pySignora

pySignora is a Python implementation of pathway gene-pair SIGnature OverRepresentation Analysis (SIGORA).

SIGORA has been published in PeerJ and implemented in R (https://cran.r-project.org/web/packages/sigora/index.html) by Foroushani et al. (2013). Shortly, the method compiles a set of weighted markers, pathway gene-pair signatures (Pathway-GPS), for each pathway in a repository. Subsequently, it identifies statistically overrepresented Pathway-GPSs in a user-specified gene list, using an adapted version of the hypergeometric test. A pathway here is defined as a set of genes with the same pathway-ontology annotation. A pathway gene-pair signature is a pair of genes that, as a combination, are specific to a single pathway. The weight of a GPS expresses the average specificity of the two gene components towards the common pathway. The weight range is [0, 1].

For more details, see the inventors' original publication: https://peerj.com/articles/229/

The two core functions, GPS-generation and signature ORA, have been implemented, but the package is still under development and many important features will be added in the future:

• Multiple-testing correction
• KEGG, Reactome, GO and other databases
• Handling hierarchical (multilevel) repositories

Usage

The input pandas dataframe (repodf) must have 3 columns: pathway, description, gene, e.g.:

import pandas as pd
repodf[:3]
#              pathwayId                           pathwayName             gene
# 1  ecadherin_1_pathway           E-cadherin signaling events  ENSG00000168036
# 2  ecadherin_1_pathway           E-cadherin signaling events  ENSG00000039068
# 3   syndecan_2_pathway  Syndecan-2-mediated signaling events  ENSG00000101680

import pysignora as ps
gps = ps.makegps(repodf)
res = ps.signora(genelist, gps)

Versioning

pySignora uses the semantic versioning scheme.

License = GPLv3+

pySignora is free software, which means the users have the freedom to run, copy, distribute, study, change and improve the software.

pySignora was written by Peter Vegh and is released under the GPLv3+ license.