Barbara Mirel

NCIBI, University of Michigan, USA

Biological Use Case: Extend knowledge about binding partners of a protein of your choice. In this case the protein BDNF having a role in neurotrophic disease.


Cytoscape version: 2.6

Plugins to Load: MiMiPlugin

GUI steps:




1. Today we will demonstrate how to interact with the MiMI database. This mechanism is a Discovery approach starting from a single protein of interest to explore other possible protein interactions

Bring up Cytoscape window, pre-loaded with TCF7L2

1.a.i. We will show how to use the Mimi Plug-in for Cytoscape to retrieve protein interactions, and to visually display them. We will highlight certain features of the Mimi Plug-in. Firstly input search parameters

Search parameters for single gene and file

1.a.ii. Secondly show the text panel for nodes

CTRL A to select all nodes. Open the text panel and select: Description, Gene, Other Gene Names, process, function, and Pathways

Show that columns can be switched and be sorted on

1.b.i. We will show the seamless integration between Cytoscape and the SAGA tool. SAGA is a subgraph approximate matching tool written by Yuanyuan Tian in the laboratory of Dr. Jignesh Patel.

Select ALL nodes (CTRL+ALT+A) Right Click and select SAGA. Show the scoring and matches between the submitted network and the KEGG Matching pathways

Blue lines are matching relationships, Red Lines are matching proteins

1.b.ii. Show how this links to the annotated KEGG Pathway

by clicking on the “Link to KEGG Picture” at the top of the match. Show how the matches to your submitted proteins align to this picture.

Tian Y, McEachin RC, Santos C, States DJ, Patel JM. SAGA: a subgraph matching tool for biological graphs. Bioinformatics. 2007; 23(2):232-9.

1.c. Edge information description

All nodes and edges should still be selected when you return to Cytoscape window. Click on the Edge Attribute tab at bottom and choose attributes: Genename, Provenance and Type. Show how to sort, and how to identify in the image which edge you’ve selected

1.d. We will show how the Mimi provenance is visible through the browser

Discuss MIMI Provenance here, and show NLP Only attributed relationships.

1.e. We will show you the integration of the parsed PubMed abstract literature data (bioNLP) , and how to connect out to pubmed for the abstract information if needed.

Returning to a single edge, we select PARP1 and TCF7L2 edge and right click to get BioNLP data. Discuss 3 different Abstracts and ability to link out. Mention MEAD summarization for large numbers of sentences

1.f. Finally how to launch the MimiWeb browser tool directly from the Cytoscape nodes

Return to Cytoscape and double click a single node CTNNB1 to launch MimiWeb

2. From the MiMI Web browser we can search by Concept or individual protein

Open MiMIWeb initial search page (preset blank page).Search for “Neurotrophic”. Discuss that from this page we can see: interactions, literature references, launch Cytoscape or Gene2Mesh, directly from line. Click in BDNF to see summary page

3. From a single protein summary page, we look at the protein interactions, and using the integrated tools, we can examine several different ways to look at protein protein interactions and relationships.

Review summary page information on protein interactions, GO terms, pathways, etc.

3.a. GIN – Gene Interaction Network. This is a NLP only summarization of the literature and tagged sentences which identify protein-protein interaction

Click on GIN and discuss the summarized nature of GIN interactions

3.b. MiSearch – uses the term you are searching to initiate a pubmed search which will organize the output order to focus on search criteria being used

Show the MiSearch link. Discuss that it is organized and displayed based on your other search criteria (profile of search pattern). Show how you can refine the search by adding authors, and show how this can link directly to Gene2Mesh, through mesh terms. Click on Brain Derived Neurotrophic Factor

May need to preload this for BDNF – running slow

3.c. Gene2Mesh – all pubmed abstracts were tagged for genes, and curated mesh terms associated with that abstract. An analysis of this reveal genes which are associated with Mesh Terms and vice versa. Fischer’s exact test was used to rank order those genes most highly correlated with mesh headings, relative to how the gene is associated with other mesh terms.

Gene2Mesh is now sorted on all genes tagged (scored) as being in articles with the mesh heading above. To pivot this search, click on a gene (BDNF) and show what other mesh headings are associated with that gene. Show that these are searchable from the MiMI summary pages as well

3.d. Additonally from the MiMI summary page we can look into MIMI itself and display the nearest neighbors to a protein using a Network Browser designed using FLASH technology. It is a portable alternative to Cytoscape and we will be working with the Cytoscape developers to add this as an alternative.

Back on the mimi summary page for BDNF, click on NetBrowser. This shows a small network of genes (from MiMI) which interact with BDNF

Come back to this after Biosearch2D to demonstrate more functions

4. As a different view into protein-protein interaction, we can start a search from a literature term rather than a protein. BioSearch2D was recently developed, and it uses a search term to identify all the abstracts in Pubmed which are related, and extracts both Mesh Terms and Gene IDs from the literature search. This is displayed in a 2 dimensional co-occurrence HeatMap. We can search for genes, highlight subsets of the map, and since our tools are all designed to be integrated, we can identify a small region of interest and ask what pathways are in that area by sending this list of genes to SAGA, and displaying them in the FLASH Network Browser

Need to have Biosearch2D pre-launched. Enter search term, “Prostate Cancer Epigenetics” - Show the heatmap and describe what we are seeing - Search for EZH2 and highlight small region for export - Show export button to show literature and gene lists - Demonstrate export to Excel button. - Send the network to Mimi Gene Browser

5. The Network Browser has some advanced features which we have just recently developed and we will be expanding on in the coming year. We can add or remove nodes (based on MiMI protein-protein interaction), and further interact with SAGA in this tool as well

Demonstrate ability to add nodes, expand and remove. Show how to highlight nodes and send them to SAGA. Return to Presentation and summary graphic

Data / Session Files: Attach (preferably) a session file or else the data files used in the workflow

Presentation: Attach your presentation

Webstart: Cytoscape with NCIBI plugins loaded (incl MiMiPlugin)


  1. Presentation related to the above recipe: Covers Mimi website, BioSearch2D, MiSearch, MimiPlugin. This is a longer presentation; 1h 22mins.

  2. Mimi plugin presentation by David States: Takes 8.5 minutes.

MiMiPpiDiscoveryRecipe (last edited 2009-02-12 01:03:01 by localhost)

Funding for Cytoscape is provided by a federal grant from the U.S. National Institute of General Medical Sciences (NIGMS) of the Na tional Institutes of Health (NIH) under award number GM070743-01. Corporate funding is provided through a contract from Unilever PLC.

MoinMoin Appliance - Powered by TurnKey Linux