(obs: the code is now hosted at google code, at the address https://code.google.com/p/biomcmc/ . The instructions below depend on a repository located at https://corn.ab.a.u-tokyo.ac.jp, which uses a self-signed security certificate. Some browsers (all those that I've tried, actually) will complain about it. Please neglect the apocalyptic warnings and accept the certificate if you want to download this version. Or you can go to google code.)
The binary files contain four programs:
To compile the program you must have a system with a gcc compiler. The compilation is based on autotools (autoconf and automake), such that it should compile on Unix and Mac OSX. For Windows I could compile in the past by using a software called interix (when installing this product, I selected everything with SDK on the name). Maybe cygwin is enough.
download the pre-compiled binaries for Mac OSX 32 bits (biomc2-1.9) [alternative location]
download the pre-compiled binaries for AMD 64 bits (biomc2-1.9) [alternative location]
download the source code (biomc2-1.9) [alternative location]
The output from biomc2 (the main program) was greatly simplified in this version. Now we have only two "small" files, post.dist and post.tree. post.dist has basically the break-point locations for each sample and post.tree has the topologies for each segment, for each sample. The program biomc2.summarise becomes then indispensable to interpret the posterior samples and summarise the information. Some summary information will be plotted to a file called recomb_freq.pdf using the libGTK2.0 library. In Debian/Ubuntu this library can be installed with the command
$ apt-get install libgtk2.0-dev
download the pre-compiled binaries for Intel 32 bits (biomc2-1.7)
download the pre-compiled binaries for AMD 64 bits (biomc2-1.7)
download the source code (biomc2-1.7)
This is version 1.7 of the biomc2 software, used in the PLoS ONE (2008) paper. If you are having trouble accessing the files, try the original location.
Please notice that the original version of the program will generate many nexus tree files (one for each segment) and a huge text file with the posterior sample of parameters, consuming a lot of disk. Some systems impose a limit on the number of concurrently open files per program, in which cases the program will fail. There are also situations where the program abort with beautiful error messages (most of the times is a malformed input file - the functions that read nexus files are very strict).
Simulated datasets used in de Oliveira Martins L, Leal E, Kishino H (2008) PLoS ONE 3(7): e2651. doi:10.1371/journal.pone.00026518 taxa: input, output, scripts and comparison files
12 taxa: input, output and scripts files
16 taxa: input, output and scripts files
The script files are used to build up the datasets and conduct the analyses "on the fly", on a unix system. They also plot the results. Perl, Ziheng Yang's PAML and R are necessary to run the scripts. The main purpose of these script files is to provide information about the simulation scenario and how to run the programs. Since they are based on version 1.7, they should not work with newer versions.
The input files are example simulated datasets, and the output files are the processed output files (posterior distribution and summarised results). The input and output files do not refer to the same datasets for 12 and 16 sequences, since when running the analysis the original input files are overwritten (lost). The input files are provided as a convenience if there is interest in doing recombination analysis using other methods (benchmark).
For 8 taxa datasets there are furthermore comparison files for the whole analysis using biomc2, cBrother and MrBayes as described in figure 3 of the PLoS ONE paper.
input files composed of the HIV-1 dataset alignment, the starting tree and the input file for the biomc2 program.
output files (processed posterior distribution) for two independent runs.