metabolic networks - hu-berlin.de · – systems biology & modelling – metabolism &...
TRANSCRIPT
![Page 1: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/1.jpg)
Metabolic Networks
Ulf Leser and Michael Weidlich
![Page 2: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/2.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 2
This Lecture
• Introduction
– Systems biology & modelling
– Metabolism & metabolic networks
• Network reconstruction
– Strategy & workflow
• Mathematical representation
– The stoichiometric matrix
– Convex analysis & solution space
– Elementary Modes & Extreme Pathways
• Constraint based network analysis
– Flux Balance Analysis & Optimization
– Tools
![Page 3: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/3.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 3
Motivation
Complex systems in biology
Metabolic network Neural network
![Page 4: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/4.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 4
„More than the sum of its parts“
• Many biological systems have emergent properties(they are called irreducible since they cannot be fullycomprehended when broken down into smaller segments)
• Studying a neuron will not explain how information isstored in the brain? (fire patterns)
![Page 5: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/5.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 5
Systems biology
"Systems biology...is about putting together rather than taking apart, integration rather than reduction. It requires that we develop ways of thinking about integration that are as rigorous as our reductionist programmes, but different....It means changing our philosophy, in the full sense of the term” - Dennis Noble
Protein structure Metabolic networks
![Page 6: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/6.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 6
Systems biology - Bridging the scales
From component to systems analysis
Whole bodyOrganTissueCell
20th century
21th century
10-6 s 10-3 s 1s 103 s 106 s 109 s
Ion channelgating
Diffusion Motility Mitosis Protein turnover
Human lifetime
![Page 7: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/7.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 7
Excursion – Virtual Liver Network
• Interdisciplinary competence network of experimental andtheoretical research groups (since 2004)
• Liver - A most relevant research object for applications in medicine, pharma research and nutrition
• Example for a german systems biology initiative
![Page 8: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/8.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 8
What is a Model ?
vs vs
A Model is a simplified image of reality.
– Models are able to reduce reality to certain aspects.
– Definition of these aspects is guided by our comprehension of whichaspects are crucial (for functionality)
– Simplification leads to understanding of complex systems
– Models should generate testable predictions (purpose driven)
![Page 9: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/9.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 9
Process of modelling
• The direct problem consists in generating data frommodels (i.e. doing simulations) relativly easy
• The inverse problem consists in generating a model fromdata generally difficult
• Modelling is an iterative process of both problems
Simulation
ExperimentModel
![Page 10: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/10.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 10
Process of modelling
• The direct problem consists in generating data frommodels (i.e. doing simulations) relativly easy
• The inverse problem consists in generating a model fromdata generally difficult
• Modelling is an iterative process of both problems
Simulation
ExperimentModel
![Page 11: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/11.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 11
Modelling - futher specified
Okay i got it! What are we going to model ?
Metabolism
![Page 12: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/12.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 12
Metabolism
• is the set of chemical reactions that happen in living organisms to maintain life, simply spoken it is the uptake, transport and conversion of chemical entities in an organism
• These processes allow for growth and reproduction, maintenance of structures and responsiveness to environmental stress.
substrates products
A B + C
![Page 13: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/13.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 13
Metabolism and Disease
• Metabolism is in a balanced state which it tries to maintain
• any permanent perturbation of this balance can lead todeseases (i.e. diabetes, cancer, etc.)
substrates products
![Page 14: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/14.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 14
Metabolism and Evolution
• Metabolism is subjected to evolutionary optimization whichleads to robustness and modularity
• Redundancy - there is often more than one solution to a certain objective
• Pathways - TCA and glycolysis are shared among manydifferent organisms
![Page 15: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/15.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 15
Key-players in metabolism
Chemical Compounds / Metabolites
• small molecules that are imported/exported and/or synthesized/degraded inside an organism.
Biochemical Reactions
• produce a set of metabolites (called products) from another set of compounds (called substrates)
• can be reversible or irreversible
• some are spontaneous, but most are enzymatic
• a set of reactions is often referred to as a pathway (still there is no consensus on the boundaries)
![Page 16: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/16.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 16
Key-players in metabolism
Enzyme
• a protein or a protein complex coded by one or several genes
• a single enzyme may accept distinct substrates and may catalyze several reactions
• a single reaction may be catalyzed by several enzymes
• the links between genes, proteins, and reactions (called GPR relationship) is nontrivial
Cofactors
• small molecules
• essential to allow the catalysis of a reaction by an enzyme
![Page 17: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/17.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 17
Example - tricarboxylic acid cycle (TCA)
![Page 18: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/18.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 18
Components of a metabolic network
![Page 19: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/19.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 19
Metabolic networks
Metabolic networks can be represented as graphs
Possibility A
Definition of a graph: G = (V,E)
• V is the set of nodes (metabolites)
• E is the set of edges (reactions)
directed graph:
• Directed edge e = (v1 ,v2) with
v1 (start node) und v2 (end node) of e.
