Welcome to pypath!§

Demo

OmniPath§

Are you interested in OmniPath data? Check out our R package OmnipathR, the most popular and most versatile access point to OmniPath, a database built from more than 150 original resources. If you use Python and don’t need to build the database yourself, try our Python client. Read more about the web service here.

Do you need pypath?§

Pypath is the database builder of OmniPath. For most people the data distributed in OmniPath is satisfying (see above), they don’t really need pypath. Typically you need pypath to:

  • Build a custom or very fresh version of the OmniPath database(s)

  • Use one of the utilities such as ID translation, homology translation, etc. (see the utils module)

  • Access the raw or preprocessed data directly from the original resources (see the inputs module)

Installation§

From PyPI:

pip install pypath-omnipath

From Git:

pip install git+https://github.com/saezlab/pypath.git

Docs§

Read the reference documentation or check out the tutorials. The most comprehensive guide to pypath is The Pypath Book.

Get help§

Should you have a question or experiencing an issue, please write us by the Github issues page.

Features§

pypath is a Python module for processing molecular biology data resources, combining them into databases and providing a versatile interface in Python as well as exporting the data for access through other platforms such as R, web service, Cytoscape and BEL (Biological Expression Language).

pypath provides access to more than 100 resources! It builds 5 major combined databases and within these we can distinguish different datasets. The 5 major databases are interactions (molecular interaction network or pathways), enzyme-substrate relationships, protein complexes, molecular annotations (functional roles, localizations, and more) and inter-cellular communication roles.

pypath consists of a number of submodules and each of them again contains a number of submodules. Overall pypath consists of around 100 modules. The most important higher level submodules:

  • pypath.core: contains the database classes e.g. network, complex, annotations, etc

  • pypath.inputs: contains the resource specific methods which directly downlad and preprocess data from the original sources

  • pypath.omnipath: higher level applications, e.g. a database manager, a web server

  • pypath.utils: stand alone useful utilities, e.g. identifier translator, Gene Ontology processor, BioPax processor, etc

Integrated databases§

In the beginning the primary aim of pypath was to build networks from multiple sources using an igraph object as the fundament of the integrated data structure. From version 0.7 and 0.8 this design principle started to change. Today pypath builds a number of different databases, exposes them by a rich API and each of them can be converted to pandas.DataFrame. The modules and classes responsible for the integrated databases are located in pypath.core. The five main databases are the followings:

  • network - core.network

  • enzyme-substrate - core.enz_sub

  • complexes - core.complex

  • annotations - core.annot

  • intercell - core.intercell

Some of the databases have different variants (e.g. PPI and transcriptional network) and all can be customized by many parameters.

Database management§

The databases above can be loaded by calling the appropriate classes. However building the databases require time and memory so we want to avoid building them more often than necessary or keeping more than one copies in the memory. Some of the modules listed above have a method get_db which ensures only one instance of the database is loaded. But there is a more full featured database management system available in pypath, this is the pypath.omnipath module. This module is able to build the databases, automatically saves them to pickle files and loads them from there in subsequent sessions. pypath comes with a number of database definitions and users can add more. The pickle files are located by default in the ~/.pypath/pickles/ directory. With the omnipath module it’s easy to get an instance of a database. For example to get the omnipath PPI network dataset:

from pypath import omnipath
op = omnipath.db.get_db('omnipath')

Important: Building the databases for the first time requires the download of several MB or GB of data from the original resources. This normally takes long time and is prone of errors (e.g. truncated or empty downloads due to interrupted HTTP connection). In this case you should check the log to find the path of the problematic cache file, check the contents of this file to find out the reason and possibly delete the file to ensure another download attempt when you call the database build again. Sometimes the original resources change their content or go offline. If you encounter such case please open an issue at saezlab/pypath#issues so we can fix it in pypath. Once all the necessary contents are downloaded and stored in the cache, the database builds are much faster, but still can take minutes.

Further modules in pypath§

Apart from the databases, pypath has many submodules with standalone functionality which can be used in other modules and scripts. Below we present a few of these.

ID conversion§

The ID conversion module utils.mapping translates between a large variety of gene, protein, miRNA and small molecule ID types. It has the feature to translate secondary UniProt ACs to primaries, and Trembl ACs to SwissProt, using primary Gene Symbols to find the connections. This module automatically loads and stores the necessary conversion tables. Many tables are predefined, such as all the IDs in UniProt mapping service, while users are able to load any table from file using the classes provided in the module input_formats. An example how to translate identifiers:

from pypath.utils import mapping
mapping.map_name('P00533', 'uniprot', 'genesymbol')
# {'EGFR'}

Homology translation§

The pypath.utils.homology module is able to find the orthologs of genes between two organisms. It uses data both from NCBI HomoloGene, Ensembl and UniProt. This module is really simple to use:

from pypath.utils import homology
homology.translate('P00533', 10090) # translating the human EGFR to mouse
# ['Q01279'] # it returns the mouse Egfr UniProt AC

It is able to handle any ID type supported by pypath.utils.mapping. Alternatively, you can access a complete dictionary of orthologous genes, or translate columns in a pandas data frame.

FAQ§

Does it run on my old Python?

Most likely it doesn’t. The oldest supported version, currently 3.9, is defined in our pyproject.toml.

Is there something similar in R?

OmniPath’s R client, besides accessing data from OmniPath, provides many similar services as pypath: ID translation, homology translation, taxonomy support, GO support, and many more.

Questions about OmniPath

Contact§

We prefer to keep all communication within the Github issues. About private or sensitive matters feel free to contact us by omnipathdb@gmail.com.

Impressum§

pypath is developed in the Saez Lab by Dénes Türei, Sebastian Lobentanzer and Ahmet Rifaioglu, and in HU Biological Data Science Lab (PI: Tunca Doğan) by Melih Darcan, Erva Ulusoy, Ömer Kaan Vural, Tennur Kılıç, Elif Çevrim, Bünyamin Şen and Atabey Ünlü. Olga Ivanova and Nicolàs Palacio also contributed in the past. The R package and the Cytoscape app are developed and maintained by Francesco Ceccarelli, Attila Gábor, Alberto Valdeolivas, Dénes Türei and Nicolàs Palacio. The Python client for the OmniPath web service has been developed and is maintained by Michael Klein in the group of Fabian Theis.

History and releases§

See here a bird eye view of pypath’s development history. For more details about recent developments see the Github releases.