Regulatory and signaling network assembly through linked open data
1. Regulatory and signaling network
assembly through Linked Open Data
Marie Lefebvre1
, Jérémie Bourdon2
, Carito Guziolowski2
, Alban Gaignard1
1
Nantes Academic Hospital, CHU de Nantes, France
2
LS2N - UMR 6004, University of Nantes, Ecole Centrale de Nantes, France
DEMO - JOBIM 2017 - LILLE
3. Marie Lefebvre - JOBIM 2017
CONTEXT
3
COMPUTATIONAL
PREDICTIONS
Signaling
networkGene regulatory network
Directed graphs
Precise
semantics
LOGIC AND
PROBABILISTIC
MODELS
bayesian,
boolean,
constraints based,
etc.
4. Marie Lefebvre - JOBIM 2017
CONTEXT
4
Signaling
networkGene regulatory network
Directed graphs
Precise
semantics
COMPUTATIONAL
PREDICTIONS
LOGIC AND
PROBABILISTIC
MODELS
bayesian,
boolean,
constraints based,
etc.
5. Marie Lefebvre - JOBIM 2017 5
CHALLENGE : NETWORK RECONSTRUCTION
6
3
22
1 - Multiple data sources
KEGG Pathways
✓ 514 pathways
Reactome
✓ 2132 pathways
✓ 10754 reactions
● In 2014, more than 1552 databases that are publicly accessible online (1)
● More than 31 databases useful for network construction(2)
(H. sapiens)
(1) Fernandez-Suarez X.M., and al. The 2014 Nucleic Acids Research Database Issue and an updated NAR online Molecular Biology Database
Collection. Nucleic Acids Res. 2014;42:D1–D6
(2) i.e. StringDB, intAct, GeneMania, HipathDB, WikiPathway, MetaCyc, ConsensusPathDB, etc.
6. Marie Lefebvre - JOBIM 2017 6
CHALLENGE : NETWORK RECONSTRUCTION
Different paths between TRAIL and FADD (apoptosis)
2 - Semantic heterogeneity
KEGG
Reactome
7. Marie Lefebvre - JOBIM 2017 7
CHALLENGE : NETWORK RECONSTRUCTION
Different paths between TRAIL and FADD (apoptosis)
2 - Semantic heterogeneity
KEGG
Reactome
8. Marie Lefebvre - JOBIM 2017
Time for 200
biological
entities input
Logic oriented
Upstream
interactions
PC Viz no result
CyPath 2 < 1min
Reactome online NA
Reactome
Cytoscape
< 1min
ChiBE < 1min
8
CHALLENGE : NETWORK RECONSTRUCTION
3 - State of the art approaches
How to discover the upstream mechanisms that trigger a list of biological entities ?
9. Marie Lefebvre - JOBIM 2017
OBJECTIVES
9
❖ Reconstruct a network from diverse biological entities
❖ Integrate diverse upstream regulators
❖ Large scale simulation-oriented networks
(→ predictive models)
10. Marie Lefebvre - JOBIM 2017 10
APPROACH
Linked Open Data
Identify &
Link Data
Querying
Reasoning
Vocabularies
Ontologies
Information
Knowledge
12. Marie Lefebvre - JOBIM 2017 12
IMPLEMENTATION
Database
● Store and disseminate knowledge about
biological pathways from 22 data
sources
● BioPAX level 3
● Freely available
Graph
SPARQL
endpoint
SPARQL
query
List of biological entities
Workflow
request
results
21. Marie Lefebvre - JOBIM 2017 21
CONCLUSION
BRAvo : Biological netwoRk Assembly
❖ Reconstruct gene regulatory network from diverse biological entities
■ From 210 entities → Time < 2min | 335 nodes & 4191 edges
❖ Integration of heterogeneous data from different data sources
❖ Designed for assembling models
❖ Available at https://github.com/symetric-group/bionets-demo
GUI Interactive reconstruction Full reconstruction BATCH
22. Marie Lefebvre - JOBIM 2017 22
FUTURE WORKS
● Extend type of biological entities : complex, molecule
● Add type of interaction : catalysis, phosphorylation
● Add filters (receptor)
Signaling network assembly
Integrate drug information
● Integration of KEGG drug / DrugBank
Simulation oriented controlled vocabulary
● Apply to several modeling frameworks
Package
● Provide JAVA and Python API