Biological Network Analysis: Metabolic Optimization Methods Tomer Shlomi Winter 2008.

Slides:



Advertisements
Similar presentations
Lets begin constructing the model… Step (I) - Definitions We begin with a very simple imaginary metabolic network represented as a directed graph: Vertex.
Advertisements

School of Computer Science, Tel-Aviv University, Tel-Aviv, Israel
Predicting essential genes via impact degree on metabolic networks ISSSB’11 Takeyuki Tamura Bioinformatics Center, Institute for Chemical Research Kyoto.
Regulation of Gene Expression in Flux Balance Models of Metabolism.
Darwinian Genomics Csaba Pal Biological Research Center Szeged, Hungary.
Effect of oxygen on the Escherichia coli ArcA and FNR regulation systems and metabolic responses Chao Wang Jan 23, 2006.
Prediction of Therapeutic microRNA based on the Human Metabolic Network Ming Wu, Christina Chan Bioinformatics Advance Access Published January 7, 2014.
The (Right) Null Space of S Systems Biology by Bernhard O. Polson Chapter9 Deborah Sills Walker Lab Group meeting April 12, 2007.
Mona Yousofshahi, Prof. Soha Hassoun Department of Computer Science Prof. Kyongbum Lee Chemical & Biological Engineering Tufts University 1.
The variation in flux through any reaction can be related to its reaction mechanism, where the flux through the reaction is described as a function of.
Multidimensional Optimality of Microbial Metabolism Robert Schuetz, Nicola Zamboni, Mattia Zampieri, Matthias Heinemann, Uwe Sauer Science 4 May 2012:
UC Davis, May 18 th 2006 Introduction to Biological Networks Eivind Almaas Microbial Systems Division.
Flux Balance Analysis. FBA articles Advances in flux balance analysis. K. Kauffman, P. Prakash, and J. Edwards. Current Opinion in Biotechnology 2003,
Integration of enzyme activities into metabolic flux distributions by elementary mode analysis Kyushu Institute of Technology Hiroyuki Kurata, Quanyu Zhao,
Models and methods in systems biology Daniel Kluesing Algorithms in Biology Spring 2009.
Regulated Flux-Balance Analysis (rFBA) Speack: Zhu YANG
Flux balance analysis in metabolic networks Lecture notes by Eran Eden.
Metabolic network analysis Marcin Imielinski University of Pennsylvania March 14, 2007.
Evolution of minimal metabolic networks WANG Chao April 11, 2006.
In silico aided metaoblic engineering of Saccharomyces cerevisiae for improved bioethanol production Christoffer Bro et al
Experimental and computational assessment of conditionally essential genes in E. coli Chao WANG, Oct
Integrated analysis of regulatory and metabolic networks reveals novel regulatory mechanisms in Saccharomyces cerevisiae Speaker: Zhu YANG 6 th step, 2006.
The global transcriptional regulatory network for metabolism in Escherichia coli exhibits few dominant functional states Speaker: Zhu Yang
1 Escheria coli K-12 undergoes adaptive evolution to achieve in silico predicted optimal growth Rafael U. Ibarra, Jeremy S. Edwards and Bernhard Ø. Palsson.
Gene regulation and metabolic flux reorganization in aerobic/anaerobic switch of E. coli Chao WANG July 19, 2006.
Constraint-Based Modeling of Metabolic Networks Tomer Shlomi School of Computer Science, Tel-Aviv University, Tel-Aviv, Israel March, 2008.
Humboldt- Universität Zu Berlin Edda Klipp, Humboldt-Universität zu Berlin Edda Klipp Systembiologie 4 – Flux Balance Analysis Sommersemester 2010 Humboldt-Universität.
This work was performed under the auspices of the U.S. Department of Energy by University of California, Lawrence Livermore National Laboratory under Contract.
Network-based data integration reveals extensive post-transcriptional regulation of human tissue-specific metabolism Tomer Shlomi*, Moran Cabili*, Markus.
Metabolic/Subsystem Reconstruction And Modeling. Given a “complete” set of genes… Assemble a “complete” picture of the biology of an organism? Gene products.
Engineering of Biological Processes Lecture 4: Production kinetics Mark Riley, Associate Professor Department of Ag and Biosystems Engineering The University.
