Guide#

Each page of the guide covers one task, with MATLAB and Python examples side by side. The pages can be read in the order below or one at a time; Getting started is the first.

The reconstruction protocol for H. polymorpha and the legacy tutorials from the original RAVEN paper are separate from the task pages. Both use MATLAB.

Functions from cobrapy#

A function that comes from cobrapy carries a cobrapy badge that links to the cobrapy documentation, and its import line shows where it lives (from cobra.io import read_sbml_model). The function mapping lists all of these functions.

Executed examples#

The Guide examples workflow runs the code on the task pages, MATLAB and Python separately. It starts when a guide page, the example data or the runner changes, and once a week against the latest RAVEN and raven-toolbox. A run fails when a block raises an error or prints output that differs from the output shown beneath it. The blocks of a page run in order in one session, so a block uses what the preceding blocks made. Reading a page starts at its Setup section.

One page and part of another are not executed. Reconstruction from KEGG downloads tens of megabytes and takes minutes, and the Human-GEM steps of Context-specific models with ftINIT take hours. Both pages quote the run that their numbers come from. A few further blocks are marked as not run on their own page, with the reason. The protocol and the legacy tutorials are not executed.

Example data#

File

Content

smallYeast.yml

Central carbon metabolism of yeast; the default example

smallYeastBad.yml

The same model with deliberate errors, used on the curation pages

empty.xml

A starter model with four metabolites and one reaction, the start of a model built from scratch

yeast-GEM.yml

yeast-GEM v9.1.0, the consensus S. cerevisiae model, for pages that need a genome-scale model

yeast-GEM.xml

yeast-GEM v9.1.0 as SBML

anaerobic.yml

A condition file: yeast-GEM under anaerobic conditions

Pages#

Foundations#

  1. Getting started: load a model and inspect it.

  2. Model structure and identifiers: how the RAVEN struct and cobra.Model correspond, field by field.

  3. Reading and writing models: SBML, RAVEN YAML, Excel and text, and the layout of a model repository.

Simulation#

  1. Simulating growth with FBA: objective, bounds, solving and reading the fluxes.

  2. Growth media and conditions: defining a medium and keeping a condition as data.

  3. Solvers and configuration: selecting a solver and reading what it returns.

Building and curating#

  1. Building a model from scratch: metabolites, reactions, genes and exchange reactions.

  2. Editing an existing model: stoichiometry, gene-reaction rules, bounds and deletions.

  3. Quality control: structure, mass balance, blocked reactions and mass produced from nothing.

  4. Table-driven curation: curation kept as reviewable tables.

  5. Biomass composition and annotation: the composition of the biomass reaction, rescaling it, and the annotation that the model keeps after the solve.

Reconstruction#

  1. Reconstruction from homology: aligning a proteome against the genes of a template model, and the effect of the cut-offs.

  2. Reconstruction from KEGG: a draft without a template model, and what such a draft lacks.

  3. Context-specific models with ftINIT: integrating omics data and extracting a cell-line model from Human-GEM and RNA-seq data.

Analysis and repair#

  1. Metabolic tasks: defining what a model must be able to do and checking that it keeps that ability.

  2. Deletions and essentiality: knockouts, essential genes and reactions, and MOMA.

  3. Gap-filling: closing the gaps of a draft model.

  4. Flux variability and alternative optima: the range of each flux and the predictions that the model fixes.

  5. Random sampling: the flux distribution behind those ranges and its conditioning on a state.

Comparing and engineering#

  1. Combining and simplifying models: merging models on metabolite names and taking a model apart.

  2. Comparing models: the differences between two models and their effect on predictions.

  3. Engineering targets: FSEOF for over-production targets and reporter metabolites for the metabolites around a transcriptional response.

Protocol and legacy tutorials#

Resource

Content

GEM reconstruction protocol, H. polymorpha

Homology-based reconstruction of hanpo-GEM, from the template models to a draft that grows on methanol. MATLAB; not executed.

Legacy tutorials

Five exercises from the original RAVEN paper (Agren et al., 2013), adapted to run with current RAVEN and otherwise unchanged. MATLAB.

The task pages are for looking up a single task. The protocol follows one reconstruction in order.

Compartment localisation#

No page covers it. The feature is in development and the two toolboxes differ in it.