CONSEL does not estimate the phylogeny by itself, but CONSEL reads the output of the existing phylogenetic packages Molphy, PAML, PAUP*, TREE-PUZZLE, and PhyML. You must be a user of one of these programs to get the "site-wise log-likelihoods" for the trees. For calculating the p-values, CONSEL does not need the molecular sequences, but only the matrix of the site-wise log-likelihoods.
The program package CONSEL must be installed on your computer. Windows users just download the self-extracting archive, and execute it. Unix users download the tar-gzipped archive of the source files, and run make. In either case, installing CONSEL is very easy.
Use option "-l name" for protml and nucml. For example,
will produce the file mam15.lls. This contains the site-wise log-likelihoods for the trees specified in 15.tpl. The program makermt in CONSEL reads the lls file to generate the bootstrap replicates;
produces the file mam15.rmt. You can omit the default extension ".lls" for the input file. Usually, makermt takes a few minutes to several hours depending on the size of the lls file. Once you get the rmt file, you can run the main program consel;
This reads mam15.rmt and produces mam15.pv. The p-values are printed by
which reads mam15.pv.
No option is needed for baseml and codeml to get the site-wise log-likelihoods. The file lnf or lfh is created automatically at the working directory. You may rename it, for example, mam15.lnf here. The program makermt in CONSEL reads the lnf file to generate the bootstrap replicates;
produces the file mam15.rmt. You can omit the default extension ".lnf" for the input file. Usually, makermt takes a few minutes to several hours depending on the size of the lnf file. Once you get the rmt file, you can run the main program consel;
This reads mam15.rmt and produces mam15.pv. The p-values are printed by
which reads mam15.pv.
*************
PAUP* outputs the site-wise log-likelihoods for the candidate trees by the option "SiteLikes=Yes" with "ScoreFile=mam15.txt", say. (I have received several emails from PAUP* users for the latest paup output formats, Thanks a lot!) Here is a sample PAUP* output file: paup2.txt. The current version (01j or later) of CONSEL ***DOES *** accept gaps (-) in the file as well as parsimony output files.
NOTE 2008/09/12: Version 0.1j includes several updates contributed by the following people.
(1) Jiaye Yu contributed PAUP header for parsimony analysis.
(2) Jeff Craig at Washington State University contributed PAUP gap (-) handling, and mac/win/dos line breaks handling.
(3) I also modified the code so that -d 2 option outputs the details of reading files. For example, seqmt -d 2 --paup paup2.txt may output what values are read now. This option helps to identify the location in the input file where an error can be found.
************
I assume a file mam15.txt is the output of PAUP* below. The program makermt in CONSEL reads the output of PAUP* to generate the bootstrap replicates;
produces the file mam15.rmt. You can omit the default extension ".txt" for the input file. Usually, makermt takes a few minutes to several hours depending on the size of the input file. Once you get the rmt file, you can run the main program consel;
This reads mam15.rmt and produces mam15.pv. The p-values are printed by
which reads mam15.pv.
TREE-PUZZLE outputs the site-wise log-likelihood file in the following steps. (Thanks for these instructions as well as the sample file from a tree-puzzle user!) (1) Start tree-puzzle using the -wsl option: treepuzzle -wsl <datafile> <treefile>, where <datafile> is the filename of the phylip format data and <treefile> is the file containing phylip format trees for which site-wise likelihoods are required. (2) Set options in tree-puzzle as required and run analysis. Note, “tree searching procedure” should be left as “evaluate user defined trees”. (3) This results in a file called <datafile>.sitelh which contains the site-wise likelihoods. Here is a sample TREE-PUZZLE output file: puzzle1.sitelh.
I assume a file mam15.sitelh is the output of TREE-PUZZLE below. The program makermt in CONSEL reads the output of TREE-PUZZLE to generate the bootstrap replicates;
produces the file mam15.rmt. You can omit the default extension ".sitelh" for the input file. Usually, makermt takes a few minutes to several hours depending on the size of the input file. Once you get the rmt file, you can run the main program consel;
This reads mam15.rmt and produces mam15.pv. The p-values are printed by
which reads mam15.pv.
PhyML outputs the site-wise likelihood file by "--print_site_lnl" option. (It is likelihood, not log-likelihood). I got a sample file (sample-phyml.txt) and the instruction from a PhyML user; thanks a lot! Since version 0.1k, CONSEL accepts this format by specifying "--phyml" either in seqmt and makermt. For example, we can generate the bootstrap replicates by
which produces the file sample-phyml.rmt. If there is anything wrong, "seqmt -d 2 --phyml sample-phyml.txt" will help you to identify where the input file was wrong. Once you get the rmt file, you can run the main program consel;
This reads sample-phyml.rmt and produces sample-phyml.pv. The p-values are printed by
which reads sample-phyml.pv.
The typical output of catpv is like this (mam15-pv.txt). In the output of catpv, the trees are sorted in the decreasing order of the log-likelihoods.