Plugin - New Network Creation

[Dario Pirola]

Hi!
I've been trying to create a new network from a plugin, using these
lines:

CyNetwork net = Cytoscape.createNetwork("imported network");
int nodeIndices = new int[resNumber];
int edgeIndices = new int[edgNumber];
for (int i = resNumber; i > 0 ;--i){
String nodeName = resList.remove(0).getLocalName();
CyNode cytoNode = Cytoscape.getCyNode(nodeName, true);
nodeIndices[0] = cytoNode.getRootGraphIndex();
cytoNode.setIdentifier(nodeName);
}
net = Cytoscape.getRootGraph().createNetwork(nodeIndices,
edgeIndices);

Problem is, in Cytoscape I get only a window rightly named "imported
network", with no nodes/edges inside.
What's missing in the code to see nodes/edges?

thank you very much

Dario

[Peng-Liang Wang]

Hi Dario

 

There is a Plugin developer's tutorial http://cytoscape.org/cgi-bin/moin.cgi/plugin_developer_tutorial

You may take a look at Tutorial#4 "How to create, modify, and destroy a network, nodes, and edges?".

 

Peng

[Steven J. Bark]

Hi.

I'm new to Cytoscape and have been going through the online tutorials. I
have a few questions.

1. In the Import Network from Database system in Cytoscape, I searched
for a protein called clusterin. I observed several human networks that
were "positive" for this protein and I loaded several. However, when
looking at these networks, I could not find clusterin in any of them. Is
there a way to find the protein you are searching for in a more direct
manner in a network?

2. I am looking for a way to map a lot of proteins observed in proteomics
experiments onto networks. I want to find the relationships between the
proteins. Is there a way to do that in Cytoscape (and if not, it would be
of extreme usefulness to proteomics science!)?

Thanks.

Steve

Steven J. Bark, Ph.D.
University of California, San Diego
Skaggs School of Pharmacy and Pharmaceutical Sciences
9500 Gilman Drive, MC 0744
La Jolla, CA 92093-0744
(858)822-6685
sb...@ucsd.edu

[Keiichiro Ono]

Hi.

> 1. In the Import Network from Database system in Cytoscape, I searched
> for a protein called clusterin. I observed several human networks that
> were "positive" for this protein and I loaded several. However, when
> looking at these networks, I could not find clusterin in any of them. Is
> there a way to find the protein you are searching for in a more direct
> manner in a network?

I assume you use Pathway Commons Client. When you search "clusterin"
for human, it returns several candidates (genes). For each
candidates, Pathway Commons client displays related pathways (from
curated databases) and interactions (usually just a collection of
binary interactions). I can see several results if I retrieve
interactions for CLUAP1.

If you want to search through the entire network, I recommend to use
EnhancedSearch plugin (see the attached screenshot).

> 2. I am looking for a way to map a lot of proteins observed in proteomics
> experiments onto networks. I want to find the relationships between the
> proteins. Is there a way to do that in Cytoscape (and if not, it would be

In this case, what's the input? Is it a list of gene names? If so,
do you want to import known interactions/pathways for those
genes/proteins?

Thanks.
Kei

> of extreme usefulness to proteomics science!)?
>
> Thanks.
>
> Steve
>
>
> Steven J. Bark, Ph.D.
> University of California, San Diego
> Skaggs School of Pharmacy and Pharmaceutical Sciences
> 9500 Gilman Drive, MC 0744
> La Jolla, CA 92093-0744
> (858)822-6685
> sb...@ucsd.edu
>
> >
>

--
Keiichiro Ono ko...@ucsd.edu

Cytoscape Core Developer Team: http://www.cytoscape.org/
UCSD Bioengineering Ideker Lab: http://chianti.ucsd.edu/idekerlab/

[Steven J. Bark]

On the first issue, I want to search entire networks. I will look at the
EnhancedSearch Plugin and see if this will solve the problem.

On the second issue, my data consists of protein itentifiers such as gene
symbols from IPI database searches. My objective is to map this set of
protein identifiers to known pathways. This will "cluster" the identified
proteins in my data into groups involved in particular networks or
pathways. This would be very powerful at focusing attention on functional
groups of proteins.

I am progressing through the advanced tutorials. I can see other
important contributions from other plugins like BinGo, MCODE and
JActiveModules.

Thanks for your help!

Steve

[allan_k...@agilent.com]

If you are using gene symbols, then you could also use the Agilent Literature Search and/or MiMI plugins to build networks. The Agilent tool builds networks from associations extracted from text, the MiMI tool builds networks by extracting information from multiple well-known protein interaction databases.

AllanK

______________________________

Allan Kuchinsky
Agilent Technologies
5301 Stevens Creek Blvd. Mailstop 54U/SC
Santa Clara, CA, 95051
* phone (408) 553-2423
* mailto:allan_k...@agilent.com

[Keiichiro Ono]

Hi.

> On the first issue, I want to search entire networks. I will look at the
> EnhancedSearch Plugin and see if this will solve the problem.

