2. Installation

2.1. System requirements

Mikado requires CPython 3.4 or later to function (Python2 is not supported). Additionally, it requires a functioning installation of one among SQLite, PostgreSQL and MySQL. Mikado has additionally the following python dependencies:

wheel>=0.28.0
pyyaml
jsonschema
cython>=0.25
numpy
networkx>=1.10
sqlalchemy>=1
sqlalchemy_utils
biopython>=1.66
intervaltree
nose
pyfaidx
scikit-learn>=0.17.0
scipy>=0.15.0
frozendict
python-magic
drmaa
snakemake
docutils!=0.13.1
tabulate
ujson

Mikado can run with little system requirements, being capable of analysing human data with less than 4GB of RAM in all stages. It benefits from the availability of multiple processors, as many of the steps of the pipeline are parallelised.

Mikado is at its core a data integrator. The Daijin pipeline has been written to facilitate the creation of a functioning workflow. To function, it requires Snakemake [Snake] and the presence of at least one supported RNA-Seq aligner and one supported transcript assembler. If you are planning to execute it on a cluster, we do support job submission on SLURM, LSF and PBS clusters, either in the presence or absence of DRMAA.

2.2. Download

Mikado is available on PyPI, so it is possible to install it with

pip3 install mikado

The source for the latest release on PyPI can be obtained with

pip3 download mikado

As the package contains some core Cython components, it might be necessary to download and compile the source code instead of relying on the wheel.

Alternatively, Mikado can be installed from source by obtaining it from our GitHub repository. Either download the latest release or download the latest stable development snapshot with

git clone https://github.com/lucventurini/mikado.git; cd mikado

If you desire, the unstable development version can be obtained with the command

git checkout development

in the Git directory. Please note that at the time of this writing development proceeds quite rapidly.

2.3. Building and installing from source

If you desire to install Mikado from source, this can be achieved with

python3 setup.py bdist_wheel

Followed by

pip3 install dist/*whl

..note:

If you want to update your installation of Mikado, the command to be executed is pip install -U dist/*whl

2.4. Testing the installed module

It is possible to test whether Mikado has been built successfully by opening a python3 interactive session and digiting:

>>  import Mikado
>>  Mikado.test()

This will run all the tests included in the suite. Although code coverage is not perfect yet, it is at 70% for the whole package and over 80% for the core components.