Workflow overview#

The diagram orders the functions that build, parameterise, tune and analyse an ecModel, with the parts that run in parallel and the alternatives. Each stage is a guide page. A solid arrow connects functions that run one after another, a dashed arrow an optional preview or an alternative way to reach the next step, a diamond a choice between paths, and a dashed outline an optional step. The button in the corner of a diagram opens it in full screen, where the mouse wheel zooms and a drag pans. In the page, the wheel zooms with Ctrl (Cmd on macOS) held. The download icon at the top of each tab saves its diagram as a PDF.

Ⓜ️ MATLAB

Download the MATLAB diagram as PDF

        ---
config:
  flowchart:
    useMaxWidth: false

---
flowchart TB
  classDef opt stroke-dasharray: 5 4
  A["<b>Getting started</b><br/>createGECKOproject<br/>Fill model_adapter.toml"]
  B["<b>Building an empty ecModel</b><br/>ModelAdapterManager.setDefault<br/>loadConventionalGEM<br/>makeEcModel (full or light)<br/>getComplexData, applyComplexData (optional)<br/>saveEcModel"]
  A --> B
  EC["EC numbers<br/>getECfromGEM or getECfromDatabase"]
  SM["Metabolite SMILES<br/>findMetSmiles"]
  HO["kcatsFromHomologs"]
  BR["kcatsFromBrenda"]
  TR["kcatsFromECTree"]
  DL["kcatsFromDLKcat"]
  OK["kcatsFromOKP"]
  CU["data/kcats/custom.tsv"]:::opt
  GA["<b>Gathering kcats</b><br/>any subset of the sources"]
  B --> GA
  GA --> EC
  GA --> SM
  GA --> HO
  EC --> BR
  EC --> TR
  SM --> DL
  SM --> OK
  AS["<b>Applying kcats</b><br/>assignKcats (preview: collectKcats)<br/>getKcatAcrossIsozymes, getStandardKcat (optional)<br/>applyKcatConstraints<br/>setProtPoolSize"]
  BR --> AS
  TR --> AS
  HO --> AS
  DL --> AS
  OK --> AS
  CU --> AS
  TU{"solveLP: growth reaches the target?"}
  AS --> TU
  SE["<b>Growth-rate tuning</b><br/>sensitivityTuning<br/>sigmaFitter"]
  CM["<b>Tuning against experimental data</b><br/>screenKcatLeverage<br/>reviewKcatAssignment<br/>cmaesKcatTuning"]
  TU -->|no| SE
  TU -->|no, several conditions measured| CM
  FM["Functional ecModel<br/>saveEcModel"]
  TU -->|yes| FM
  SE --> FM
  CM --> FM
  PR["<b>Proteomics integration</b><br/>loadProtData, fillEnzConcs<br/>levelComplexConcs, constrainEnzConcs<br/>calculateFfactor, setProtPoolSize<br/>loadFluxData, constrainFluxData"]:::opt
  FM --> PR
  RL{"Growth reaches the target?"}
  PR --> RL
  FX["<b>Relaxing</b>, gradual<br/>flexibilizeEnzConcs"]
  GR["<b>Relaxing</b>, fast<br/>relaxProteomicsGreedy"]
  RL -->|no, gradual| FX
  RL -->|no, fast| GR
  RD["ecModel for simulation and analysis"]
  RL -->|yes| RD
  FX --> RD
  GR --> RD
  FM -.->|without proteomics| RD
  I1["<b>Simulation</b><br/>ecFVA, randomSampling<br/>ecFSEOF<br/>getSubsetEcModel"]
  I2["<b>Enzyme usage</b><br/>getEnzymeUsage<br/>getEnzymeBottlenecks"]
  RD --> I1
  RD --> I2

