Modelling of biochemical systems:  Day 2 Vangelis Simeonidis Manchester Centre for Integrative Systems Biology
Case study 1: Yeast Glycolysis Silicon Cell website http://jjj.biochem.sun.ac.za/database/index.html Teusink et al. glycol...
Flux Balance Analysis  <ul><li>Key idea: look for steady state flux patterns that optimise a given objective function </li...
Mechanistic (kinetic) Constraint-based   (stoichiometric) Find an exact solution Find a range of allowable solutions
[c] : 13BDglcn + h2o --> glc-D [e] : 13BDglcn + h2o --> glc-D [c] : udpg --> 13BDglcn + h + udp [c] : udpg --> 16BDglcn + ...
genome-scale networks
Stoichiometric matrix (~1700x1300)
Flux Balance Analysis (FBA) Stoichiometric Matrix : signifies if and how a metabolite takes part in a certain reaction A B...
Flux Balance Analysis (FBA)    easy to solve     only stoichiometry required    no  substrate concentrations Stoichiome...
Biomass “reaction” A+B+………Z    biomass v b
A E B C D G H I K L M N O F L 1  ≤ v 1  ≤ U 1 L 2  ≤ v 2  ≤ U 2 … .............. L n  ≤ v n  ≤ U n max M S  .  v = 0 Chasi...
1 1 1 1 1 1 1 1 X Y 1 1 1 How does FBA work? A E B C D G H I K L M 1
1 1 1 1 X Y 1 1 1 A E B C D L N 1 XX YY ZZ WW UU How does FBA work?
What pathways?  GLC DHAP G6P F6P FDP G3P 13PG 3PG 2PG PYR PEP ACALD CO2 ETOH AKG 3PHP PSEP GLU SER GLY CO2 GLYC3P GLYC OAA...
Some of the problems with FBA    no  substrate concentrations    not  always realistic    solution degeneracy
FBA and metabolite concentrations linlog (with correct elasticities): Teusink:  o linlog (with estimated elasticities): <u...
In general an FBA problem can have more than one optimal solution.  FBA and solution degeneracy
FBA and unrealistic solutions
Discussion: Can a biologist fix a radio?
Discussion: Can a biologist fix a radio?
Modelling of Signalling <ul><li>Biochemists use pictorial models </li></ul><ul><ul><li>Useful for understanding interactio...
NF-  B Signaling Pathway TLR LPS IL-1R Tollip IL IL-1RAcP IRAK1/2  MyD88 DD DD TNFR1 TNF- α TRADD  TRAF6  ECSIT TAK1 TAB ...
Computational Modeling of the NF-  B Pathway Ihekwaba et al 2004 26 signaling species 64 reactions (uindirectional) Oscil...
Mammalian MAPK Pathways Stimulus MAPKKK MAPKK MAPK Response Raf, Tpl2 ERK1 /2 MKK1 /2 Growth Factors Proliferation Differe...
Generic MAP Kinase Pathway Characteristics <ul><li>3 kinase cascade conserved from yeast to humans </li></ul>Substrates Ex...
The p38 MAP Kinase Signalling Pathway TLRs ILRs TNFR RAC CDC42 ASK1 TAK1 MEKK4 MLK2 MYD88 TOLLIP IRAK TRAF6 TAB1 TAB2 TRAF...
Modelling Approach <ul><li>Chemical equations governing interactions & transformations… </li></ul><ul><ul><li>Association/...
p38 MAP Kinase Model <ul><li>Starts with TLR7 activation </li></ul><ul><li>Ends with activated p38  </li></ul><ul><li>60 ...
Regulation of metabolic pathways mRNA genes enzymes Regulating metabolite Systems thinking
SBML: Systems Biology Markup Language <ul><li>computer-readable format for representing models  of biological processes </...
Computational systems biology Computational model Experimental data Analysis Better model New experiments Improved underst...
Computational modelling strategy Biological model Biochemical model Kinetic model Mathematical model “… For example, in th...