If you mean search multiple networks, you need to merge them first
because EnhancedSearch or other search/filter function supports one
network at a time. So if you want to search genes in multiple
networks/pathways, you need to make a one big network from them.

> On the second issue, my data consists of protein itentifiers such as gene
> symbols from IPI database searches. My objective is to map this set of
> protein identifiers to known pathways. This will "cluster" the identified
> proteins in my data into groups involved in particular networks or
> pathways. This would be very powerful at focusing attention on functional
> groups of proteins.

I see. Currently, Cytoscape supports one data source (interaction
database) at a time, so the following is for NCBI Entrez Gene.

1. Install NCBIClient and NCBIClientUI plugins
2. File-->Import-->Network from Web Service
3. Select NCBI as data source
4. Enter search term. This uses exact same syntax as Entrez Gene web site uses:
http://www.ncbi.nlm.nih.gov/sites/entrez?db=gene
See sample searches. Here, I typed "clusterin AND human[ORGN]" I
specifyed only one gene in this example, but you can enter list of
genes instead.
5. Import the interaction network. This takes while...
6. File-->Import-->Import Attributes from NCBI Entrez Gene...
7. Check annotations you want from NCBI Gene database. "Pathway"
option imports all known pathways in KEGG and Reactome.
7. Import. This takes several minutes depends on your network connection.
8. Use the imported attributes to visualize data. In my example,
genes on known pathways are colored, otherwise they are gray.

The resulting visual is something like the attached image.

Basically, you can start from both list of genes/protein IDs or free
keyword. This is a common usecase, so I'll put this somewhere on the
wiki.

Thanks.
Kei

[Keiichiro Ono]

By the way, the example I posted starts from interaction networks then
map known pathways onto it. However, you can also extend this by
importing those pathways and then merge interaction networks and
pathway diagram. This is a bit complicated example, so I'll post it
later.

Thanks.
Kei

2008/10/8 Keiichiro Ono <ko...@ucsd.edu>:

[Steven J. Bark]

How about if you don't know an interaction network, but only know the
identity of the proteins. How would you map these protein identities onto
a known pathway?

Steve

[Steven J. Bark]

Thanks for this information.

I tried to find the NCBIClientUI plugin, but could not. I found the
NCBIClient v0.6 and installed that. Where is the NCBIClientUI plugin? I
also went to Import Network from Database window and installed the extra
web service clients. This installed the IntAct Service Client.

I will wait for you reply on where to get the NCBIClientUI plugin before
trying your method.

Thanks!

Steve

[Steven J. Bark]

I have run into another error. This time, used the Import Network from
Database window, searched the IntAct database and searched for some
proteins. It found 12 records for CPE (carboxypeptidase E), but when I
clicked "yes" to create new network from the search result, I get an error
"Failed to load network from web service." The error details are

Root Error
Caused by: java.lang.IndexOutOfBoundsException: Index:1, Size: 1

[Keiichiro Ono]

Sorry, the actual plugin name is NCBIEntrezGeneUserInterface. It is
available from Plugin Manager.

Kei

2008/10/9 Steven J. Bark <sb...@ucsd.edu>:

[Keiichiro Ono]

Hi.
I'm not sure this is a bug or network issue, but I'll track down the
problem today.

Thanks.
Kei

2008/10/9 Steven J. Bark <sb...@ucsd.edu>:
>

--

[Keiichiro Ono]

Hi. I found the cause. The Intact web service made a minor change to
their return data type, so I'll fix the client to accept the new data
format. I'll release the new version within an hour.

Thanks.
Kei

2008/10/9 Keiichiro Ono <ko...@ucsd.edu>:

[Steven J. Bark]

I have tried to update my Cytoscape, but it doesn't recognize a new
version for the IntAct web services plugin.

Is it available yet?

Steve

[Keiichiro Ono]

If you cannot see IntAct version 0.60 from Plugin Manager, uninstall
the old one and try again.

[Steven J. Bark]

1. The update indicates that I already have version 0.60. Therefore, it
won't update.

2. I can't figure out how to uninstall the webservices client. I
performed a complete uninstall of Cytoscape. After reinstalling the
program, I am still unable to use the IntAct webservices browser. Where
is this client located as it is not affected by uninstallation of the
entire Cytoscape program?

3. IntAct Webservices still will not load a network.

HEEEEEEEEEELP!!!!

Steve

[Steven J. Bark]

I figured out that uninstall can be done using the manager. I am trying
that now and will report the results.

Steve

[Steven J. Bark]

The manager delete worked. I found that the webservices 1.1 may be the
problem. It may be that any IntAct webserves appear already installed
from that 1.1 version, even if they are not. I did not install from the
webservices 1.1 and manually installed the IntAct WebServices Client 0.6.

Steve

[Keiichiro Ono]

Hi.
Seems current Web Service client pack is a bit confusing, so I'll try
to create more up-to-date client packs.

Thanks.
Kei

2008/10/10 Steven J. Bark <sb...@ucsd.edu>: