Introduction Qualitative data Consistency Applications Conclusion Adding missing post-transcriptional regulations to a regulatory network Carito Guziolowski, Sylvain Blachon, Anne Siegel Project Symbiose - IRISA-INRIA - Rennes Journ´ ee satellite JOBIM 2008
33
Embed
Adding missing post-transcriptional regulations to a ...Introduction Qualitative data Consistency Applications Conclusion Adding missing post-transcriptional regulations to a regulatory
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Introduction Qualitative data Consistency Applications Conclusion
Adding missing post-transcriptionalregulations to a regulatory network
Introduction Qualitative data Consistency Applications Conclusion
Consistency: causal boolean rule
”The variation of one molecule in the network must be explainedby an influence received from at least one of its predecessors inthe network” (Siegel et al. 2006 Biosystems)
General rule : transcriptional regulations
A, B observed A, B observed. A, B, C observed.
A B C+ + +– – –
A B C+ – ?– + ?
A B C+ + –– – +
C predicted C NOT predicted.unfixed INCONSISTENCY
Introduction Qualitative data Consistency Applications Conclusion
Consistency: Validation and Lessons
Validated on E. coli transcriptional network (Guziolowski et al.2006 JBPC)
mRNA variation ≈ active protein variation
ProblemsInconsistencies
False positive predictions
Unfixed values (too many)
SolutionsMissing post-transcriptionalinteractions
Separate mRNA from activeproteins
Refine general rule into specificboolean rules (avoid dynamic)
All solutions include post-transcriptional reasoning
Introduction Qualitative data Consistency Applications Conclusion
Consistency: Validation and Lessons
Validated on E. coli transcriptional network (Guziolowski et al.2006 JBPC)
mRNA variation ≈ active protein variation
ProblemsInconsistencies
False positive predictions
Unfixed values (too many)
SolutionsMissing post-transcriptionalinteractions
Separate mRNA from activeproteins
Refine general rule into specificboolean rules (avoid dynamic)
All solutions include post-transcriptional reasoning
Introduction Qualitative data Consistency Applications Conclusion
Consistency: Validation and Lessons
Validated on E. coli transcriptional network (Guziolowski et al.2006 JBPC)
mRNA variation ≈ active protein variation
ProblemsInconsistencies
False positive predictions
Unfixed values (too many)
SolutionsMissing post-transcriptionalinteractions
Separate mRNA from activeproteins
Refine general rule into specificboolean rules (avoid dynamic)
All solutions include post-transcriptional reasoning
Introduction Qualitative data Consistency Applications Conclusion
Consistency: Validation and Lessons
Validated on E. coli transcriptional network (Guziolowski et al.2006 JBPC)
mRNA variation ≈ active protein variation
ProblemsInconsistencies
False positive predictions
Unfixed values (too many)
SolutionsMissing post-transcriptionalinteractions
Separate mRNA from activeproteins
Refine general rule into specificboolean rules (avoid dynamic)
All solutions include post-transcriptional reasoning
Introduction Qualitative data Consistency Applications Conclusion
Consistency: Validation and Lessons
Validated on E. coli transcriptional network (Guziolowski et al.2006 JBPC)
mRNA variation ≈ active protein variation
ProblemsInconsistencies
False positive predictions
Unfixed values (too many)
SolutionsMissing post-transcriptionalinteractions
Separate mRNA from activeproteins
Refine general rule into specificboolean rules (avoid dynamic)
All solutions include post-transcriptional reasoning
Introduction Qualitative data Consistency Applications Conclusion
Inconsistent data with networkProposed correction (Atlung et al.
1996) External signal
Consistent after adding post-transcriptional effect0,01% of network products changed as {+, –}
Introduction Qualitative data Consistency Applications Conclusion
Applications: E. coli (2)New rule added: protein-complex behaviour rule
”The weakest takes it all” (Radulescu et al. FOSBE 2007)
The concentration of a protein complex at the steady state follows theconcentration of the limiting subunit
A, B observed, Cprotein complex
A, B observed, Cprotein complex. B
limited
A B C+ – ?
A B C+ – –
C NOT predicted C predicted
Applied to the IHF protein complex in E. coli30% products of the network changed as {+, –}
Introduction Qualitative data Consistency Applications Conclusion
Applications: E. coli (2)New rule added: protein-complex behaviour rule
”The weakest takes it all” (Radulescu et al. FOSBE 2007)
The concentration of a protein complex at the steady state follows theconcentration of the limiting subunit
A, B observed, Cprotein complex
A, B observed, Cprotein complex. B
limited
A B C+ – ?
A B C+ – –
C NOT predicted C predicted
Applied to the IHF protein complex in E. coli30% products of the network changed as {+, –}
Introduction Qualitative data Consistency Applications Conclusion
Applications: E. coli (2)New rule added: protein-complex behaviour rule
”The weakest takes it all” (Radulescu et al. FOSBE 2007)
The concentration of a protein complex at the steady state follows theconcentration of the limiting subunit
A, B observed, Cprotein complex
A, B observed, Cprotein complex. B
limited
A B C+ – ?