Computational modelling strategy Biological model Biochemical model Kinetic model Mathematical model “… For example, in th...
Computational modelling strategy Biological model Biochemical model Kinetic model Mathematical model Difficult Easier Auto...
Computational modelling strategy Biological model Biochemical model Kinetic model Mathematical model Difficult Easier Auto...
Key points about building models <ul><li>Don’t be shy of making assumptions & estimates – they are vital for ‘first pass’ ...
Generalised kinetics <ul><li>Use stoichiometric structure of the biochemical network </li></ul><ul><ul><li>n substrates, m...
Estimates of kinetic parameters <ul><li>You can always use typical values for </li></ul><ul><ul><li>protein association ra...
 
Modelling of biochemical systems THANK YOY FOR LISTENING!
Upcoming SlideShare
Loading in …5
×

Modelling of biochemical networks - Day 2

701
-1

Published on

Slides for MRes and DTC course

Published in: Education, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
701
On Slideshare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
22
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide
  • Explaining how FBA works, and what the challenges are
  • Explaining how FBA works, and what the challenges are
  • Explaining how FBA works, and what the challenges are
  • The graph of No of enzymes relative to Carbon flux
  • Modelling of biochemical networks - Day 2

    1. 1. Modelling of biochemical systems: Day 2 Vangelis Simeonidis Manchester Centre for Integrative Systems Biology
    2. 2. Case study 1: Yeast Glycolysis Silicon Cell website http://jjj.biochem.sun.ac.za/database/index.html Teusink et al. glycolysis model (Eur J Biochem 267:5313, 2000) <ul><li>Find steady state flux through: </li></ul><ul><ul><li>Glucose Transporter (GLT) </li></ul></ul><ul><ul><li>Glycogen (GLYCO) </li></ul></ul><ul><ul><li>Trehalose (Treha) </li></ul></ul><ul><ul><li>Phosphofructokinase (PFK) </li></ul></ul><ul><li>How long does it take system to reach steady state? </li></ul><ul><li>What is the effect of decreasing extracellular glucose level from 50 mM to 2 mM on flux through ADH (flux of ethanol)? </li></ul>
    3. 3. Flux Balance Analysis <ul><li>Key idea: look for steady state flux patterns that optimise a given objective function </li></ul><ul><ul><li>biomass production </li></ul></ul><ul><ul><li>product yield in metabolically engineered cells </li></ul></ul><ul><li>Use stoichiometric matrix only – flux patterns must satisfy </li></ul><ul><li>Ignore kinetics – just have max/min bounds on fluxes </li></ul><ul><li>Find balancing fluxes that maximise flux of product or biomass </li></ul>
    4. 4. Mechanistic (kinetic) Constraint-based (stoichiometric) Find an exact solution Find a range of allowable solutions
