PyPathway

integrated Python toolkit for pathway based analysis

PyPI version Build Status

Installation

Install PyPathway via Anaconda is recommended.

  1. Download and install anaconda from Anaconda site
  2. Install PyPathway by
conda install -c steamedsheep pypathway

NOTE: If you want to install pypathway via pypi, please refer to the Installation section

Features

  • Public databases APIs: STRING, BioGRID, KEGG, Reactome and WikiPathway
  • Functional set based and network based enrichment analysis algorithms implemented: ORA, GSEA and SPIA
  • Performance optimize for denovo enrichment algorithm MAGI and Hotnet2.
  • Network propagation algorithms random walk, RWR and heat kernel.
  • Interactive visualization and web page exportation for pathway, graph and analysis result.
  • Integrated with pandas, networkx and numpy. Most of the methods accept both text file and data structure from these packages
  • Dynamic visualization for IPython notebook.
  • Most classes implement __repr__ method for interactive environment.

Network process

Intuitive APIs for querying and retrieval interaction network from public database. The return object are stored in networkx.Graph object.

Support databases

  • KEGG
  • Reactome
  • WikiPathway
  • STRING
  • BioGRID
from pypathway import PublicDatabase
kg = PublicDatabase.search_kegg('CD4')
wp = PublicDatabase.search_wp('CD4')
rt = PublicDatabase.search_reactome('CD4')

Load

pathway = r[0].load()

Plot

pathway.draw()

IPython notebook examples

Enrichment Analysis

Support methods

  • ORA
  • GSEA
  • Network enrichment (SPIA and Enrichment)
  • denovo enrichment (MAGI and Hotnet2)

Implementation / Interface

  • Staticmethod run() for the starting of the analysis
r = SPIA.run(all=c.background, de=c.deg, organism='hsa')
  • table, plot() and graph() method for the presentation of the analysis
res.table


res.plot()


res.graph()

IPython examples

Modeling

  • the Python Interface and optimize for MAGI
  • several c extension for Hotnet permutation performance

Propagation

Implemented algorithms

  • Random walk
random_walk(G, h)
  • Random walk with restart
random_walk_with_restart(G, h, rp=0.7, n=-1)
  • Heat kernel
diffusion_kernel(G, h, rp=0.8, n=100)

Implementation detail

image source: Network propagation: a universal amplifier of genetic associations

IPython notebook examples

Utility and Performance

  • The Id converter
  • GMT file manager
  • network and expression data sets.
  • numpy implementation of SPIA
  • node swap c extension for Hotnet2
  • multi-threading for MAGI