🐍 Python

Download the Python diagram as PDF

        ---
config:
  flowchart:
    useMaxWidth: false

---
flowchart TB
  classDef opt stroke-dasharray: 5 4
  A["<b>Getting started</b><br/>geckopy init<br/>Fill model_adapter.toml"]
  B["<b>Building an empty ecModel</b><br/>ModelAdapter.from_folder<br/>load_conventional_gem<br/>make_ec_model (full or light)<br/>get_complex_data, apply_complex_data (optional)<br/>save_ec_model"]
  A --> B
  EC["EC numbers<br/>fill_eccodes_from_gem or fill_eccodes_from_database"]
  SM["Metabolite SMILES<br/>find_met_smiles"]
  HO["kcats_from_homologs"]
  BR["kcats_from_brenda"]
  TR["kcats_from_ec_tree"]
  DL["kcats_from_dlkcat"]
  OK["kcats_from_okp"]
  CU["data/kcats/custom.tsv"]:::opt
  GA["<b>Gathering kcats</b><br/>any subset of the sources"]
  B --> GA
  GA --> EC
  GA --> SM
  GA --> HO
  EC --> BR
  EC --> TR
  SM --> DL
  SM --> OK
  AS["<b>Applying kcats</b><br/>assign_kcats (preview: collect_kcats)<br/>fill_kcats_from_isozymes, assign_standard_kcat (optional)<br/>apply_kcat_constraints<br/>set_prot_pool_size"]
  BR --> AS
  TR --> AS
  HO --> AS
  DL --> AS
  OK --> AS
  CU --> AS
  TU{"model.optimize: growth reaches the target?"}
  AS --> TU
  SE["<b>Growth-rate tuning</b><br/>sensitivity_tuning<br/>fit_sigma"]
  CM["<b>Tuning against experimental data</b><br/>screen_kcat_leverage<br/>review_assignment<br/>cmaes_kcat_tuning"]
  TU -->|no| SE
  TU -->|no, several conditions measured| CM
  FM["Functional ecModel<br/>save_ec_model"]
  TU -->|yes| FM
  SE --> FM
  CM --> FM
  PR["<b>Proteomics integration</b><br/>load_prot_data, fill_enz_concs<br/>level_complex_concs, constrain_enz_concs<br/>calculate_f_factor, set_prot_pool_size<br/>load_flux_data, apply_flux_data_constraints"]:::opt
  FM --> PR
  RL{"Growth reaches the target?"}
  PR --> RL
  FX["<b>Relaxing</b>, gradual<br/>flexibilize_enz_concs"]
  GR["<b>Relaxing</b>, fast<br/>relax_proteomics_greedy"]
  RL -->|no, gradual| FX
  RL -->|no, fast| GR
  SA["<b>Relaxing</b>, smallest total amount<br/>relax_enz_concs"]
  RL -->|no, smallest total amount| SA
  RD["ecModel for simulation and analysis"]
  RL -->|yes| RD
  FX --> RD
  GR --> RD
  SA --> RD
  FM -.->|without proteomics| RD
  I1["<b>Simulation</b><br/>ec_fva, random_sampling<br/>ec_fseof<br/>get_subset_ec_model"]
  I2["<b>Enzyme usage</b><br/>enzyme_usage<br/>get_enzyme_bottlenecks"]
  RD --> I1
  RD --> I2

Stages, order and alternatives#

Each row is a stage with the stage it runs after, its functions and what it leaves behind. "Any subset" marks functions that run independently of each other, "alternative" a path that replaces another, and "one of" a choice of algorithm.

