Installation

Fundamental dependencies

python v2.7.*
gcc
hdf5

Python packages

numpy>=1.7.1
pysam == 0.10.0
h5py >= 2.0.1
pbcore >= 0.9.4
scipy >= 0.12.0
biopython >= 1.6.1
matplotlib >= 1.5.0

All but Numpy will be installed automatically during the standard installation as described below. Numpy, however, must be installed prior to mBin installation (see below):

Setting up virtualenv

mBin should be installed in a Python virtual environment using virtualenv, which creates a clean and isolated Python environment in which to install packages and their dependencies.

Virtualenv can be installed using pip:

$ pip install virtualenv

Once installed, navigate to the directory where you would like to keep the virtual environment and create a virtual environment called venv:

$ virtualenv venv

Finally, activate this virtual environment venv:

$ . venv/bin/activate

Once activated, you are now operating inside the venv and should see the following on you command line:

(venv)<COMMAND LINE>

Installing mBin

With the virtual environment activated, install mbin using pip:

$ pip install mbin

Installing t-SNE

In order to create 2-D maps of methylation (and other) features for binning using mapfeatures, we must install the Barnes-Hut implementation of the t-SNE algorithm. Full details on the BH-tSNE algoritm and wrapper script can be found here. First, we pull the repository from GitHub and enter the directory:

$ git clone https://github.com/lvdmaaten/bhtsne.git
$ cd bhtsne

Next we compile the source code to get the executable bh_tsne:

$ g++ sptree.cpp tsne.cpp tsne_main.cpp -o bh_tsne -O2

Once the executable bh_tsne is compiled, add this directory to your $PATH environmental variable:

$ export PATH=$PATH:`pwd`

If bh_tsne is accessible in the path, the following should list usage instructions for mapfeatures:

$ mapfeatures --help