A + B C
C D
![Page 20: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/20.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 20
Metabolic networks
Metabolic networks can be represented as graphs
Possibility B
Bipartite Graph G = (U,V,E)
• U is one type of nodes (metabolites)
• V is the other type of nodes (reactions)
• E is the set of edges with e = (u,v) or e=(v,u)
• directed bipartite graphs are very common(not shown here)
Which graph to choose depends on the type of questions asked.
A + B C
C D
![Page 21: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/21.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 21
Graph analysis
• Topological network properties like clustering, centrality, degree, density, distance and neighborhood etc. can becalculated
• But as these measures have already been covered in thelecture about protein interaction networks they will beskipped in favor of flux balance analysis which yields moreinteresting results for metabolic networks
![Page 22: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/22.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 22
This Lecture
• Introduction
– Systems biology & modelling
– Metabolism & metabolic networks
• Network reconstruction
– Strategy & workflow
• Mathematical representation
– The stoichiometric matrix
– Convex analysis & solution space
– Elementary Modes & Extreme Pathways
• Constraint based network analysis
– Flux Balance Analysis & Optimization
– Tools
![Page 23: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/23.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 23
(Genome Scale) Network Reconstruction
Three stages
• Use gene annotationfor enzyme activity
• Link enzyme activitywith reactions
• Compile thereactionlist
![Page 24: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/24.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 24
Genome Annotation
• In silico annotation methods typically lead to a functionalassignment of 40-70% of identified ORFs on a freshlysequenced microbial genome
• Automatic network reconstruction from genome annotationhas been attempted but needs intensive manualverification of the components and links of a network toproduce high quality reconstructions.
![Page 25: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/25.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 25
(Genome Scale) Network Reconstruction
Three stages
• Use gene annotationfor enzyme activity
• Link enzyme activitywith reactions
• Compile thereactionlist
![Page 26: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/26.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 26
GPR association
• Not all genes have a one-to-onerelationship with their correspondingenzymes or metabolic reactions
– Promiscous enzymes catalyzing morethan one reaction
– enzyme complexes consisting of different proteinsubunits which all need to be present for thereaction to occur
– Iso enzymes /Isozymes catalyzing the same reaction
![Page 27: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/27.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 27
(Genome Scale) Network Reconstruction
Three stages
• Use gene annotationfor enzyme activity
• Link enzyme activitywith reactions
• Compile thereactionlist
![Page 28: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/28.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 28
Compiling the reaction list
Based on Evidence
• Biochemistry : direct enzyme assay
• Genomics: functional assignments to ORFs
• Physiology: indirect information
• In silico modelling: inferred reactions
• Make extensive use of existing knowledge (databases)
All in all lots of manual refinement & curation
Betterevidence
![Page 29: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/29.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 29
Databases
• GenomesENSEMBL
• PathwayKEGG, REACTOME, METACYC
• EnzymeBRENDA
• CompoundsKEGG, CHEBI, PUBCHEM
• Reconstruction Tools PATIKA, PathwayTools
![Page 30: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/30.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 30
Database Screenshots
![Page 31: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/31.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 31
This Lecture
• Introduction
– Systems biology & modelling
– Metabolism & metabolic networks
• Network reconstruction
– Strategy & workflow
• Mathematical representation
– The stoichiometric matrix
– Convex analysis & solution space
– Elementary Modes & Extreme Pathways
• Constraint based network analysis
– Flux Balance Analysis & Optimization
– Tools
![Page 32: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/32.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 32
Stoichiometry
• The set of chemical reactions that comprise a network can be represented as a set of chemical equations
• Embedded in these chemical equations is information about reaction stoichiometry (the quantitative relationships of the reaction's reactants and products)
• Stoichiometry is invariant between organisms for the same reactions and does not change with pressure, temperature, or other conditions.
• All this stoichiometric information can be represented in a matrix form; the stoichiometric matrix, denoted by S.
![Page 33: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/33.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 33
Stoichiometric matrix
m rows m metabolites
n columns n reactions
Sij number of molecules of the ith metabolite produced in jth reaction.
• Consumption is understood as negative production.
![Page 34: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/34.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 34
Example 1 - Reversible Conversion
![Page 35: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/35.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 35
Example 2 - Bimolecular association
![Page 36: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/36.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 36
Example 3 - Cofactor-Coupled Reaction
![Page 37: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/37.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 37
Mathematical Interpretation
Mathematically, the stoichiometric matrix S is a linear transformation of the flux vector
to a vector of time derivatives of the concentration
or
![Page 38: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/38.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 38
Example 4 – system of 5 metabolites
![Page 39: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/39.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 39
Example 4 – viewed from a row
![Page 40: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/40.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 40
First constraint
• If we think about metabolism, we remember that it tendsto maintain a balanced state.
• So a good assumption might be that under normal environmental conditions a constant flow through thesystem would be optimal.
• This implies that the relative concentration of metaboliteswould remain constant over time, which in fact is thecentral constraint for flux balance analysis
substrates products
![Page 41: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/41.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 41
Steady state
Balance
• A state in which no macroscopic events take place
• For every reaction the forward and reverse directioncancel each other out
• Analogy: „calm lake“
Steady state
• A state in which every variable is temporally constant
• For every substance generating and consuming reactionscancel each other out such a system can „live“
• Analogy: „calm river“
![Page 42: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/42.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 42
Convex analysis
• Equalities (in this case Sv = 0) and inequalities (in this case, 0 ≤ vi ≤ vi,max) lead to convex analysis.
