1. Building Metabolic Models
© 2013 SRI International

2. Metabolic Model Building
Wonderful puzzles!
Convert molecules in the environment into the molecules needed to sustain life

3. Flux-Balance Analysis
Nutrients
Metabolic Reaction List
Secretions A A B C D D
Biomass Metabolites X

4. What Could Possibly Go Wrong?
Incomplete knowledge
Missing reactions
Genome annotation is incomplete
Genome annotation is incorrect
MetaCyc reactions sometimes in incorrect directions
Failures in reaction instantiation
Model uses the wrong reactions
Utilization of reactions that are inactivated
Nutrients, secretions, BMs uncertain

5. Reactions in the Metabolic Model
Reactions in the metabolic model are obtained from the PGDB
Metabolic pathways in PGDB not relevant to metabolic model
Note also that .fba file can modify reactions present in model
Reactions exist in PGDB because
Presence inferred by PathoLogic from enzymes in annotated genome
Metabolic pathway inference imports reactions in MetaCyc that may lack enzymes in the PGDB
Reactions added or deleted manually by the user

6. PathoLogic Reactome Inference
For each protein in the organism
Infer reaction(s) it catalyzes
Import reaction(s) from MetaCyc into current PGDB
Match protein functions to MetaCyc reactions
Enzyme names (uncontrolled vocabulary)
EC numbers
Gene Ontology terms
Genome annotation subject to false positives and false negatives
Reactome inference subject to false positives and false negatives

7. PathoLogic Enzyme Name Matcher
Name matcher generates alternative variants of each name and matches each to MetaCyc
Strips extraneous information found in enzyme names
Putative carbamate kinase, alpha subunit
Flavin subunit of carbamate kinase
Cytoplasmic carbamate kinase
Carbamate kinase (abcD)
Carbamate kinase ( )

8. Read the Literature for Your Organism
Chemical composition of growth medium
Energy source, elemental sources
Quantitative data
Cellular doubling time
P/O ratio if cell respires
Anaerobic respiration electron acceptors?
Cellular chemical composition – BMs and their coefficients
% fraction of dry cell weight for protein, RNA, etc
Membrane composition
Length of quinol chains
Known pathways, enzymes, transporters
On/off under different growth conditions
Cellular architecture

9. The Metabolic Maze Seven major and intertwined themes:
Atomic assimilation of carbon, nitrogen, sulfur, phosphorus
Electron transfer
Nucleotide phosphorylation
Production of ion gradients

10. Starting Point Place nutrients into the cytoplasm
Set an upper bound of 10 on one nutrient
Leave other nutrients unbounded
Define as unbounded nutrients and secretions:
CO2, WATER, PROTON

11. Prop Compounds
Add this set of temporary prop compounds as both nutrients and secretions
ATP[CCO-CYTOSOL]
ADP[CCO-CYTOSOL]
Pi[CCO-CYTOSOL]
NAD[CCO-CYTOSOL]
NADH[CCO-CYTOSOL]
NADP[CCO-CYTOSOL]
NADPH[CCO-CYTOSOL]
WATER[CCO-CYTOSOL]
Reduced-ferredoxins[CCO-CYTOSOL]
Oxidized-ferredoxins[CCO-CYTOSOL]
Reduced-flavodoxins[CCO-CYTOSOL]
Oxidized-flavodoxins[CCO-CYTOSOL]
CO-A[CCO-CYTOSOL]
ACP[CCO-CYTOSOL]
HYDROGEN-MOLECULE[CCO-CYTOSOL]
PROTON[CCO-CYTOSOL]
THF[CCO-CYTOSOL]
5-METHYL-THF[CCO-CYTOSOL]
METHYLENE-THF[CCO-CYTOSOL]
10-FORMYL-THF[CCO-CYTOSOL]

12. Achieve Production of all Biomass Metabolites

13. Finally Remove prop compounds
Implement transport so nutrients are no longer placed in the cytoplasm
Some compounds change form during transport

14. Bottom-Up vs Top-Down Model Building
Proceed incrementally from nutrients to selected intermediates and BMs
Slow, methodical, carefully ordered
Controls complexity by incrementally expanding the needed network
Ensure the correct reactions are used to produce each BM
Expand explicit reaction list along the way
Top down:
Throw all nutrients, reactions, BMs into a pot and stir
Let gap filler figure out what is missing
Faster, but less controlled
Gap filler is powerful, but minimal-cost solution is not always correct
BMs may be produced by the wrong reactions
Both approaches can be accelerated by gap filler and blocked-reaction tools

15. Bottom-Up Model Building
Reach all biomass metabolites, remove all prop compounds
Work forward toward one biomass metabolite at a time
See bottom-up-methodology.docx

16. Validating Your Model Small-scale validation Large-scale validation
All BMs can be made from expected nutrient set
Relative nutrient uptake rates and associated cellular growth rate
Reaction usage, reaction fluxes
Large-scale validation
Gene knock-outs
Different nutrient conditions