Chapter 3. Tutorial

1.0

Table of Contents

Introduction
Description of the ProViz interface
The menu bar
The tool bar
Right panel
Left panel
Using ProViz with a dataset in Tulip file format
Opening the file
Filtering Data
Saving modifications
Using graph layout
Using ProViz with a dataset in PSI-MI canonical format
Importing a PSI-mi canonical file
Saving modifications
Conclusion

Introduction

This tutorial aims to provide a quick tour of available ProViz functionalities. We will first show you these functionalities through an example using a dataset stored in a Tulip file. Then, we will do similar exploration but with a dataset stored in a PSI canonical file.

Description of the ProViz interface

As you can see on the figure 1, ProViz interface can be splitted into four major parts :

  • Menu bar
  • Tool bar
  • Right panel
  • Left panel

Figure 1 : ProViz main interface

The menu bar

The menu bar is composed of three menus (figure 2) to access the global functionalities of ProViz. The first menu (``File'') enables you to open, import, save or print graphs. It also gives an option to exit from ProViz and list the last opened files. The ``View'' menu proposes to center the graph view or apply a layout algorythm on your graph. The last menu provides classical help functionalities.

Figure 2 : ProViz menu and tool bars

The tool bar

The tool bar provides mostly used functionalities of ProViz, like layout algorythms, zoom box and center view functions, contextual help (``what's this?'' button) and open/save buttons (figure 2). A cursor like button is also available to switch between zoom box and selection modes.

Right panel

The right panel is dedicated to graph visualization. The top part of the panel is the visualization of the graph and the botom part provides clone/close view options and scale nodes (enables you to increase or decrease nodes size) (figure 3).

Figure 3 : ProViz right panel

Left panel

The left panel is composed of four tabs. The first one (called ``Views'') displays an history of the graph views you have created (figure 4). It is the first panel displayed when you load a graph. On this panel, you can also change the background color of the graph and choose whether nodes labels should be displayed or not.

Figure 4 : Views panel

The second tab (called ``Protein Ontology'') provides nodes filtering functions (figure 5), based on the Gene Ontology terms linked to the nodes and on the evidences that have lead to the association of the GO term with the node. This panel displays the tree that stands for the Gene Ontology Tree. You can find Gene ontology terms in the tree using the search function (regexp search). You can also use the expand/collapse all function (``+/- all'' button) that acts on the Gene Ontology tree. At last, three buttons at the bottom of the page enable you to see the result of your filtering. The first one (``Highlight'') displays selected nodes and edges with a sligthly transparent purple background. The second (``View'') displays your selection in the same window (hiding non selected nodes/edges) and ask you to name the view. The last one (``New View'') will open a new tab in the right panel and ask you to name it. As a result you will obtain a new tab displaying your selection.

Figure 5 : Protein ontology panel

The third tab (figure 6) provides filtering functions over the interaction Controlled Vocabulary(CV). This ontology is based on PSI Controlled Vocabulary and allow interaction filtering according to the experimental methods used for their detection. It provides the same selection functionalities as the previous tab.

Figure 6 : Interactions ontology panel

The last tab (figure 7) enables you to visualize properties of proteins and interactions, and to add comments on them. There is also a function to search for a protein according to some criteria. You can either search with a word in the definition, the name or other properties. Once found, each protein/interaction can be selected and appears as surrounded by a slight purple circle. The link nox provide an access to external references of proteins and interactions (Swiss-Prot, Gene Ontology, IntAct database, etc...). Clicking on a link open the link in Mozilla (if you hae mozilla installed) or print the complete link in your terminal window (this function will be soon improved).

Figure 7 : Properties panel

Using ProViz with a dataset in Tulip file format

ProViz is based on the Tulip software that performs highly interactive graph drawing and exploration. Tulip has its own file format that is used as the first format in ProViz. We provide two datasets in Tulip format that contains Yeast Two-Hybrid data (y2hAnnotated.tlp.gz) and Cellzome data (complexes.tlp.gz). As you might have noticed with the filenames cited below, ProViz can transparently handle file compressed with gzip. This is usefull since text files can be highly compressed, so that they can be exchanged over the network more efficiently. In this example, we will use the first file (y2hAnnotated.tlp.gz), but you can apply the same protocol to the other file (obviously, you won't obtain the same visualization results).

