Network Enrichment Analysis: Method, Software, And Web-Site For
Functional Interpretation Of 'Omics' Data In Global Networks Andrey Alexeyenko
User-friendly implementation on-line
https://www.evinet.org/
E-mail: [email protected] Telefon: +46 8 52481513
http://research.scilifelab.se/andrej_alexeyenko/ http://ki.se http://bils.se
Three inputs to NEA
Functional
gene set
e.g. pathway
?
Network enrichment analysis
Experimentally
defined
(altered)
genes
0 50 100 150 200 250 300 350 400 450
No. of positives, random groups
0
200
400
600
800
1000
1200
1400
No
. o
f p
osi
tives,
rea
l gro
ups
GEA_SIGN
GEA_REST
NEA_SIGN
NEA_REST
Gene set enrichment analysis
Alexeyenko A., Lee W., … Pawitan Y. Network enrichment analysis:
extension of gene-set enrichment analysis to gene networks.
BMC Bioinformatics, 2012
1) Your gene/protein set: mutations, rare variants,
candidate disease genes, differentially
expressed genes.
2) A collection of previously known functional sets:
pathways, Gene Ontology terms etc.
3) A global network of interactions and relations
between genes and proteins (known and/or
predicted)
Comparison to state-of-the-art method
P53 signaling
Cell cycle
Apoptosis
0 2 4 6 8
0.0
0.2
0.4
0.6
0.8
1.0
Years since surgery
Re
lap
se
-fre
e s
urv
iva
l
Predictors:
GO:0001666 response to hypoxia
GO:0005154 EGFR binding
GO:0005164 TNF binding
GO:0070848 response to growth factor stimulus
Example applications
a) Distinguishing
between driver
and passenger
cancer mutations
Red: drivers
Black: passengers
b) Identifying driver copy number changes in cancer
confidence
copy number distribution in the cohort
c) Network enrichment scores as predictors of clinical features When used as biomarkers, the scores are often superior
to gene expression and mutation profiles
Network enrichment analysis: the concept
The overall goal of our tools for network enrichment analysis
(NEA) is functional exploration of experimental data in either
hypothesis-driven or hypothesis-free manner. Similarly to any
biological observation, statistical significance of network
patterns should always be estimated. This is done by
comparing the network connectivity score either between or
within gene sets to that expected by chance.
The method extends the gene set enrichment analysis
(GSEA) into the network domain. The output and
interpretation of NEA are similar to those of GSEA, but the
sensitivity of NEA is higher ~5-10-fold. In addition, NEA is
applicable to both gene sets and single genes, including
those lacking any GO and pathway annotations.
NEA can also accommodate any custom/proprietary network
Alexeyenko A. and Sonnhammer E. Global networks of functional
coupling in eukaryotes by
comprehensive data integration
Genome Research, 2009