1 Introduction to Biological Modeling Steve Andrews Brent lab, Basic Sciences Division, FHCRC Lecture 3: Metabolism Oct. 6, 2010.
VL Netzwerke, WS 2007/08 Edda Klipp 1 Max Planck Institute Molecular Genetics Humboldt University Berlin Theoretical Biophysics Networks in Metabolism.
Richard Notebaart Systems biology / Reconstruction and modeling large biological networks.
Biological Network Analysis: Introduction to Metabolic Networks Tomer Shlomi Winter 2008.
Metabolic Model Describing Growth of Substrate Uptake By Idelfonso Arrieta Anant Kumar Upadhyayula.
Lecture #23 Varying Parameters. Outline Varying a single parameter – Robustness analysis – Old core E. coli model – New core E. coli model – Literature.
Genetic modification of flux (GMF) for flux prediction of mutants Kyushu Institute of Technology Quanyu Zhao, Hiroyuki Kurata.
Transcriptional Regulation in Constraints-based metabolic Models of E. coli Published by Markus Covert and Bernhard Palsson, 2002.
The Optimal Metabolic Network Identification Paula Jouhten Seminar on Computational Systems Biology
Improving NADPH availability for natural product biosynthesis in Escherichia coli by metabolic engineering 汇报人:刘巧洁.
Solution Space? In most cases lack of constraints provide a space of solutions What can we do with this space? 1.Optimization methods (previous lesson)
Steady-state flux optima AB RARA x1x1 x2x2 RBRB D C Feasible flux distributions x1x1 x2x2 Max Z=3 at (x 2 =1, x 1 =0) RCRC RDRD Flux Balance Constraints:
A chemostat approach to analyze the distribution of metabolic fluxes in wine yeasts during alcoholic fermentation Quirós, M. 1, Martínez-Moreno, R. 1,
Lecture 6: Product Formation Stoichiometry
BIOINFORMATICS ON NETWORKS Nick Sahinidis University of Illinois at Urbana-Champaign Chemical and Biomolecular Engineering.
Genome-scale constraint-based metabolic model of Clostridium thermocellum Chris M. Gowen 1,3, Seth B. Roberts 1, Stephen S. Fong 1,2 1 Department of Chemical.
1 Departament of Bioengineering, University of California 2 Harvard Medical School Department of Genetics Metabolic Flux Balance Analysis and the in Silico.
Introduction: Acknowledgments Thanks to Department of Biotechnology (DBT), the Indo-US Science and Technology Forum (IUSSTF), University of Wisconsin-Madison.
10 AM Tue 20-Feb Genomics, Computing, Economics Harvard Biophysics 101 (MIT-OCW Health Sciences & Technology 508)MIT-OCW Health Sciences & Technology 508.
Metabolic pathway alteration, regulation and control (3) Xi Wang 01/29/2013 Spring 2013 BsysE 595 Biosystems Engineering for Fuels and Chemicals.
In silico gene targeting approach integrating signaling, metabolic, and regulatory networks Bin Song Jan 29, 2009.
Purpose of the Experiment  Fluxes in central carbon metabolism of a genetically engineered, riboflavin-producing Bacillus subtilis strain were investigated.
Flexibility in energy metabolism supports hypoxia tolerance in Drosophila flight muscle: metabolomic and computational systems analysis Jacob Feala 1,2.
19. Lecture WS 2003/04Bioinformatics III1 Computational Studies of Metabolic Networks - Introduction Different levels for describing metabolic networks:
Lecture #19 Growth states of cells. Outline Objective functions The BOF The core E. coli model The genome-scale E. coli model Using BOF.
Essence of Metabolic Engineering
Project 2 Flux Balance Analysis of Mitochondria Energy Metabolism Suresh Gudimetla Salil Pathare.
V15 Flux Balance Analysis – Extreme Pathways
Virginia Commonwealth University Department of Chemical and Life Science Engineering Evolutionary Engineering Laboratory
Flexibility in energy metabolism supports hypoxia tolerance in Drosophila flight muscle: metabolomic and computational systems analysis Jacob Feala Laurence.
BT8118 – Adv. Topics in Systems Biology
BT8118 – Adv. Topics in Systems Biology
Structural analysis of metabolic network models
Building Metabolic Models
Nutrigenomics/pharmacogenomics
System Biology ISA5101 Final Project
BT8118 – Adv. Topics in Systems Biology
BT8118 – Adv. Topics in Systems Biology
Optimality principles in adaptive evolution.
Presentation transcript:

