mattools — utilities for distance matrices

Phylip distance matrices are a common intermediate format in phylogeny reconstruction. This tool set consists of a small set of utilities for their manipulation, formatting and comparison.

This readme covers all necessary instructions to get the mattools up and running.

Usage

Let us assume you have mat installed via the way described below. Then you get help to mat and all its subcommands via the option --help.

$ mat --help
The available commands are:
 compare     Compute the distance between two matrices
 format      Format the distance matrix
 grep        Print submatrix for names matching a pattern
 nj          Convert to a tree by neighbor joining

Use 'mat <command> --help' to get guidance on the usage of a command.

Comparison

To check whether two distance matrices are equal, use mat compare. The input matrices will be interpreted as two vectors and their euclidean distance computed. Thus a distance of zero indicates equality. To circumvent problem with differently sized matrices, only those values are included in the computation, whose corresponding names equal.

Formatting

Unfortunately, the phylip distance matrix format is poorly designed, described, and badly implemented in different tools. With mat format all these formatting differences can be removed.

$ cat lowertriangle.mat
2
A  
B  0.5
$ mat format lowertriangle.mat
2
A          0.0000e+00 5.0000e-01
B          5.0000e-01 0.0000e+00

To verify that the distance matrix is indeed a distance matrix in the mathematical sense, the option --validate can be used. The mat tools will then hunt for errors and try to fix them, where possible.

Filtering

To remove or extract individual lines and submatrices mat grep can be used. It takes a regular expression and checks it against the names. Names, not matching the pattern are discarded from the output. This behaviour can be changed with the flag --invert-match.

Neighbor Join

The mattools also come with a module for building a phylogeny via neighbor joining. The resulting tree also contains support values computed via quartet analysis. See the following paper for a description of the process: Klötzl & Haubold (2016).

Building

The mattools require the BOOST library as a dependency. Next, clone this repository and then build the programs.

$ git clone https://github.com/evolbioinf/mattools
$ autoreconf -fi
$ ./configure
$ make
$ make install  # try sudo

License

This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. The full license text is available at http://gnu.org/licenses/gpl.html.

Individual files may be licensed differently.

Contact

In case of bugs or unexpected errors don't hesitate to send me a mail: kloetzl@evolbio.mpg.de

EvolBioInf / mattools

readme