Exercises in ChemProt and network pharmacology

During this session you will have hands-on training on our new web-server, ChemProt. ChemProt gives you bioactivity information associated to small compounds. One specificity of ChemProt is that you can get biological information for a protein when is part of disease complexes.

The second part of the exercise will focus on the network pharmacology profile of a specific drug compound, based on protein-protein interaction data extracted from the InWeb interactome network of CBS. The networks will be visualized and analyzed using Cytoscape. The query drug is citalopram, which is a well-known antidepressant defined as SSRI (Selective Serotonin-Reuptake Inhibitor).

ChemProt Server

⇒  Go to the ChemProt server

http://www.cbs.dtu.dk/services/ChemProt/

⇒  Type Citalopram in the "Type a compound name:" field or paste in the SMILES for the Citalopram in the text box (1).

       Click on Submit (2).

• By moving the cursor on the Compound ID (in pink) you can see the 2D structure and general physico-chemical properties associated to the compound.

• By clicking on Disease Complexes (last column) you will be directed to a server that takes the protein target as input and retrieves its protein-protein interaction complexes. If you click on a complex, you can see both the schematic view of the network as well as relevant biological information from a variety of sources (OMIM, GO terms, HPA,...).

Questions

Q1: How many proteins associated to Citalopram?
       Hint: You can download the output file and sort the information by proteins in Excel

Click on one disease complex associated to a protein target (e.g. Human Dopamine Receptor D4).

Q2: What are the diseases associated to this target?

Q3: In which tissues is its protein-protein interaction complex essentially expressed?

Network pharmacology of citalopram

You need to download and install this version of Cytoscape 2.6.3 to get the plugins to work.

In the second part of the exercise, we will use Cytoscape to visualize and analyze the network pharmacology of the antidepressant citalopram.

We gathered from the literature 49 genes annotated to citalopram and used them as queries to form a protein-protein interaction network. The output network consists of 629 proteins with 4141 interactions.

First, you need to download the following plugins for Cytoscape:

• NetworkAnalyzer.zip (Download and unzip it and save the all 24 JAR-files in your plugin-directory.
  

      Mac OS X /Applications/Cytoscape_v2.6.0/plugins
      Windows XP C:/Program Files/Cytoscape_2.6.0/plugins)

• MCODE v.1.31 (install from Cytoscape)

• jActiveModules.jar v. 2.23 (install from Cytoscape)

The two last plugins can be downloaded directly from Cytoscape: (Plugins ⇒ Manage Plugins ⇒ Analysis)

Import network

To visualize the citalopram network, open Cytoscape and load the file citalo_ppi.cys by selecting from the menu: File--> Open --> citalo_ppi.cys.

Network layout and nodes selection

You can try different layouts (circular, organic, hierarchical and random) by selecting the appropriate layout in the "yFiles" under Layout. One of the most common layouts for network biology is the organic layout (similar to the spring layout).

In the main Cytoscape window (with the network view) you can select and move nodes by clicking and dragging them with the left mouse button.

Each node has an ID label. The node representing serotonin transporter is number 14424. Instead of searching by ID we can change the search option by clicking on:

• Configure search option (1)
• click on dropdown menu (2)
• change to Label (3)
• click on Apply (4)

Questions

• Q4: How many proteins interact directly with SERT (SLC6A4)?

• Q5: Have these proteins been already associated to SERT in the literature? (search in PubMed)

• Q6: Have they been associated to citalopram?

Network statistics

In this part you will work with the NetworkAnalyzer plugin. First deselect all nodes in the network (just click somewhere on the white space). Then apply the NetworkAnalyzer plugin to the network by selecting Network Analysis from the Plugins menu followed by Analyze Network (treat the network as undirected). Ignore the update version of Networkanalyzer (press OK). This should open a "Simple Parameters" window with some statistics. Browse through the various network statistics/parameters and try to answer the following questions:

• Q7: What is the average degree (connectivity) of the network?

• Q8: What is the network diameter, radius and density?

NetworkAnalyzer Online Help

Identification of clusters

As the average cluster coefficient is relatively high, we expect that some clusters in the network are formed. We will try to identify clusters using the MCODE algorithm.

To start MCODE: Plugins--> MCODE --> Start MCODE. At the bottom of the MCODE panel (bottom left), click Analyze.

This will identify several complexes. By clicking on complex 1, the complex will be highlighted in yellow on the network

• Q9: Is SERT or a protein neighborhood of SERT part of the cluster?

• Q10: How many proteins are included in this cluster?

• Q11: After both exercises, what do you think about the definition of citalopram as a SSRI (Selective Serotonin-Reuptake Inhibitor)?