Biological Network Analysis: Metabolic Optimization Methods Tomer Shlomi Winter 2008

Linear Programming c, l, A, b, α, β are parameters Problem may be either feasible or infeasible If the problem has an unique optimal value: –It may either have a single optimal solution –Or a space of optimal solutions Alternatively, the problem may be unbounded

CBM Example (I)

CBM Example (II)

CBM Example (III)

Flux Balance Analysis Searches for a steady-state flux distribution v: Satisfying thermodynamic and capacity constraints: S∙v=0 v min ≤v ≤v max With maximal growth rate Max v biomass

Lecture Outline 1. Growth rate predictions a.Phenotypic Phase Plane (PPP) analysis 2. Gene knockout lethality predictions a.FBA b.Minimization of Metabolic Adjustment (MOMA) c.Regulatory On/Off Minimization (ROOM) 3. Predicting knockout strategy for metabolic production a.OptKnock b.OptStrain 4. Gene function prediction

1. Growth rate predictions

Flux Balance Analysis (reminder) Searches for a steady-state flux distribution v with maximal growth rate: S∙v=0 v min ≤v ≤v max Max v biomass Requires bounds on metabolite uptake rates (b1)

Phenotype Phase Planes (PPP) (I) X axis – Succinate uptake rate Y axis – Oxygene uptake rate Z axis - Growth rate (maximal value of the objective function as function of succinate and oxygen uptake) Line of optimality

Phenotype Phase Planes (PPP) (II) Observations: Schilling 2001 Metabolic network is unable to utilize succinate as sole carbon source in anaerobic conditinos. Region 1: oxygen excess – this region is wasteful – (less carbon is available for biomass production since it is oxidized to eliminate the excess oxygen.) Succinate Oxygene Growth rate Region 3- the uptake of additional succinate has a negative effect. Cellular resources are required to eliminate excessive succinate.

Does E. coli behave according to Phenotype Phase Planes? (I) E. coli was grown with malate as sole carbon source. A range of substrate concentrations and temperatures were used in order to vary the malate uptake rate (MUR). Oxygen uptake rate (OUR) and growth rate were measured

Does E. coli behave according to Phenotype Phase Planes? (II) Malate/oxygen PPP Ibarra et al., Nature 2002 The experimentally determined growth rate were on the line of optimality of the PPP !

Does E. coli behave according to Phenotype Phase Planes? (III) Malate/oxygen PPP Ibarra et al., Nature 2002 Is the optimal performance on malate stable over prolonged periods of time? Evolution of E. coli on malate was studied for 500 generations in a single condition… 2- An adaptive evolution was observed with an increase of 19%in growth rate! 3- Same adaptive evolution was observed for succinate and Malate!

Does E. coli behave according to Phenotype Phase Planes? (III) Same experiments were made using glycerol as sole carbon source Day 0 – Sub optimal growth Day 1-40 – evolution toward optimal growth Day 40 –optimal growth Day 60 –optimal growth (no change) Why?

2. Gene Knockout Lethality

Predicting Knockout Lethality (I) A gene knockout is simulated by setting the flux through the corresponding reaction to zero The corresponding reactions are identified by evaluating the Boolean gene-to-reaction mapping in the model

Predicting Knockout Lethality (II) A gene is predicted essential if it’s knockout yields a significant drop in the maximal possible growth rate v1 is essential for growth v6 is not essential for growth

Gene knockout lethality: E. coli in glycerol minimal media In total, 819 out of the 896 mutants (91%) showed growth behaviors in glycerol minimal medium in agreement with computational predictions 69% correct prediction out of the experimental essential genes

2. Gene essentiality prediction Gene knockout lethality: Resolving Discrepancies (I)

Gene knockout lethality: Resolving Discrepancies (II)

