MATISSE User Manual

Installation & Execution

Prior to installation please make sure that Java JRE version 5.0 or higher is installed at your computer. Java JRE is available here. Unzip the installation file into a directory of your choice. Use MATISSE.bat to activate the program. If your computer has more than 2GB RAM, use MATISSE_2GB.bat.

Getting started

Upon opening MATISSE, two options are available - to start a new session or to load an existing session.

If you select to start a new session, a start dialog is shown that allows you to load some data into the program. The initial specification includes the selection of the species and an option to load one interaction network and one expression dataset:


The first step is species selection, which is precondition for the other specifications. Currently the following species are supported:

Note that the species selection affects the main identifier that should be loaded into the program in all the data files. For S. cerevisiae the expected identifiers are ORF names. For all other organisms MATISSE expects
Entrez Gene identifiers. It is also possible to select "Other species", in which case any identifiers can be used, but no additional information will be available for the genes.

For network loading the following options are available:

You can also choose to skip loading the network right away and to load a network in a subsequent stage using the Load network command in the Network menu. In this case the program will be initialized with an empty network.

Expression data can be loaded either in the start dialog or afterwards using the Load expression data option in the Expression menu. For expression loading the following options are available:

For expression data, note that if you intend to use the MATISSE algorithm, it is best to load data which is most suitable for computing Pearson correlations between gene patterns. If possible, it is best to provide log-ratios data. If the original data is Affymetrix or Illumina normalized ratio, they should be log-transformed (which can be done using the Expression->Transform dataset menu). Missing values can be estimated in MATISSE using the Expression menu.

Finding new modules

In order to find new modules using network and/or expression data use the button in the main tool bar or the Find modules option in the File menu. Some algorithms are only available when network/expression data are properly loaded. Note that algorithms utilizing both network and expression data will only work if both data are loaded with matching identifiers (preferably with the default identifiers for the species in question, see above). Currently the following options are available for finding modules:


    The MATISSE algorithm is relevant when both expression and network data are available. The following options can be adjusted:

    Notes on using the MATISSE algorithm:


    The DEGAS algorithm is described here. It is relevant when both expression and network data are available and when interaction confidence values are available. The options are largely similar to those in MATISSE. The only additional parameter is 'Confidence threshold' which specifies the q parameter of CEZANNE - the minimal confidence that at least one edge connects any two parts of a reported module.

  3. DEGAS

    The DEGAS algorithm is described here and it is relevant when both expression and network data are available and the expression dataset compares groups of hetherogenous samples (as in case-control studies). Therefore it is possible to execute DEGAS only once at least one sample parameter with at least one value has been defined (using Expression->Define sample parameters menu). The following options can be adjusted:

  4. deMATISSE

    The deMATISSE algorithm is described here. It is relevant when you are interested in identifying subnetworks of genes that are correlated with each other and also with some sample parameter. Therefore it is possible to execute DEGAS only once at least one sample parameter with at least one value has been defined (using Expression->Define sample parameters menu).

    Most of the options are similar to those of MATISSE. The parameters specific to deMATISSE are:

  5. Custom MATISSE

    This is a version of MATISSE in which the user can supplied a custom weight matrix (which should have both positive and negative weights).

    Most of the options are similar to those of MATISSE. The only specific parameter is Weight matrix file that allows the user to specify the location of the weight matrix. The format of this file is very simple - it is a tab delimited table file, where the first row and column should correspond to gene identifiers (EntrezGene), or any other identifiers that match to the network nodes. A sample weight matrix file can be found here. All the genes which appear in the matrix will be considered as front nodes.

  6. Co-clustering

    The Co-clustering algorithm for simultaneous clustering of network and expression data is relevant when both of these data types are available. The following parameters can be adjusted:

  7. Expression k-means

    The k-means clustering algorithm (applied to expression data). The following parameters can be adjusted:

  8. Expression t-test

    This option applies the t-test to one of the loaded expression datasets. The following options are available:

  9. CAST network clustering

    This CAST algorithm for network clustering applied to the loaded network. The following options can be adjusted:

When all the options are adjusted use the button to execute the module finder. After the module finding is over, a module set will be produced and the modules will be listed in a table in the upper left corner of the screen

Analyzing results

Every module set produced by a module finder or imported from an external file is represented within the program as a module set. The environment can maintain multiple module sets, but only one module set is the active module set at any given time. The active module set is selected in a drop-down list in the main toolbar. It is possible to add additional empty module sets using the Create new module set->Empty module set option in the Module set menu.

