pahmm 0.1.2

Pairwise statistical phylogenetic distance estimation using pair hidden Markov models for Python.

paHMM-Tree Library for Python and C

paHMM-Tree is short for "pairwise statistical phylogenetic distance estimation using pair hidden Markov models". It's a tool developed by Marcin Bogusz and Simon Whelan. This library, "libpahmm" exposes the tool's capabilities to Python as well as C.

What is libpahmm?

It's a library for finding evolutionary distances between a set of sequences of nucleotides or amino-acids (such as DNA, RNA, proteins, etc.). The library lets you configure the way distance calculations are made. These distances can then be used to create phylogenetic trees to help visualize genetic relationships.

Why use libpahmm?

• It allows for more accurate distance calculations that take into account multiple substitutions per site on the sequences.
• It has good algorithmic time complexity: O(n^2 * L^2) where n is the number of sequences and L is their length (assuming all have the same length).
• It's implemented in C++ internally, most work is done by compiled code and not Python.
• It's free and open-source.

Basic usage

Let's see how libpahmm can be used in Python and C. This following examples will calculate the distance between two sequences from a FASTA-file containing DNA-data.

Using Python

from pahmm import *

# 1. Create a banding estimator object
be = BandingEstimator()

# 2a. (Alternative a) Load a file
be.set_file_input("sequences.fasta")

# 2b. (Alternative b) Load a string
be.set_str_input(b"[sequences in FASTA-format here]")

# 3. Load the desired distance-calculation method
sequences = be.execute_hky85_model()

# 4. Calculate distance between two sequences
d = sequences.get_distance_from_names(b"seq1", b"seq2")

Using C

1. Create a banding estimator object:

EBCBandingEstimator *be = ebc_be_create();

2a. (Alternative a) Load a file:

ebc_be_set_input_from_file(be, "sequences.fasta");

2b. (Alternative b) Load a string:

ebc_be_set_input(be, "[sequences in FASTA-format here]");

3. Load the desired distance-calculation method:

EBCSequences *sequences = ebc_be_execute_hky85_modelv2(be);

4. Calculate distance between two sequences:

double d = ebc_seq_get_distance(sequences, "seq1", "seq2");

5. Clean up resources:

ebc_seq_free(sequences);
ebc_be_free(be);

Installation

Python-library installation can easily be done as such:

python3 -m pip install pahmm

Windows notice: The library build-scripts (CMakeList.txt and setup.py) have not been tested on Windows, only on Linux and MacOS. Therefore, the library does not have official Windows support at the time being.

C-library notice: Currently there are no binary packages distributed for the C-library. If you wish to use libpahmm as a C-library, you could compile it yourself by following the instructions under the Compilation-section of this README.

Compilation

To compile libpahmm you'll need to install CMake. This can be done with the following command on Linux (Ubuntu):

sudo apt install cmake

On MacOS I'd recommend using Brew for easy installation:

brew install cmake

Then navigate to "libpahmm/", and enter the following command to build for Python 3:

python3 setup.py install

Note: Instead of "install" you could write "develop" (installs, but doesn't copy any binaries to Python's site-packages folder). Use "clean" to clean up the source directory from any build files (excluding tests/).

To build and install the C-library, just run:

mkdir build
cd build
cmake ..
make
sudo make install

To uninstall, look inside install_manifest.txt and delete everything that's listed there on your system. This can be done in the following way. Make sure to check that install_manifest.txt contains the files you actually want to delete:

cat install_manifest.txt

If everything you want to remove is listed in the file, then run the following command:

sudo xargs rm < install_manifest.txt

Directory structure

The following is a description of what's in every directory in the root of the repository.

• auxiliary - Files needed for the Python build script.
• dlib - The dlib library, it's a dependency of paHMM-dist.
• examples - Example code that shows how the library can be used.
• include - Header files for the C pahmm library.
• paHMM-dist - A slightly modified version of the original paHMM-dist tool.
• python - The pahmm Python library.
• src - The source code for the C pahmm library, it's a thin wrapper around paHMM-dist.
• tests - Tests to verify the correctness of the library.

Tests

There are tests that you can run if you wish to verify the correctness of the library output, the test-suite can be found under tests/. They verify that the output from the library match the output from the paHMM-Tree tool. You need to have the pahmm Python library installed before you run the tests.

License

Libpahmm and paHMM-Tree are licensed under the GNU General Public License version 3.0 (GPLv3).

Dlib is licensed under Boost Software License - Version 1.0.

Credits

• Marcin Bogusz and Simon Whelan for the development of paHMM-Tree: https://github.com/marbogusz/paHMM-Tree

• Lars Arvestad for refactoring paHMM-Tree: https://github.com/arvestad/paHMM-dist

• Mazen Mardini for the paHMM-Tree library code and more refactoring of paHMM-Tree: https://github.com/mazen-mardini/libpahmm