3. MOMA and ROOM

Minimization of Metabolic Adjustment (MOMA) (I) FBA assumes optimality of growth for wild type – evolution drives the growth rate towards optimality This assumption is not necessarily correct following a gene knockout! What other objective can capture the biological essence of these mutations? (hint – the title of this slide)

Minimization of Metabolic Adjustment (MOMA) (II) Assumption: following the knockout, the mutant remains as close as possible to the wild-type strain The flux distribution of mutant should also satisfy all constraints as in FBA

Minimization of Metabolic Adjustment (MOMA) (III) Formally: w – the wild-type optimal growth vector (obtained via FBA). v – a vector in mutant flux space. Find V m which minimizes the Euclidian distance to V wt : Min (w -v)², - minimize Euclidian distance s.t S∙v = 0, - mass balance constraints v min  v  v max - capacity constraints v j = 0, j  G - knockout constraints Solved using Quadratic Programming (QP) w v

Validating MOMA: Gene essentiality prediction

Validating MOMA: Experimental fluxes

Regulatory On/Off Minimization (ROOM) (I) Assumption: The organism adapts by minimizing the set of flux changes (via the regulatory system) Search for a feasible flux distribution with minimal number of changes from the wild-type ABC D E byp cof byp cof Wild-type solution Knockout solution

Regulatory On/Off Minimization (ROOM) (II) Min  y i - minimize changes s.t v – y ( v max - w)  w- distance constraints v – y ( v min - w)  w- distance constraints S∙v = 0,- mass balance constraints v j = 0, j  G - knockout constraints Integer variables are required to track the ‘number of changes in flux’ from the wild-type Use Boolean auxiliary variables y to reflect changes in flux between the wild-type and mutant y i =0if and only if v i = w i Formulate a MILP problem to find a pair of v and y with a minimal sum of y i ’s.

Validating ROOM: Alternative pathways ROOM identifies short alternative pathways to re-route metabolic flux following a gene knockout, in accordance with experimental data

Validating ROOM: Experimental fluxes (I) Intracellular fluxes measurements in E. coli central carbon metabolism Obtained using NMR spectroscopy in C labelling experiments Knockouts: pyk, pgi, zwf, and gnd in Glycolysis and Pentose Phosphate pathways Glucose limited and Ammonia limited medias FBA wild-type predictions above 90% accuracy 13 Emmerling, M. et al. (2002), Hua, Q. et al. (2003), Jiao, Z et al. (2003) (*) Based on a figure from Jiao, Z., et al.

Validating ROOM: Experimental fluxes (II) ROOM flux predictions are significantly more accurate than MOMA and FBA in 4 out of 8 experiments ROOM growth rate predictions are significantly more accurate than MOMA

4. Metabolite Production

Constraint-based Modeling: Biotechnological Applications Design bacteria that produces chemicals of interest Bacteria Objective: Grow Fast Vanillin The major compound in Vanilla Bioengineering Objective: Produce Vanillin Bioengineering Objective Produce Vanillin

OptKnock Designing microbial organisms for efficient production of metabolites Finds reactions whose removal increases the production of metabolite of interest

OptKnock: Optimization problem (I) A nested (bi-level) optimization problem is needed

OptKnock: Optimization problem (I) A nested (bi-level) optimization problem is needed Reactions to remove Cells have to grow Removed reactions have zero flux The max number of reactions to remove

Succinate Production Strains

OptStrain An integrated framework for redesigning microbial production systems Step 1: Creation of universal reactions DB Step 2: Compute maximal theoretical metabolite production yield Step 3: Identifying the minimal number of required to be added to an organism to achieve the maximal production yield. Step 4: Adding the identified reactions and finding gene deletions that ensure metabolite secretion (OptKnock)

OptStrain: Step 1 Creation of universal reactions DB Download set of known reactions from KEGG (Kyoto Encyclopedia of Genes and Genomes) Validate reaction data consistency – remove unbalanced reaction Define a universal stoichiometric matrix S.

