2 -odirectory_outfile |
string | Output directory | Any string | |||||||
"-outtreefile" associated outfile qualifiers | ||||||||||
-odirectory | string | Output directory | Any string | |||||||
General qualifiers | ||||||||||
-auto | boolean | Turn off prompts | Boolean value Yes/No | N | ||||||
-stdout | boolean | Write first file to standard output | Boolean value Yes/No | N | ||||||
-filter | boolean | Read first file from standard input, write first file to standard output | Boolean value Yes/No | N | ||||||
-options | boolean | Prompt for standard and additional values | Boolean value Yes/No | N | ||||||
-debug | boolean | Write debug output to program.dbg | Boolean value Yes/No | N | ||||||
-verbose | boolean | Report some/full command line options | Boolean value Yes/No | Y | ||||||
-help | boolean | Report command line options and exit. More information on associated and general qualifiers can be found with -help -verbose | Boolean value Yes/No | N | ||||||
-warning | boolean | Report warnings | Boolean value Yes/No | Y | ||||||
-error | boolean | Report errors | Boolean value Yes/No | Y | ||||||
-fatal | boolean | Report fatal errors | Boolean value Yes/No | Y | ||||||
-die | boolean | Report dying program messages | Boolean value Yes/No | Y | ||||||
-version | boolean | Report version number and exit | Boolean value Yes/No | N |
These programs are intended for the use of morphological systematists who are dealing with discrete characters, or by molecular evolutionists dealing with presence-absence data on restriction sites. One of the programs (PARS) allows multistate characters, with up to 8 states, plus the unknown state symbol "?". For the others, the characters are assumed to be coded into a series of (0,1) two-state characters. For most of the programs there are two other states possible, "P", which stands for the state of Polymorphism for both states (0 and 1), and "?", which stands for the state of ignorance: it is the state "unknown", or "does not apply". The state "P" can also be denoted by "B", for "both".
There is a method invented by Sokal and Sneath (1963) for linear sequences of character states, and fully developed for branching sequences of character states by Kluge and Farris (1969) for recoding a multistate character into a series of two-state (0,1) characters. Suppose we had a character with four states whose character-state tree had the rooted form:
1 ---> 0 ---> 2 | | V 3
so that 1 is the ancestral state and 0, 2 and 3 derived states. We can represent this as three two-state characters:
Old State New States --- ----- --- ------ 0 001 1 000 2 011 3 101
The three new states correspond to the three arrows in the above character state tree. Possession of one of the new states corresponds to whether or not the old state had that arrow in its ancestry. Thus the first new state corresponds to the bottommost arrow, which only state 3 has in its ancestry, the second state to the rightmost of the top arrows, and the third state to the leftmost top arrow. This coding will guarantee that the number of times that states arise on the tree (in programs MIX, MOVE, PENNY and BOOT) or the number of polymorphic states in a tree segment (in the Polymorphism option of DOLLOP, DOLMOVE, DOLPENNY and DOLBOOT) will correctly correspond to what would have been the case had our programs been able to take multistate characters into account. Although I have shown the above character state tree as rooted, the recoding method works equally well on unrooted multistate characters as long as the connections between the states are known and contain no loops.
However, in the default option of programs DOLLOP, DOLMOVE, DOLPENNY and DOLBOOT the multistate recoding does not necessarily work properly, as it may lead the program to reconstruct nonexistent state combinations such as 010. An example of this problem is given in my paper on alternative phylogenetic methods (1979).
If you have multistate character data where the states are connected in a branching "character state tree" you may want to do the binary recoding yourself. Thanks to Christopher Meacham, the package contains a program, FACTOR, which will do the recoding itself. For details see the documentation file for FACTOR.
We now also have the program PARS, which can do parsimony for unordered character states.
7 6 Alpha1 110110 Alpha2 110110 Beta1 110000 Beta2 110000 Gamma1 100110 Delta 001001 Epsilon 001110 |
If the Camin-Sokal parsimony method (option C or S) is invoked and the A option is also 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 state changes. If these are tied, then it may not be possible for the program to infer the state in the internal nodes, and these will all be printed as ".". If this has happened and you want to know more about the states at the internal nodes, you will find helpful to use MOVE to display the tree and examine its interior states, as the algorithm in MOVE shows all that can be known in this case about the interior states, including where there is and is not amibiguity. The algorithm in PENNY gives up more easily on displaying these states.
If the A option is not used, then the program will assume 0 as the ancestral state for those characters following the Camin-Sokal method, and will assume that the ancestral state is unknown for those characters following Wagner parsimony. If any characters have unknown ancestral states, and if the resulting tree is rooted (even by outgroup), a table will be printed out showing the best guesses of which are the ancestral states in each character. You will find it useful to understand the difference between the Camin-Sokal parsimony criterion with unknown ancestral state and the Wagner parsimony criterion.
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 program finds multiple trees tied for best, all of these are written out onto the output tree file. Each is followed by a numerical weight in square brackets (such as [0.25000]). This is needed when we use the trees to make a consensus tree of the results of bootstrapping or jackknifing, to avoid overrepresenting replicates that find many tied trees.
Penny algorithm, version 3.69 branch-and-bound to find all most parsimonious trees Wagner parsimony method requires a total of 8.000 3 trees in all found +-----------------Alpha1 ! ! +--------Alpha2 --1 ! ! +-----4 +--Epsilon ! ! ! +--6 ! ! +--5 +--Delta +--2 ! ! +-----Gamma1 ! ! +--Beta2 +-----------3 +--Beta1 remember: this is an unrooted tree! +-----------------Alpha1 ! --1 +--------------Alpha2 ! ! ! ! +--Epsilon +--2 +--6 ! +-----5 +--Delta ! ! ! +--4 +-----Gamma1 ! ! +--Beta2 +--------3 +--Beta1 remember: this is an unrooted tree! +-----------------Alpha1 ! ! +-----Alpha2 --1 +--------2 ! ! ! +--Beta2 ! ! +--3 +--4 +--Beta1 ! ! +--Epsilon ! +--6 +--------5 +--Delta ! +-----Gamma1 remember: this is an unrooted tree! |
(Alpha1,((Alpha2,((Epsilon,Delta),Gamma1)),(Beta2,Beta1)))[0.3333]; (Alpha1,(Alpha2,(((Epsilon,Delta),Gamma1),(Beta2,Beta1))))[0.3333]; (Alpha1,((Alpha2,(Beta2,Beta1)),((Epsilon,Delta),Gamma1)))[0.3333]; |
Program name | Description |
---|---|
eclique | Largest clique program |
edollop | Dollo and polymorphism parsimony algorithm |
edolpenny | Penny algorithm Dollo or polymorphism |
efactor | Multistate to binary recoding program |
emix | Mixed parsimony algorithm |
epenny | Penny algorithm, branch-and-bound |
fclique | Largest clique program |
fdollop | Dollo and polymorphism parsimony algorithm |
fdolpenny | Penny algorithm Dollo or polymorphism |
ffactor | Multistate to binary recoding program |
fmix | Mixed parsimony algorithm |
fmove | Interactive mixed method parsimony |
fpars | Discrete character parsimony |
Please report all bugs to the EMBOSS bug team (emboss-bug © emboss.open-bio.org) not to the original author.
Converted (August 2004) to an EMBASSY program by the EMBOSS team.