Network reconstruction tutorial

**Figure:** Reconstruction of a metabolic network by a team of curators. The datasets are stored on a central password-protected web server. The information in the dataset as well as optional dictionary files are used to export an SBML file.

This tutorial demonstrates how Metannogen is used by curators to reconstruct a metabolic network. It shows how reactions in already published networks can be copied and subsequently modified. The resulting SBML file will be created from the information stored in the datasets. Start Metannogen by clicking

. The left upper panel contains the networks "KEGG", tab "RECON1" and tab "EHMN" as a browsable tree. The network components in these tree can be used to define reactions for the network to be reconstructed.

Creating Datasets

Open the KEGG network tree (first tab in the upper left panel) and open the of an enzymatic reaction, i.e. expand a pathway node and chose one of the reactions. Then, click the menu item " New dataset". A new dataset for this enzymatic reaction is created.

**Figure:** Input mask for a dataset.

The identifier and the equation is taken from the source reaction. Modify the equation. Add metabolites. Since identifiers may be hard to remember, metabolites can also be typed as metabolite names. word completion may help with long and complicated metabolite names. Check out word completion using the tabulator key. After completing a metabolite name, it is translated automatically into the respective ID. A metabolite name is completed when the user types a space character or the at-sign. Also try the alternative way to insert metabolites using the "+Compound" button. Since the editable text field for the equation always contains metabolites as identifiers, the equation is also shown with human readable metabolite names below text field. Finally press the upload-tool button "-> To Server" to save the data on the server.

Compartments

In the following we are going to explain two different ways to specify compartments for biochemical conversions. For this purpose, create a dataset with an enzymatic reaction. When creating or modifying datasets you can always check, what reaction objects would be generated from the respective dataset. Find the "Menu▼" pull down menu within the dataset form. Use the sub-menu "View" to get a list of all reactions defined by the dataset.

Compartment suffices after the metabolite identifiers

Append @cyto (at-sign) and the compartment ID to each metabolite identifier in the equation. Then check the list of reactions. This method is generally applicable. It works even for processes at compartment borders as the following example.

Reactions that involve more than one compartment

Create a transport process for NH3 from "cyto" to "endo". you could type the equation:

NH3@cyto <=> NH3@endo

If you have loaded the KEGG network, Metannogen will replace NH3 by C00014 as you type. Please learn from this example that transport processes over membranes are just biochemical reactions where the sub-cellular locations of the metabolites are not identical.

Compartments by choice menu

Appending the compartment after each metabolite is tedious and unnecessary if all metabolites have the same compartment. In the following we will present a more convenient method. First create a dataset with a biochemical equation. Do not add compartment suffices. Instead find the compartment choice menu at the top of the dataset form. Select . Now check the created reactions. All metabolites inherit the selected compartment, in this case "cyto". If the respective enzyme would also be active in other compartments such as "mitMx" at the same time, further compartments may be specified. Please generate an additional compartment choice menu with the button and select a compartment . If you look at the list of generated reactions you will find two reactions, one for cyto and one for mitoMx. This method has two advantages over the suffix-method:

Less typing
Several localized reactions are defined by one single dataset

Unspecific reactions

Motivation: Some reactions or transporters act not on one specific metabolite, but on a variety of different metabolites. When recording conversions of unspecific enzymes or carriers, curators could create several datasets with slightly different biochemical equation. All other information apart from the reaction would be identical. Redundancy, however, should be avoided and typing must be reduced.

Solution: To avoid creation of different datasets, can be used. To try brace expansion create a transport dataset with the following reaction string:

        { Alanine Proline}@cyto  <=> { Alanine Proline}@mito

Look at the list of created reaction objects. Add further metabolites of your choice and compare the list of created reactions again. You see that in this example the first group of metabolites is exactly the same as the second one. Consequently, you may substitute the second group by "{}". There must not be a space between the two braces:

        { Alanine Proline}@cyto  <=> {}@mito

Make up an example where the two groups differ such that the second group cannot be abbreviated by "{}". (Hint: Transaminases, Desaturases of Fatty acids of different length). Try what happens if the number of elements in each group is not identical.

For transport processes the subcellular localization of the metabolites are written as suffices. Here brace expansion can also be applied. The following describes the transmembranal glucose transport across the endoblasmic reticulum membrane and across the cell membrane.

        C00031@{cyto cyto} <=> C00031@{ext endo}

It expands to

        C00031@cyto <=> C00031@ext 
          C00031@cyto <=> C00031@endo

_{An attribute declaration (See Attributes) like $FUNCTION="Metabolism" is inherited by all reactions expanded from the braces.
It might be necessary to assign attributes only to one specific
reaction. With the example above one might want to create an attribute only for the reaction with Alanine but not with Proline.
In this case use the member in the parentheses as a suffix: $FUNCTION{Alanine}="Metabolism"

See Metannogen brace expansion for more.}

Reaction Attributes

The multi-line text field contains free text which does not to be structured in a certain way. Usually, it is not possible to include information from this text-field in the exported model file. However, content that is captured in form of a variable will be available for further data processing. Add a simple variable declaration in the multi-line text field. If you know programming languages like or you will be familiar with the syntax.

        $THE_FUNCTION=break-down

You can use any variable name starting with a letter and consisting of letters, digits and underscore. It is recommended to register variable names in Menu-bar>Customize>Variables. The advantage is that registered variables are color-highlighted. This allows one to notice when variables are mistyped. To refer only to the reaction in one compartment add a compartment suffix:

        $THE_FUNCTION@cyto=break-down

or several compartments separated by comma

        $THE_FUNCTION@cyto,endo=break-down

For text containing white space use quotes:

        $THE_FUNCTION@cyto="break  down"

Obtain the list of reaction objects of the dataset and observe what reaction[s] the attributes are assigned to. The variable content can be used for the data export. See Tutorial SBML-Export.

Export as SBML

Finally, use the "File" menu in the Metannogen's menu-bar to export the reconstructed network as SBML.