OptStrain: Step 2 Determination of maximal theoretical yield of a metabolite of interest Yield – metabolite production rate per unit of substrate uptake Use LP to find the maximal yield for different substrates, denoted R

OptStrain: Step 3 Identification of minimum number of non-native reactions for a host organism MILP formulation – y i represented whether reaction i should be added to the organism

OptStrain: Step 4 Incorporating the non-native reactions into the host organism’s stoichiometric model Eliminate genes such that biomass production is coupled with the production of the metabolite of interest OptKnock

Case study: Hydrogen production The highest hydrogen yield (0.126 g/g substrate consumed) is obtained for methanol

Case study: Hydrogen production (I) Testing E. coli on glucose media Step 3 reveals that new reactions are needed for E. coli on glucose

Case study: Hydrogen production (II) C. acetobutylicum - the "Weizmann Organism", after Chaim Weizmann, who in 1916 helped discover how C. acetobutylicum culture could be used to produce acetone, butanol, and ethanol from starch The knockout of 2 reactions tightly couple biomass production and metabolite hydrogen secretion

Case study: Vanillin production (I) Vanillin is an important flavor and aroma molecule (found in vanilla pods) Maximal theoretical production rate: 0.63 (g/g glucose) E. coli needs 3 new reactions to achieve this vanillin yield Previous bioengineering experiments have already involved the extraction of these 3 reactions from Neurospora crassa and their addition to E. coli However, the resulting vanillin production rate was only 0.15

Case study: Vanillin production (II) OptStrain predicts knockout sets that provide a vanillin yield of 0.57 (g vanillin/g glucose) in E. coli This is close to the maximal theoretical production rate

4. Gene function prediction

Refining Genome Annotation A substantial fraction of the genes have unknown function An integrated computational/experimental approach for predicting gene function: –Identify discrepancies between model predictions and growth phenotyping in E. coli –An algorithm then identifies missing reactions whose addition could reconcile model predictions and experimental observations –Search for ORFs that might be responsible for these missing activities based on literature searches, sequence-homology, etc –experimental verification of the algorithm’s predictions via growth phenotypes of single-deletion strains

Refining Genome Annotation

Refining E.coli’s Annotation Identify 50 minimal medium conditions in which the model cannot explain the observed (experimental) growth Identify reactions whose addition enables growth for 26 of the environemnts 6 cases are investigated in depth

New Transporter Genes (I) Growth on propionate and 5-keto-D-gluconate require the addition of relevant transporters Currently, such transporters are unknown 8 potential transporters are identified via literature searches, sequence-homology, etc Only putP deletion showed reduced growth in propionate Only idnT deletion showed reduced growth in 5- keto-D-gluconate Both genes show increased expression level on these media

New Transporter Genes (II) The algorithm found a missing reaction that secretes a byproduct in thymidine metabolism – thymine Experimental inspection of the growth media support this finding The identity of the transporter gene remained unclear

Growth on D-Malate The algorithm finds two missing reactions: –D-Malate transporter –Decarboxilation of D-Malate to Pyruvate

References: Edwards JS, Ramakrishna R, Palsson BO Characterizing the metabolic phenotype: a phenotype phase plane analysis. Biotechnol Bioeng 77(1): Ibarra RU, Edwards JS, Palsson BO Escherichia coli K-12 undergoes adaptive evolution to achieve in silico predicted optimal growth. Nature 420(6912): Segre D, Vitkup D, Church GM Analysis of optimality in natural and perturbed metabolic networks. Proc Natl Acad Sci U S A 99(23): Shlomi T, Berkman O, Ruppin E Regulatory on/off minimization of metabolic flux changes after genetic perturbations. Proc Natl Acad Sci U S A 102(21): Burgard AP, Pharkya P, Maranas CD Optknock: a bilevel programming framework for identifying gene knockout strategies for microbial strain optimization. Biotechnol Bioeng 84(6): Pharkya P, Burgard AP, Maranas CD OptStrain: a computational framework for redesign of microbial production systems. Genome Res 14(11): Reed JL, Patel TR, Chen KH, Joyce AR, Applebee MK, Herring CD, Bui OT, Knight EM, Fong SS, Palsson BO Systems approach to refining genome annotation. Proc Natl Acad Sci U S A 103(46):