A B C+ – –
C NOT predicted C predicted
Applied to the IHF protein complex in E. coli30% products of the network changed as {+, –}
Introduction Qualitative data Consistency Applications Conclusion
Applications: E. coli (3)
Validation of predictionsCompared to microarray outputs (Faith et al. 2007 PLoS Biology)80% of the reported changes were in agreement
And the rest ?
Expression DataihfA = +ihfB = –fic = +
Introduction Qualitative data Consistency Applications Conclusion
Applications: E. coli (3)
Validation of predictionsCompared to microarray outputs (Faith et al. 2007 PLoS Biology)80% of the reported changes were in agreement
And the rest ?
Expression DataihfA = +ihfB = –fic = +
Predictioncrp = –
ompA = –rpoD = –rpoS = +IHF = –
Introduction Qualitative data Consistency Applications Conclusion
Applications: E. coli (3)
Validation of predictionsCompared to microarray outputs (Faith et al. 2007 PLoS Biology)80% of the reported changes were in agreement
And the rest ?
Expression DataihfA = +ihfB = –fic = +
Predictioncrp = –
ompA = –rpoD = –rpoS = +IHF = –
Rsd protein forms a complex with RpoD preventing it tobind RNA-polymerase (Jishage and Ishihama 1998)active RpoD predicted not rpoD mRNA
Introduction Qualitative data Consistency Applications Conclusion
Applications: E. coli (3)
Validation of predictionsCompared to microarray outputs (Faith et al. 2007 PLoS Biology)80% of the reported changes were in agreement
And the rest ?
Expression DataihfA = +ihfB = –fic = +
Predictioncrp = –
ompA = –rpoD = –rpoS = +IHF = –
Rsd protein forms a complex with RpoD preventing it tobind RNA-polymerase (Jishage and Ishihama 1998)active RpoD predicted not rpoD mRNA
Introduction Qualitative data Consistency Applications Conclusion
Applications: E. coli (3)
Validation of predictionsCompared to microarray outputs (Faith et al. 2007 PLoS Biology)80% of the reported changes were in agreement
And the rest ?
Expression DataihfA = +ihfB = –fic = +
Predictioncrp = –
ompA = –rpoD = –rpoS = +IHF = –
Rsd protein forms a complex with RpoD preventing it tobind RNA-polymerase (Jishage and Ishihama 1998)active RpoD predicted not rpoD mRNA
Introduction Qualitative data Consistency Applications Conclusion
Ewing’s network
Include protein complex ruleMany post-transcriptional interactions → Dividing mRNAfrom active protein (Baumuratova et al. SBMC 2008)
Before After
Consequences:Distinguish the type of predicted moleculeNew way of modeling phosphorylations
Introduction Qualitative data Consistency Applications Conclusion
Ewing: Modelling phosphorylation
The RB1 exampleactive protein RB1 = hypophosphorylated RB1
Old model: New one:
Introduction Qualitative data Consistency Applications Conclusion
Ewing: Modelling competitions (1)The RB1, (E2F.) example
RB1 does not directly inhibit (E2F.)Both proteins and the cell cycle S may be active (+) at thesame moment
Old model: New one:
Introduction Qualitative data Consistency Applications Conclusion
Ewing: Modelling competitions(2)
Active RB1 wins active (E2F.) in the cell cycle S control (DeGregori 2002 BBA)
Model:
Known behaviour:(E2F.) RB1 E2F:RB1 cc S
+ + + –– + – –+ – – +– – – –
We need is to add a new qualitative boolean rule:f(E2F, RB1) −− > cell cycle S
Introduction Qualitative data Consistency Applications Conclusion
Ewing: Modelling competitions(2)
Active RB1 wins active (E2F.) in the cell cycle S control (DeGregori 2002 BBA)
Model:
Known behaviour:(E2F.) RB1 E2F:RB1 cc S
+ + + –– + – –+ – – +– – – –
We need is to add a new qualitative boolean rule:f(E2F, RB1) −− > cell cycle S
Introduction Qualitative data Consistency Applications Conclusion
Conclusion
Reasoning from a general rule allow us to retrieve missing post-transcriptionalmechanisms (E. coli)
Any given literature curated network model may be post-transcriptionallyrefined (Ewing):
Division between mRNA/active proteinTreating phosphorylation interactionsInhibitor-activator competition models mapped into logical functions
Direct results of including post-transcriptional:Better validation of experimental outputsEnlarging predictions to (mRNA, active protein, protein complex)
Boolean rules may be appropiate to describe equilibria-shifts (work in progress)
New Bioquali: including logical rules
Introduction Qualitative data Consistency Applications Conclusion
Acknowledgements:
RennesAnne SiegelSylvain BlachonMichel Le BorgneJeremy GruelOvidiu RadulescuPhilippe VeberTatiana Baumaratova