Opening the file

To load the file, just click on the folder icon or either on the ``File'' menu and choose the ``open'' option (figure 8). A window should pop up and ask you to choose a file in .tlp or .tlp.gz format (figure 9). Just browse your hard drive to find your file and then click on the ``open'' button.

ProViz then computes the Tulip content of the file and displays the graph. Your window should look like the figure 1.

Figure 8 : Opening a Tulip file

Figure 9 : Finding a Tulip file

Filtering Data

Just switch to the ``Protein Ontology'' tab and expand the Gene Ontology tree by clicking on the ``+/- all'' button. You can see all the GO terms displayed in three different colors (figure 6). To know the semantic of these colors, click on the contextual help icon in the tool bar (``what's this?'' icon).

Then, search for the ``RNA-nucleus export'' GO terms by typing ``RNA*'' in the search box (* is a wildcard so that you can find any terms that begin with RNA, in a case insensible search mode) and hiting next until you reach the desired term. Select this term by clicking on it. As you can see, all childs have been selected. To refine the selection, click on r, sn and tRNA-nucleus to unselect them.

Choose ``Experimental Support'' as an evidence filter (at the bottom of the left panel), click on the ``New View'' button and give a name to this new view when asked to do so. You can notice that a new tab has just appeared on the top of the right panel, with the name you gave as title (figure 10). The focus is now on this tab and you can apply a layout to obtain a better visualization. Click on the button to apply the GEM layout and then click on the ``center view'' button to obtain the best view.

Figure 10 : Filtered graph

In this graph, you can see two major complexes, each one composed by four partners. The first one is composed by PSE1, NUMP57, NSP1 and NUP49 (I) and the second one by NUP53, NIC96, POM152 and NUP157 (II). So click on the ``properties'' tab at the top of the left panel and then click on each node of complexes I and II to check protein properties (figure 11).

Figure 11 : Adding a comment on a protein

As expected, each protein annotation indicates an implication in a nuclear pore, means that these proteins and protein complexes are likely to be involved in the transport of RNA from the nucleus to the cytosol.

You may also notice that PSE1 is annotated as Karyopherin and that NUP53 is annotated as Karyopherin docking complex component. So, these two proteins might be in interaction via another protein that has not the chosen GO term (since it is not in the graph). We want to remember this fact, so we put an annotation on each protein : PSE1 likely to be in interaction with NUP53 and vice versa. Just write this comment in the ``comment box'' in the properties panel (figure 11). If you choose another node, check its properties and then go back to NUP53, you can see that your comment has been kept. This comment will be stored in your file if you save it.

Saving modifications

So you have explored the graph, maybe found interesting informations and put comments onto proteins and interactions, and you want to keep your work without overwriting the original file. So, in the ``File'' menu, choose the ``Save As'' option (figure 8) and put a name in the appropriate field (myFile.tlp for a normal tlp file, myFile.tlp.gz for a compressed one). You are now working on this saved file, without any risk to overwrite the original one.

Using graph layout

Lets imagine you have found an amazing relationship between TFB1, TFA1 and TFA2 (surprising, isn't it?) and that you want to publish an article about this, with a pretty image of this complex. First of all, lets isolate this complex from the entire network. Go on the root view and then, in the ``protein ontology'' panel, search for ``transcription* *regulator*'' and select it when found.

Then click on the ``new view'' button and center the view on the complex by zooming on it, either with the wheel of your mouse or with the zoom box. For this last option, click on the ``zoombox'' button in the tool bar and draw a selecting box around the complex. Then, click on the ``selection cursor'' in the toolbar (figure 12).

Figure 12 : Zooming with the zoom box

You now have a view of the complex and can save it as a PNG file or print it. If the background color is not your favorite one, you can change it in the ``Views'' tab of the left panel. Click on the ``Background color'' button and choose your color in the pop-up window, then validate by clicking on ``ok''.