• It leads to the definition of a set of nonnegative generating vectors that span the solution space, the so calledExtreme Pathways.
• All possible flux distributions of a metabolic network liewithin the cone circumscribed by these pathways.
![Page 43: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/43.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 43
Elementary Flux Modes
Three Conditions for SEFM
1. Steady state
2. Thermodynamic feasibility ei ≥ 0 if reaction i irreversible
3. Non-decomposability no nontrivial vector v fulfilling thetwo conditions above and such that
Where P(v) denotes the set of reaction with non-zero fluxin the mode v.
![Page 44: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/44.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 44
Extreme Pathways
Two additional Conditions for SEP
4. Network reconfiguration: reactions must be classified either as exchange or internal. All reversible internal reactions are splitted into two separate, irreversible reactions with opposite direction. Internal flux can only admit non-negative value. Exchange reactions can be reversible.
5. Systemic independance: the set of EPs in a network is the minimal set of EFMs such that all feasible steady-state fluxes are non-negative linear combinations of these extremal pathways.
![Page 45: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/45.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 45
Extreme Pathways (Definition)
Papin 2004 Comparison of network-based pathway analysis methods - PMID: 15283984
![Page 46: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/46.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 46
EP vs EFM
• EP is the minimal set of EFM in the sense that no element in EP is the convex combination of other elements.
• EP = EFM in the case where all exchange reactions (reactions connecting a metabolite with the outside of a metabolic system) are irreversible (e.g. glucose can only go into and ammonia can only go out of the system).
![Page 47: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/47.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 47
Examples (EP & EFM)
![Page 48: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/48.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 48
How to compute Extreme Pathways (1)
![Page 49: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/49.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 49
How to compute Extreme Pathways (2)
![Page 50: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/50.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 50
How to compute Extreme Pathways (3)
![Page 51: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/51.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 51
How to compute Extreme Pathways (4)
![Page 52: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/52.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 52
How to compute Extreme Pathways (5)
![Page 53: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/53.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 53
How to compute Extreme Pathways (6)
![Page 54: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/54.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 54
How to compute Extreme Pathways (7)
• Now we balanced all the species (net change is zero) and the fluxcombination recorded on the left panel corresponds to the extremalpathways
![Page 55: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/55.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 55
Extreme Pathway Classification
Biologically infeasible
![Page 56: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/56.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 56
This Lecture
• Introduction
– Systems biology & modelling
– Metabolism & metabolic networks
• Network reconstruction
– Strategy & workflow
• Mathematical representation
– The stoichiometric matrix
– Convex analysis & solution space
– Elementary Modes & Extreme Pathways
• Constraint based network analysis
– Flux Balance Analysis & Optimization
– Tools
![Page 57: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/57.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 57
Optimization based methods (FBA)
• Identify an objective function Z to be minimized or maximized depending on the task at hand
• Set upper and lower bounds on the flux variables vi,min ≤ vi ≤ vi,max for i={1,…,n}
• Solve Linear Programming Problem (i.e. Simplex Algorithm)
![Page 58: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/58.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 58
Linear Programming
![Page 59: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/59.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 59
Possible Solutions for LP (visual)
vi,min ≤ vi ≤ vi,max unique degeneratedempty
![Page 60: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/60.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 61
Minimization - Objective Functions
• Minimize ATP production : most energy-efficient state
• Minimize nutrient uptake : the fittest state under nutrient-shortage
• Minimize the Manhattan distance or Euclidian distance ofthe flux vector : minimize the overall flux
![Page 61: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/61.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 62
Maximization - Objective Functions
• Maximize metabolite production : the biochemicalproduction capabilities of certain desirable metabolite such as lysine, phenylalanine etc.
• Maximize biomass formation : maximal growth rate
• Maximize biomass as well es metabolite production : thetrade-off between cell-growth and forced metaboliteproduction
![Page 62: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/62.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 63
Additional constraints restraining the solution space
![Page 63: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/63.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 64
Summary Flux Balance Analysis
![Page 64: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/64.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 65
Tools for FBA
![Page 65: Metabolic Networks - hu-berlin.de · – Systems biology & modelling – Metabolism & metabolic networks • Network reconstruction – Strategy & workflow • Mathematical representation](https://reader035.vdocument.in/reader035/viewer/2022062603/5f095b437e708231d426724e/html5/thumbnails/65.jpg)
Ulf Leser and Michael Weidlich: Bioinformatics, Sommersemester 2011 66
Further Reading
• Llaneras 2008 Stoichiometric Modelling of Cell MetabolismPMID: 18295713
• Papin 2004 Comparison of network-based pathway analysismethods PMID: 15283984
• Harald Marx 2007 „Ausarbeitung Extreme Pathways“
• More kinetic analysis in cellular systems visit the lectures (Systemsbiology, Modelle zellulärer Prozesse) of Prof. Dr. Edda Klipp (Head of Theoretical Biophysics)