Ⓜ️ MATLAB

Stage	Run after	Functions	Output
Getting started	none	`createGECKOproject`, then edit `model_adapter.toml`; `loadDatabases('kegg')` (optional)	project folder
Building an empty ecModel	project and model adapter	`ModelAdapterManager.setDefault`, `loadConventionalGEM`, `makeEcModel` (full or light), `getComplexData, applyComplexData` (optional), `saveEcModel`	ecModel file
Gathering kcats	built ecModel	any subset of `kcatsFromBrenda`, `kcatsFromECTree`, `kcatsFromHomologs`, `kcatsFromDLKcat`, `kcatsFromOKP`; `updateKcatSources` gathers all sources of `model_adapter.toml`	`data/kcats/<source>.tsv`
Applying kcats	at least one source file or `custom.tsv`	`assignKcats` (preview: `collectKcats`), `getKcatAcrossIsozymes` and `getStandardKcat` (optional), `applyKcatConstraints`, `setProtPoolSize`	`ec.kcat`, constrained ecModel
Growth-rate tuning	applied kcats, growth rate below the target	`sensitivityTuning`, then `sigmaFitter`	tuned `ec.kcat`
Tuning against experimental data	applied kcats, growth rates and fluxes of several conditions	alternative to sensitivity tuning: `screenKcatLeverage`, `reviewKcatAssignment`, `cmaesKcatTuning`	tuned `ec.kcat`
Proteomics integration	functional ecModel	`loadProtData`, `fillEnzConcs`, `levelComplexConcs`, `constrainEnzConcs`, `calculateFfactor, setProtPoolSize`, `loadFluxData, constrainFluxData` (optional)	proteomics-constrained ecModel
Relaxing overconstrained proteomics	proteomics-constrained ecModel below the target growth rate	one of `flexibilizeEnzConcs`, `relaxProteomicsGreedy`	relaxed ecModel
Simulation and analysis, Enzyme usage and bottlenecks	any ecModel	independent of each other: `ecFVA`, `randomSampling`, `ecFSEOF`, `getSubsetEcModel`, `getEnzymeUsage`, `getEnzymeBottlenecks`	results

🐍 Python

Stage	Run after	Functions	Output
Getting started	none	`geckopy init`, then edit `model_adapter.toml`; `geckopy kegg-download` (optional)	project folder
Building an empty ecModel	project and model adapter	`ModelAdapter.from_folder`, `load_conventional_gem`, `make_ec_model` (full or light), `get_complex_data, apply_complex_data` (optional), `save_ec_model`	ecModel file
Gathering kcats	built ecModel	any subset of `kcats_from_brenda`, `kcats_from_ec_tree`, `kcats_from_homologs`, `kcats_from_dlkcat`, `kcats_from_okp`; `update_kcat_sources` gathers all sources of `model_adapter.toml`	`data/kcats/<source>.tsv`
Applying kcats	at least one source file or `custom.tsv`	`assign_kcats` (preview: `collect_kcats`), `fill_kcats_from_isozymes` and `assign_standard_kcat` (optional), `apply_kcat_constraints`, `set_prot_pool_size`	`ec.kcat`, constrained ecModel
Growth-rate tuning	applied kcats, growth rate below the target	`sensitivity_tuning`, then `fit_sigma`	tuned `ec.kcat`
Tuning against experimental data	applied kcats, growth rates and fluxes of several conditions	alternative to sensitivity tuning: `screen_kcat_leverage`, `review_assignment`, `cmaes_kcat_tuning`	tuned `ec.kcat`
Proteomics integration	functional ecModel	`load_prot_data`, `fill_enz_concs`, `level_complex_concs`, `constrain_enz_concs`, `calculate_f_factor, set_prot_pool_size`, `load_flux_data, apply_flux_data_constraints` (optional)	proteomics-constrained ecModel
Relaxing overconstrained proteomics	proteomics-constrained ecModel below the target growth rate	one of `flexibilize_enz_concs`, `relax_proteomics_greedy`, `relax_enz_concs`	relaxed ecModel
Simulation and analysis, Enzyme usage and bottlenecks	any ecModel	independent of each other: `ec_fva`, `random_sampling`, `ec_fseof`, `get_subset_ec_model`, `enzyme_usage`, `get_enzyme_bottlenecks`	results

Reading the paths#

Kcat sources: each kcatsFrom* function (Python kcats_from_*) writes its own file in data/kcats. assignKcats reads the files that [kcat_assignment] of model_adapter.toml lists, so any combination of sources works. The EC numbers feed the BRENDA, EC tree and homolog sources, and the metabolite SMILES feed DLKcat and OpenKineticsPredictor (see Gathering kcats). updateKcatSources (Python update_kcat_sources) gathers the sources that model_adapter.toml needs in one call.
Tuning: sensitivity tuning raises kcats one at a time until one target growth rate is reached. Tuning against experimental data fits many kcats to the measured growth rates and fluxes of several conditions.
Relaxation: flexibilizeEnzConcs and relaxProteomicsGreedy are alternatives with different stringency (see Relaxing overconstrained proteomics).
Analysis: the analysis functions work on the finished ecModel and do not depend on each other.

Workflow overview#

Stages, order and alternatives#

Reading the paths#

See also#