Using ProViz with a dataset in PSI-MI canonical format

The Proteomic Standards Initiativeaims to define common exchange formats for proteomic data. PSI format for PPIN is named PSI-MI and can be found in two forms. The canonical form (or normalized/compact) is more appropriated for computer uses. The expanded form (denormalized) is more human readable. To convert one form into the other, you can use XSLT stylesheets provided by the PSI group.

Many protein interaction databases (BIND, DIP, MINT) have chosen to deliver their data in PSI-MI format, so that most of PPIN data will be soon available in this format. Moreover, the IntAct project is currently developping functionalities to retrieve the results of queries in IntAct as PSI-MI files.

So, we have developped a plugin that enables you to import PSI-MI canonical files into ProViz. ``Classical'' \PSI files can be handled without any problem, but to have more functionalities available in ProViz, some informations should be in the PSI-MI file (GO terms and evidences for each interactor, controlled vocabulary on edges, distance between start nodes and other nodes, etc...).

Lets try an example with the PSI-MI canonical file yeast_normalized.xml, which is based on a data sample available on the CVS repository of the PSI-mi project on sourceforge. The second file is a modified version of the first one, with informations added (Go terms and evidences). You can find three other PSI-MI canonical files along with ProViz.

Importing a PSI-mi canonical file

Please see the section on Graph Extraction Policies and Raw Graph Extraction to simply import a PSI-MI file into a graph.

Just switch to the ``Protein Ontology'' tab and expand the Gene Ontology tree by clicking on the ``+/- all'' button. You can see that all GO terms seem to be disabled. Actually they are, since the yeast_normalized.xml file do not contain any GO term. To provide a better example, we have added GO terms and evidences to this file and renamed it to yeastAnnotated.xml.gz. Now, import this file with the same protocol as described above. You can see now all the GO terms displayed in three different colors (see figigure 6).

Then, search for the ``RNA-nucleus export'' GO terms by typing ``RNA*'' in the search box and hiting next until you reach the desired term. Select this term by clicking on it. As you can see, all childs have been selected. To refine the selection, click on r, sn and tRNA-nucleus to unselect them.

Choose ``Experimental Support'' as an evidence filter (at the bottom of the left panel), click on the ``New View'' button and give a name to this new view when asked to do so. You can notice that a new tab has just appeared on the top of the right panel, with the name you gave as title (figure 13). The focus is now on this tab and you can apply a layout to obtain a better visualization. Click on the button to apply the GEM layout and then click on the ``center view'' button to obtain the best view.

Figure 13 : Results of the first filtering on GO terms and evidences

In this graph, you can see one complexe composed by NPR3, RLR1 and YDL175C. So click on the ``properties'' tab at the top of the left panel and then click on each node to check protein properties.

As expected, each protein annotation indicates an implication in a nuclear pore, means that these proteins and protein complexes are likely to be involved in the transport of RNA from the nucleus to the cytosol.

No other crucial information is available for this view of the graph. As you may have noticed, these proteins where not found in the previous dataset with the same filtering. The first dataset (y2hAnnotated.tlp.gz) contains Yeast Two Hybrid data and this dataset (yeastAnnotated.xml.gz) contains Yeast PPIN detected by Liquid Chromatography following by Tandem Mass spectrometry (LC-MS/MS).

Saving modifications

So you have explored the graph, maybe found interesting informations and put comments onto proteins and interactions, and you want to keep your work. You can save your work in the Tulip file format. Open the ``File'' menu, choose the ``Save'' option (figure 8) and put a name in the appropriate field (myFile.tlp for a normal tlp file, myFile.tlp.gz for a compressed one). You are now working on this saved file.

Conclusion

We got an overview of ProViz functionalities and we have described a general schema for searching in a PPIN. You can load PPIN stored within a Tulip file or PSI-MI canonical file, you can choose layouts to obtain a better view of the network and are able to perform several types of search.

The most important fact to remind from this tutorial is that the more informations you have in your files, the more ProViz is able to give you interesting results. You can try to demonstrate that fact by loading and exploring the two yeast PSI files (yeast_normalized and yeastAnnotated) and examining the functionalities that you can use or not. Several enhancements are planned for the next version and are described in the ``TODO'' file delivered with ProViz distribution, but you are encouraged to send any comment or feature request.