    5. 5. [c] : 13BDglcn + h2o --> glc-D [e] : 13BDglcn + h2o --> glc-D [c] : udpg --> 13BDglcn + h + udp [c] : udpg --> 16BDglcn + h + udp [c] : 23camp + h + h2o --> amp2p 2dda7p[c] <==> 2dda7p[m] 2dhp[c] <==> 2dhp[m] [c] : 2doxg6p + h2o --> 2dglc + pi [m] : 2hpmhmbq + amet --> ahcys + h + q6 [m] : 2hp6mp + o2 --> 2hp6mbq + h2o [m] : 2hp6mbq + amet --> 2hpmmbq + ahcys + h [m] : 2hpmmbq + (0.5) o2 --> 2hpmhmbq 2mbac[c] --> 2mbac[e] 2mbald[c] <==> 2mbald[e] 2mbald[c] <==> 2mbald[m] 2mbtoh[c] <==> 2mbtoh[e] 2mbtoh[c] <==> 2mbtoh[m] 2mppal[c] <==> 2mppal[e] 2mppal[c] <==> 2mppal[m] 2oxoadp[m] --> 2oxoadp[c] 2phetoh[e] <==> 2phetoh[c] 2phetoh[m] <==> 2phetoh[c] [m] : 34hpp + h + nadh --> 34hpl + nad [c] : 34hpp + o2 --> co2 + hgentis …………………………………………………… ………………………………………………… ... …………………………………………………… 34hpp[c] + h[c] <==> 34hpp[m] + h[m] 34hpp[c] + h[c] <==> 34hpp[x] + h[x] 3c3hmp[c] <==> 3c3hmp[m] 3c4mop[c] <==> 3c4mop[m] [m] : 3dh5hpb + amet --> 3hph5mb + ahcys + h 3dh5hpb[c] <==> 3dh5hpb[m] [c] : 3dsphgn + h + nadph --> nadp + sphgn [c] : 3hanthrn + o2 --> cmusa + h [m] : 3hph5mb --> 2hp6mp + co2 [c] : 3c2hmp + amet --> 3ipmmest + ahcys 3mbald[c] <==> 3mbald[e] 3mbald[c] <==> 3mbald[m] [c] : 3mob + h --> 2mppal + co2 3mob[c] <==> 3mob[m] [c] : 3mop + h --> 2mbald + co2 3mop[c] <==> 3mop[m] [c] : 3ophb_5 + (0.5) o2 --> 3dh5hpb 3ophb_5[c] <==> 3ophb_5[m] 4abutn[c] <==> 4abutn[m] 4abut[c] <==> 4abut[m] 4abz[c] <==> 4abz[m] 4h2oglt[c] <==> 4h2oglt[m] 4h2oglt[c] <==> 4h2oglt[x] [m] : coa + coucoa + h2o + nad --> 4hbzcoa + accoa + h + nadh ………………………………………………………… . ………………………………………………………… …………………………………………………………
    6. 6. genome-scale networks
    7. 7. Stoichiometric matrix (~1700x1300)
    8. 8. Flux Balance Analysis (FBA) Stoichiometric Matrix : signifies if and how a metabolite takes part in a certain reaction A B … G r 1 r 2 …. r n a 1 b 1 … . g 1 a 2 b 2 … . g 2 … . … . … . … . a n b n … . g n Flux Vector : Each component represents the flux through the corresponding reaction v 1 v 2 … . v n v dA/dt dB/dt … . dG/dt = Steady State condition 0 0 … . 0 = L 1 ≤ v 1 ≤ U 1 L 2 ≤ v 2 ≤ U 2 … .............. L n ≤ v n ≤ U n requires minimal biological data to make quantitative inferences about network behaviour S . v = 0
    9. 9. Flux Balance Analysis (FBA)  easy to solve  only stoichiometry required  no substrate concentrations Stoichiometric Matrix : signifies if and how a metabolite takes part in a certain reaction A B … G r 1 r 2 …. r n a 1 b 1 … . g 1 a 2 b 2 … . g 2 … . … . … . … . a n b n … . g n Flux Vector : Each component represents the flux through the corresponding reaction v 1 v 2 … . v n v dA/dt dB/dt … . dG/dt = Steady State condition 0 0 … . 0 = L 1 ≤ v 1 ≤ U 1 L 2 ≤ v 2 ≤ U 2 … .............. L n ≤ v n ≤ U n  not detailed enough S . v = 0
    10. 10. Biomass “reaction” A+B+………Z  biomass v b
    11. 11. A E B C D G H I K L M N O F L 1 ≤ v 1 ≤ U 1 L 2 ≤ v 2 ≤ U 2 … .............. L n ≤ v n ≤ U n max M S . v = 0 Chasing the flux: Flux Balance Analysis
    12. 12. 1 1 1 1 1 1 1 1 X Y 1 1 1 How does FBA work? A E B C D G H I K L M 1
    13. 13. 1 1 1 1 X Y 1 1 1 A E B C D L N 1 XX YY ZZ WW UU How does FBA work?
