Disease-protein network

[Majid]

Hi,
I have a list of proteins that are associated to several diseases with
different evidence of association (P-value). I like to connect
proteins to diseases using lines that are colored in a heat scale,
according to their P-values (e.g. in a shade of red to blue) also I
want to make the size of disease circle nodes to be proportional to
the number of genes that are connected to them. I wonder, Dose
Cytoscape can be used to create such a graph? I’ll be grateful for
your comments and suggestions.

Kindly regards,
M

[allan_k...@agilent.com]

Yes, you can do this sort of thing in Cytoscape. Importing the network and coloring the edges based upon P-values would be pretty straightforward, but you would need to do a little programming, probably via a small plugin, to make the sizes of the disease nodes proportional to the number of edges they have. This is because you will have two kinds of nodes -- for proteins and diseases, respectively, so the visual mapping is a little more complicated than can be handled in the VizMapper user interface. What you'd want to do is create the network, building edges for each protein-disease pair, with an edge attribute for the p-value. Each node should have a 'node-type' attribute, identifying it as either a 'protein' or a 'disease'. Then you would need to calculate a node attribute, call it 'connectedness', that can be used to create a continuous visual mapping to node size. My suggestion is that if the node-type is 'protein' then set 'connectedness' to 1, if it's a 'disease' then set 'connectedness' to the number of edges for the node. Then you need to create a visual style that has a continuous mapping from 'connectedness' to node size.

There is a plugin developer's tutorial on the Cytoscape wiki at http://cytoscape.wodaklab.org/wiki/plugin_developer_tutorial, which describes how to do some of these things and has example code.

There may be simpler ways to do this, perhaps other people on this list may have suggestions.

Let me know if this is helpful.

AllanK


Allan Kuchinsky
Principal Project Scientist
Molecular Tools  
Agilent Laboratories

Agilent Technologies, Inc.
Santa Clara, CA, USA
408.553.2423
allan_k...@agilent.com

--

You received this message because you are subscribed to the Google Groups "cytoscape-helpdesk" group.
To post to this group, send email to cytoscape...@googlegroups.com.
To unsubscribe from this group, send email to cytoscape-helpd...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/cytoscape-helpdesk?hl=en.

[Russ]

I don't think you need to do any programming to achieve your aims
here.
a)Import your data in such a way that you can select for the disease
nodes within cytoscape.
b)Once in CSC select the disease nodes and perform degree-sorted
circle layout on those nodes only - this will give a degree value for
all your disease nodes, but not for the genenodes
c)Relayout the network however you want
d)In Vizmapper's Visual Mapping Browser double-click on 'NodeSize',
select 'Degree' from the dropdownmenu and set the mapping type to
continuous then you can play about with the graphic so that size
varies linearly with degree.

For edge colours you use the same sort of protocol.

Regards
Russ
U.Liverpool

[allan_k...@agilent.com]

Excellent idea. I learn something new about Cytoscape every day.

Would you have any suggestions about how he might import his data in such a way that he can select for the disease nodes within Cytoscape? I am thinking that he can import the data from a table, where the column that contains the diseases will be the target, then use filters to select all targets. What do you think?

AllanK


Allan Kuchinsky
Principal Project Scientist
Molecular Tools  
Agilent Laboratories

Agilent Technologies, Inc.
Santa Clara, CA, USA
408.553.2423
allan_k...@agilent.com



-----Original Message-----

[Russ]

On Dec 3, 3:17 pm, <allan_kuchin...@agilent.com> wrote:
> Excellent idea.  I learn something new about Cytoscape every day.  
>
> Would you have any suggestions about how he might import his data in such a way that he can select for the disease nodes within Cytoscape?  I am thinking that he can import the data from a table, where the column that contains the diseases will be the target, then use filters to select all targets.  What do you think?
>
> AllanK
>

Yep, but that's a bit of an easy problem so he'd benefit from solving
it himself. (.na)

Ciao
Russ
U. Liverpool

[Majid]

Thanks a lot guys for the quick and helpful replies, I will go ahead
with your comments and try to make the graphics.
Cheers,
M

[allan_k...@agilent.com]

Please keep us posted. It will be very interesting to see what you come up with.

AllanK


Allan Kuchinsky
Principal Project Scientist
Molecular Tools  
Agilent Laboratories

Agilent Technologies, Inc.
Santa Clara, CA, USA
408.553.2423
allan_k...@agilent.com



-----Original Message-----
From: Majid [mailto:majid....@gmail.com]
Sent: Thursday, December 03, 2009 10:21 AM
To: cytoscape-helpdesk
Subject: [cytoscape-helpdesk] Re: Disease-protein network

[Majid]

I installed the Cytoscape and defined the disease names column as
target interaction and followed the Russ's comment. Things worked just
fine and I got the figure.
Thanks a lot and have a nice day,
M

On Dec 3, 1:31 pm, <allan_kuchin...@agilent.com> wrote:
> Please keep us posted.  It will be very interesting to see what you come up with.
>
> AllanK
>
> Allan Kuchinsky
> Principal Project Scientist
> Molecular Tools  
> Agilent Laboratories
>
> Agilent Technologies, Inc.
> Santa Clara, CA, USA
> 408.553.2423

> allan_kuchin...@agilent.com

[Russ]

Cool.
I thought Allan's method for filtering disease nodes a bit long-winded
to be honest. Had I done this, I'd have introduced a node attribute
'Disease' which would take the value 1 for all your disease nodes.
Selecting the disease nodes is then trivial using either the new or
old filters. Importantly, once the attribute is set up, you could also
alter the shape/colour of the disease nodes relative to the protein
nodes for improved visuals.

Regards
Russ

[Majid]

Interesting
Thanks Russ, I am going to use it.
Best,
M

[Diego Bonatto]

Dear researchers,

My research group developed a web server (BioNet) that allow to
collect the proteins associated with a specific disease from OMIM site
[http://www.ncbi.nlm.nih.gov/omim] and use these data to generate
protein-protein interaction network from the site STRING
[http://string.embl.de/]. The archives can be saved in different
formats to be used in Cytoscape.
Please feel free to make suggestions to improve the web server. BioNet
is still under construction and any suggestion is welcome. The link to
access BioNet is www.biosoft.bio.br
Regards,
Diego Bonatto

> --
>
> You received this message because you are subscribed to the Google Groups "cytoscape-helpdesk" group.
> To post to this group, send email to cytoscape...@googlegroups.com.
> To unsubscribe from this group, send email to cytoscape-helpd...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/cytoscape-helpdesk?hl=en.
>
>
>

--
Dr. Diego Bonatto
Laboratorio de Genetica Toxicologica - 206
Instituto de Biotecnologia
Centro de Ciencias Biologicas e da Saude
Universidade de Caxias do Sul - UCS
Rua Francisco Getulio Vargas 1130 – Bloco 57
Caixa Postal 1352
Caxias do Sul – Rio Grande do Sul
BRAZIL
95070-560
Phone: 55-54-3218-2100 (Extension: 2682)
Fax: 55-54-3218-2293
Alternative e-mail: dbon...@ucs.br