Once a module set is selected more options of the environment become available. This is an outline of the main window of the environment:


The main screen features available for analyzing modules are:

Module subnetworks area - In this area, module windows are presented. A window is opened for a module when in is selected in the modules list:

Modules list - in this table, all the modules in the active module set are listed. The table contains three columns: the name of the module, the number of genes it contains and an additional column (module attribute column), the contents of which are controlled by a drop-down list under the module table. This list contains all the module attributes which are available for the active module set. New attributes are added by some of the options which are described later.

By default all the modules in the active module set are listed in the modules list. It is possible to show only a subset of the modules by selecting a module filter from the roll-down menu above the list. The following filters are available:

Selecting a module in the table makes it the active module and opens a corresponding module subnetwork frame in the Module subnetworks area.

Enrichments list - in this list its is possible to view different enrichments of the active module in two modes - the TANGO mode and the Annotation mode.

Sequence elements list - This list is currently not supported in the public version of MATISSE.

Bottom pane - In this pane two unrelated panels can be presented:

Genes list - This table contains a list of all the genes in the active module. The list can be sorted by the names of the genes in ascending/descending order by clicking the header of the table. Selecting a gene in the list paints it in white and scrolls the active module subnetwork window such that the corresponding gene becomes visible.

The module subnetwork window

After a module is selected in the modules list its subnetwork window opens in the module subnetworks area. This window presents all the genes (nodes) in the active module and the interactions between them.

The layout of the window is dynamic and it is possible to drag the nodes of the graph using the mouse left button. It is also possible to perform various layout operation, including automatic layout, from the Layout menu.

Note that directed edges generally correspond to protein-DNA interactions, and undirected ones to other types of interactions. Genetic interactions are denoted by green dashed lines.


Additional operations are available in a pop-up menu that appears when the canvas is right-clicked:

If a right-click is performed on one of the nodes it is possible to open a link to one of the databases that describe that gene. It is possible to get additional information by positioning the mouse over nodes/edges.

File menu: Saving and loading results

The best way to load/save results is through the "Load session"/"Save session" options in the File menu. Sessions are stored in a single .zip file that contains all the information about the network/expression data/modules/annotations.

In addition, it is possible to save all the active module sets into a simpler and smaller text file by by using the Save all module sets option in the File->Export module set(s) menu. Additional options available in the File menu:

Network menu: Accessing network information

The Network menu has various options for handling network data loaded into the program:

Expression menu: Analyzing expression data

Once a module is selected and expression data is loaded it is possible to view the submatrix representing the expression values of the module's genes using the button in the main toolbar. Details about this view are given here.

The Expression menu has the following options:

Module set menu

The following options are available in the Module set menu:

Module menu

Options are available in the Module menu all refer to operations performed on the genes in the active module.

Annotation menu: annotating the modules

Two types of enrichment analyses of annotations are currently supported: (a) GO annotations using the TANGO algorithm which are corrected for multiple testing and annotation overlaps by random re-sampling; (b) annotations from different annotation databases which are not corrected for annotation overlap etc. (there are called annotations throughout the program and the manual) .

The following options are available in the Annotation menu:

Layout menu

In the Layout menu it is possible to control the layout of the genes in the active module subnetwork window. The following options are available:

Similarity matrix display

This display shows correlations/co-occurances between two groups of objects, or between objects belonging to the same group. One group of objects is presented as the rows of the similarity matrix and the other as the columns. The matrix cells are color-coded based on the similarity values. Blue indicates positive values and red negative ones. The scale of the colors is -1..1 by default, but if the similarity values exceed this region, the thresholds are scaled such that full blue and red colors represent the 90% percentile of the value range.

Using the toolbar of the similarity matrix view it is possible to zoom in/out of the view and to shift the matrix in order to allow a better visibility of the row/column labels..


Expression matrix display

This view presents the expression values of a list of genes (rows) in a series of conditions (columns). The color scale can be set by clicking the color scale in the tool bar. It is also possible to adjust the view (zoom in/out etc.) using the toolbar.


For any questions, contact Igor Ulitsky

Back to the MATISSE homepage