    14. 14. What pathways? GLC DHAP G6P F6P FDP G3P 13PG 3PG 2PG PYR PEP ACALD CO2 ETOH AKG 3PHP PSEP GLU SER GLY CO2 GLYC3P GLYC OAA ASP G1P UDPG 13BDGLCN AC MAN6P MAN1P GDPMANN DOLMANP MANNAN 14GLUN GLYCOGEN
    15. 15. Some of the problems with FBA  no substrate concentrations  not always realistic  solution degeneracy
    16. 16. FBA and metabolite concentrations linlog (with correct elasticities): Teusink: o linlog (with estimated elasticities): <ul><li>Good fit in most cases </li></ul><ul><li>Can easily incorporate experimental information to improve the fit </li></ul>
    17. 17. In general an FBA problem can have more than one optimal solution. FBA and solution degeneracy
    18. 18. FBA and unrealistic solutions
    19. 19. Discussion: Can a biologist fix a radio?
    20. 20. Discussion: Can a biologist fix a radio?
    21. 21. Modelling of Signalling <ul><li>Biochemists use pictorial models </li></ul><ul><ul><li>Useful for understanding interactions </li></ul></ul><ul><ul><li>Limited to very simple predictions </li></ul></ul><ul><li>System biologists use computational model (ODEs) </li></ul><ul><ul><li>Quantitative </li></ul></ul><ul><ul><li>Predicts shape of the response profile </li></ul></ul><ul><ul><li>Can make detailed predictions </li></ul></ul><ul><ul><li>Can help design more sophisticated therapies </li></ul></ul>Transient ERK activity t ERK t ERK Sustained ERK activity Differentiation Proliferation
    22. 22. NF-  B Signaling Pathway TLR LPS IL-1R Tollip IL IL-1RAcP IRAK1/2 MyD88 DD DD TNFR1 TNF- α TRADD TRAF6 ECSIT TAK1 TAB MEKK1 NIK/? IKK γ IKK Complex IKK α IKK β p65 p50 p65 p50 p65 p50 p65 p50 P P p65 p50 P P ub ub ub NF- κB/ IκB Complex I κB Phosphorylation Ubiquitination 26S Proteosome I κB Degradation Cell membrane I κB Nucleus Nuclear translocation RIP TRAF2 IkB translation IkB transcription
    23. 23. Computational Modeling of the NF-  B Pathway Ihekwaba et al 2004 26 signaling species 64 reactions (uindirectional) Oscillations of nuclear NF-kB levels
    24. 24. Mammalian MAPK Pathways Stimulus MAPKKK MAPKK MAPK Response Raf, Tpl2 ERK1 /2 MKK1 /2 Growth Factors Proliferation Differentiation Apoptosis MKK4 /7 Tpl2, MEKK, MLK, TAK, ASK JNK p38 MKK3/6 Inflammatory Cytokines, Stress Inflammation, Apoptosis Development, Proliferation
    25. 25. Generic MAP Kinase Pathway Characteristics <ul><li>3 kinase cascade conserved from yeast to humans </li></ul>Substrates Extracellular Stimuli MAP Kinase Kinase Kinases MAP Kinase Kinases MAP Kinases Transcription factor Gene T-x-Y +P +P +P -P -P P P P P P P P -P
    26. 26. The p38 MAP Kinase Signalling Pathway TLRs ILRs TNFR RAC CDC42 ASK1 TAK1 MEKK4 MLK2 MYD88 TOLLIP IRAK TRAF6 TAB1 TAB2 TRAF2 MKK4 MKK3 MKK6 P38  P38  P38  P38  MKP-1 MKP-5 PP2C  PP2C  MAPKAP-K2 MAPKAP-K3 MSK1 MSK2 ELK1 CHOP ATF1 ATF2 HSP27 CREB MNK1 SRF 4EBP1 Cytoplasm Nucleus Heat shock, H 2 O 2 Stress down regulation MAPKKKs MAPKKs MAPKs Extra-cellular GADD45 PAC1 PP2C  LPS, ssRNA TLRs TAK1 TRAF6 TAB1 TAB2 MKK3 MKK6 P38  MKP-1 MKP-5 PP2Ce PP2Cb PP2Ca
