doc: sketch of draft protocol

tutorials/protocol.m

@@ -0,0 +1,76 @@
+%% This script sketches the blueprint for the GECKO3 protocol.
+% At a later stage, it can likely be converted into a script that is
+% referenced in the protocol paper. Right now, this script might contain
+% comments and clarifications that should rather be included in the
+% protocol than in here, while other comments might be best kept in this
+% script. At the moment it is just a sketchbook.
+
+%% Prepare for ec model reconstruction
+% - Prepare a folder in userData, or select another path whether all model-
+%   specific files and scripts will be kept
+% - Find a high-quality GEM of your species of interest
+%   - For reference, might be best to be stored together with the
+%     model-specific files, e.g. userData/ecYeastGEM/model/yeast-GEM.xml
+% - Install RAVEN & GECKO
+%   - RAVEN can be installed in many different ways, refer to GitHub Wiki.
+%   - GECKO can be installed via cloning or direct download of ZIP file.
+%     Should GECKO also be distributed as MATLAB Add-On, just like RAVEN?
+%   - GECKO should be added to the MATLAB path.
+% - On Uniprot:
+%   - Search a proteome for your species and note the Organism ID / TaxId
+%   - Check which gene identifiers are used in the model, and see where
+%     this is represented in Uniprot. E.g. for yeast, the genes are in
+%     "Gene Names (Ordered locus)", which via the API can be gather via
+%     "gene_oln" according to https://www.uniprot.org/help/return_fields
+% - On KEGG:
+%   - Find the organism code of your species
+% - Prepare a ModelAdapter file with some of the above information and
+%   more.
+
+%% Summary
+% Pipeline can look different dependent on user preferences. Getting an
+% all-DLKcat model would look something like:
+% makeEcModel -> findMetSmiles -> writeDLKcatInput -> readDLKcatOutput ->
+% selectKcatValue -> applyKcatConstraints -> ...
+% while a fuzzy matching to BRENDA approach looks like:
+% makeEcModel -> getECfromDatabase -> fuzzyMatching -> applyKcatConstraints
+% -> ...
+
+%% Initiate model reconstruction
+% Load the model with RAVEN's importModel
+modelY = importModel(fullfile(findGECKOroot,'userData','ecYeastGEM','models','yeast-GEM.xml'));
+
+% Set the ModelAdapter correctly
+setGECKOModelAdapter('userData/ecYeastGEM/ModelAdapter.m')
+
+ecModel = makeEcModel(modelY);
+% Read makeEcModel documentation to get a list of all it does: it prepare
+% the new model.ec structure and prepares the S-matrix by splitting
+% reversible reactions, isozymes etc.
+% Can also make a geckoLight model with makeEcModel(..,..,true);
+
+%% Gather kcat values
+% Different approaches are possible: (1) DLKcat; (2) fuzzy matching; (3)
+% manual kcatList. For (1) and (2), ecRxns parameter can indicate that you
+% do not want to get kcats for all reactions. You might want to first run
+% fuzzy matching, only keep the ones without any EC wildcards, and then run
+% DLKcat for the remaining reactions.
+
+% (1) DLKcat
+% Requires metabolite SMILES:
+model.metSmiles = findMetSmiles(model.metNames);
+% Currently, a DLKcatInput.tsv file is written that can be used by DLKcat,
+% and the DLKcatOutput.tsv file can be loaded into MATLAB again. Feiran is
+% also working on providing DLKcat as a package that can directly be called
+% by GECKO/MATLAB.
+writeDLKcatInput(model);
+kcatList = readDLKcatOutput(model);
+ecModel  = selectKcatValue(ecModel,kcatList);
+ecModel  = applyKcatConstraints(ecModel);
+
+% (2) fuzzy matching
+% Requires EC-codes
+ecModel  = getECfromDatabase(ecModel);
+kcatList = fuzzyKcatMatching(ecModel);
+ecModel  = selectKcatValue(ecModel,kcatList);
+ecModel  = applyKcatConstraints(ecModel);