Camoco is a python library modeled after the popular package pandas for building and analyzing co-expression networks. It exposes a nice API and is effecient for working with such large data structures. It was created to be used interactively through something like iPython or normal scripting. All it takes to get started is an import:

import camoco as co 

This will expose many of the cool classes camoco implements. Most of the classes in the library inherit directly from the Camoco class. This is because many of the data structures the library implements are expensive to make and the library itself relies heavily on caching. sqlite3 and hdf5 files will be created from you input datasets in a directory specified in ~/.camoco.conf

The base camoco class is almost entirely abstract. This means that you never really interact with a camoco class directly, instead many other classes inherit properties from the camoco class. This is nice because, as I said before, many of the data structures are expensive to compute. What camoco does is to require datasets to have names and descriptions, so after they are built, they can be cached on disk using sqlite and hdf5. This makes it fast to come back! On disk objects are stored in a directory specified as basedir in the configuration file: ~/.camoco.conf . This is all done automatically, so if camoco is doing its job, you shouldn't ever have to interact with it directly. It should just work! (not guaranteed ... ).

The other methods which are exposed when you import camoco are housekeeping methods. You can view, delete, move, and rename camoco datasets using the following methods:

# See docstrings for more information
co.available_datasets()
co.del_dataset()
co.mv_dataset()
co.redescribe_dataset() 

Other than that, Camoco has three main classes which each have some companion classes. These classes/subclasses are as follows:

• RefGen and Locus
• COB and Expr
• Ontology and Term

Python has a fantastic range of built in data types. It lacks however in a data type which succinctly represents a locus. Loci cover a large range of biological data types; a gene is a locus, as is a SNP, or a QTL, or an entire chromosome. It all depends on context.

Tagged releases are available on PyPi: pip install camoco. Otherwise, you should be able to run python setup.py install once you have the following dependencies resolved:

Required: python>=3.3

Hints on how we routinely build camoco can be found in our Travis-CI configuration file, .travis.yml. Following these steps should result in a successful build (at least on our test platform).

We decided to use APSW as an sqlite3 python wrapper, instructions on building apsw can be found here

Required (install in this order to save pain):

• cython>=0.16
• numpy>=1.9.1
• scipy>=0.15
• ipython>=3.1.0
• pandas>=0.16
• numexpr>=2.0.0
• tables>=3.0.0 (hdf5)
• apsw>=3.8
• matplotlib>=1.4.3
• statsmodels>=0.6.1
• termcolor>=1.1.0
• pyaml==15.6.3

Unit tests within Camoco are implemented using the pytest framework. Unit tests as well as Build tests can be found in the test directory.