    27. 27. Modelling Approach <ul><li>Chemical equations governing interactions & transformations… </li></ul><ul><ul><li>Association/Dissociation: </li></ul></ul><ul><ul><ul><ul><ul><li>TAK1(P) + MKK3   TAK1(P)-MKK3 </li></ul></ul></ul></ul></ul><ul><ul><li>Catalysis (phosphorylation or dephosphorylation): </li></ul></ul><ul><ul><ul><ul><ul><li>TAK1(P)-MKK3 ’ TAK1(P) + MKK3(P) </li></ul></ul></ul></ul></ul><ul><ul><ul><ul><ul><li>PP2C  -MKK3(P) ’ PP2C  + MKK3 </li></ul></ul></ul></ul></ul><ul><li>… are converted to ordinary differential equations (ODEs): </li></ul><ul><ul><li>d/dt[MKK3] = -k 1 [TAK1(P)][MKK3] + k -1 [TAK1(P)-MKK3] </li></ul></ul><ul><ul><li>+ k 3 [PP2C  -MKK3(P)] </li></ul></ul><ul><li>assume mass-action kinetics </li></ul>
    28. 28. p38 MAP Kinase Model <ul><li>Starts with TLR7 activation </li></ul><ul><li>Ends with activated p38  </li></ul><ul><li>60 species, 90 reactions </li></ul><ul><li>Developed in Sentero (in-house network analysis tool) </li></ul><ul><li>Interactive tool – SBML Compliant, interfaces with MATLAB for solution </li></ul>Activation of TLR7 and formation of TAK1 Complex Activation of TAK1 Activation of MKK3 & MKK6 Translocation to Cytosol Negative feedback Activation of p38 
    29. 29. Regulation of metabolic pathways mRNA genes enzymes Regulating metabolite Systems thinking
    30. 30. SBML: Systems Biology Markup Language <ul><li>computer-readable format for representing models of biological processes </li></ul><ul><li>applicable to simulations of metabolism, cell-signaling, and many other topics </li></ul><ul><li>evolving since </li></ul><ul><li>Close to 200 software supporting it </li></ul><ul><li>website: http://www.sbml.org </li></ul><ul><li>Model database: http://biomodels.net </li></ul>
    31. 31. Computational systems biology Computational model Experimental data Analysis Better model New experiments Improved understanding Current understanding Assumptions Approximations Estimates
    32. 32. Computational modelling strategy Biological model Biochemical model Kinetic model Mathematical model “… For example, in the current model, ternary complex binds to the 40S subunit prior to binding of the mRNA. The assignment of this order of binding does not appear to be strong, however... A priori there does not seem to be any reason why unwinding of the mRNA’s 5&quot;-UTR and loading of the message onto the 40S subunit by the factors could not take place, at least some of the time, prior to binding of the ternary complex….” Difficult Easier Automatic Stoichiometry of translation initiation Rate equations Concentrations Molecule No./ cell/1e5 Reference eIF2B 0.3 von der Haar 11 eIF2GDP 1.8 von der Haar 11 tRNA 20 Estimate eIF1 2.5 von der Haar 11 eIF3 1 von der Haar 11 eIF5 0.483 Ghaemmaghami 12 eIF1A 0.5 von der Haar 11 40S 2 French 13 eIF4E 3.4 von der Haar 11 eIF4G 0.175 von der Haar 11 mRNA 0.15 Ambro Van Hoof 14 PabP 1.98 Ghaemmaghami 12 eIF4A 8 von der Haar 11 eIF4B 1.55 von der Haar 11 eIF5B 0.134 Ghaemmaghami 12 60S 2 French 13
    33. 33. Computational modelling strategy Biological model Biochemical model Kinetic model Mathematical model “… For example, in the current model, ternary complex binds to the 40S subunit prior to binding of the mRNA. The assignment of this order of binding does not appear to be strong, however... A priori there does not seem to be any reason why unwinding of the mRNA’s 5&quot;-UTR and loading of the message onto the 40S subunit by the factors could not take place, at least some of the time, prior to binding of the ternary complex….” Difficult Easier Automatic Stoichiometry of translation initiation
    34. 34. Computational modelling strategy Biological model Biochemical model Kinetic model Mathematical model Difficult Easier Automatic Stoichiometry of translation initiation Rate equations Concentrations Molecule No./ cell/1e5 Reference eIF2B 0.3 von der Haar 11 eIF2GDP 1.8 von der Haar 11 tRNA 20 Estimate eIF1 2.5 von der Haar 11 eIF3 1 von der Haar 11 eIF5 0.483 Ghaemmaghami 12 eIF1A 0.5 von der Haar 11 40S 2 French 13 eIF4E 3.4 von der Haar 11 eIF4G 0.175 von der Haar 11 mRNA 0.15 Ambro Van Hoof 14 PabP 1.98 Ghaemmaghami 12 eIF4A 8 von der Haar 11 eIF4B 1.55 von der Haar 11 eIF5B 0.134 Ghaemmaghami 12 60S 2 French 13
    35. 35. Computational modelling strategy Biological model Biochemical model Kinetic model Mathematical model Difficult Easier Automatic Rate equations Concentrations Molecule No./ cell/1e5 Reference eIF2B 0.3 von der Haar 11 eIF2GDP 1.8 von der Haar 11 tRNA 20 Estimate eIF1 2.5 von der Haar 11 eIF3 1 von der Haar 11 eIF5 0.483 Ghaemmaghami 12 eIF1A 0.5 von der Haar 11 40S 2 French 13 eIF4E 3.4 von der Haar 11 eIF4G 0.175 von der Haar 11 mRNA 0.15 Ambro Van Hoof 14 PabP 1.98 Ghaemmaghami 12 eIF4A 8 von der Haar 11 eIF4B 1.55 von der Haar 11 eIF5B 0.134 Ghaemmaghami 12 60S 2 French 13
    36. 36. Key points about building models <ul><li>Don’t be shy of making assumptions & estimates – they are vital for ‘first pass’ model building </li></ul><ul><li>An imperfect computational model is better than no model </li></ul><ul><li>Avoid approximations that throw away prior biochemical knowledge </li></ul><ul><li>You can always construct a kinetic (dynamic) model from a biochemical (stoichiometric) model: </li></ul><ul><ul><li>Generalised kinetics </li></ul></ul><ul><ul><li>Estimated rate constants </li></ul></ul>
    37. 37. Generalised kinetics <ul><li>Use stoichiometric structure of the biochemical network </li></ul><ul><ul><li>n substrates, m products </li></ul></ul><ul><ul><li>Assume no allosteric effects </li></ul></ul><ul><li>e.g. 2-keto-3-deoxy-d-gluconate (KDG) kinase in glucose metabolism of Sulfolobus solfataricus </li></ul><ul><ul><li>Irreversible mass action kinetics </li></ul></ul><ul><ul><li>Reversible Michaelis-Menten </li></ul></ul><ul><ul><li>other general kinetics linlog, random order, “convinience” kinetics, … </li></ul></ul>KDG KDG kinase ATP ADP KDGP
    38. 38. Estimates of kinetic parameters <ul><li>You can always use typical values for </li></ul><ul><ul><li>protein association rate constant (Schlosshauer & Baker 2004) </li></ul></ul><ul><ul><li>protein dissociation rate constant (Fekkes et al. 1995) </li></ul></ul><ul><ul><li>catalytic rate constant (e.g. phosphorylation) (Wilkinson et al. 2008) </li></ul></ul><ul><li>Number of papers with parameters & kinetics is not increasing </li></ul>
    39. 40. Modelling of biochemical systems THANK YOY FOR LISTENING!
    1. A particular slide catching your eye?

      Clipping is a handy way to collect important slides you want to go back to later.

    ×