Force Directed or Spring Embedded Layout

[KTBCN]

Hi all,

I have a very simple problem and it's driving me nuts. I have a gene
list with about 400 genes. The file is 3 columns plain and simple:
column 1 is always "Per1" my gene of interest, column 2 is a list of
various genes which show high expression correlation with Per1 (i.e.
>= 0.7), and column 3 is the expression correlation value which is
calculated using Pearson ranging from 0.7 to 1.0.

I first load column 1 as source node and column 2 as target nodes and
column 3 as edge attributes. I want to display the network in such a
way that the more weight an edge has (i.e. the higher the Pearson
value is -- closer to 1) the closer the node is situated to the hub
gene Per1. And the lower the Pearson value is (i.e. the close it is to
0.7 my minimum) the farther it is.

I have read in the documentation at
http://opentutorials.cgl.ucsf.edu/index.php/Tutorial:Introduction_to_Cytoscape#Force-Directed_Layouts
and it seems that the force on a spring or spring length should be the
weight on the edge.

Why is it that neither "Edge-Weighted Spring Embedded" and "Edge-
Weighted Force Directed (BioLayout)" seem to do anything? The nodes
are just distributed randomly and not based on their Pearson value.

Thanks for your help.
Kiana

[Daniele Merico @BaderLab]

Hey Kiana!

I usually use 'Force Directed / Weighted' (that's what we use in our
plugin, Enrichment Map, btw); you should be able to select your edge
attribute from the layout menu as: Force Directed / ... (as an
alternative to 'unweighted' *). Try contrasting it to organic, which
does not use weights; sometimes, when networks are highly connected,
force directed results in overlapping nodes and label, a problem you
won't have with organic. Sometimes, spring embedded has the tendency
to create more disperse networks (with some nodes really far apart),
but it may serve you well in specific situations. I remotely remember
Gary mentioning problems with the BioLayout, not sure though.

* Make sure you can access your edge attribute from the attribute
browser, if you have doubts.

Also, you may want to play around with the correlation threshold a
bit, of course.

Ciao!
Daniele

On 11 Feb, 10:11, KTBCN <kiana.toufi...@gmail.com> wrote:
> Hi all,
>
> I have a very simple problem and it's driving me nuts. I have a gene
> list with about 400 genes. The file is 3 columns plain and simple:
> column 1 is always "Per1" my gene of interest, column 2 is a list of
> various genes which show high expression correlation with Per1 (i.e.>= 0.7), and column 3 is the expression correlation value which is
>
> calculated using Pearson ranging from 0.7 to 1.0.
>
> I first load column 1 as source node and column 2 as target nodes and
> column 3 as edge attributes. I want to display the network in such a
> way that the more weight an edge has (i.e. the higher the Pearson
> value is -- closer to 1) the closer the node is situated to the hub
> gene Per1. And the lower the Pearson value is (i.e. the close it is to
> 0.7 my minimum) the farther it is.
>

> I have read in the documentation athttp://opentutorials.cgl.ucsf.edu/index.php/Tutorial:Introduction_to_...

[KTBCN]

Hi Daniele,

Thanks for your reply. My edge attributes are in fact accessible from
the "Edge Attribute Browser" but when I invoke the Edge-Directed Force-
Directed (Biolayout) it doesn't seem to be working on any logic. For
example, as you said some of the nodes are really really far. One in
particular, so one would assume that one node to either have the
maximum Pearson correlation value of 1 or the minimum of 0.7 but it's
not!! My network is basically one hub connected to everything else
with a weight on each edge. There are no closed loops and like I said
the connections only go one way: between the hub node and everybody
else! So why is this not working?

Thanks in advance for your help.
Frustrated Kiana.


On Feb 12, 7:49 pm, "Daniele Merico @BaderLab"

[Ruth Isserlin]

Hi Preeti, 

Sometimes I find that applying the default force-directed or other types of layouts form the layout menu don't change much.  In order to better tweak them click on layout->Settings...  Select the layout you want to use and adjust the parameters to get better results.

Ruth 

On Fri, May 24, 2013 at 11:33 AM, P. Iyer <kpreet...@gmail.com> wrote:

Hi,

did you obtain a solution for this problem? If yes, would it be possible to know how? I am currently facing a similar situation.

Many thanks in advance!

Regards,

Preeti

--
You received this message because you are subscribed to the Google Groups "cytoscape-discuss" group.

To unsubscribe from this group and stop receiving emails from it, send an email to cytoscape-disc...@googlegroups.com.

To post to this group, send email to cytoscap...@googlegroups.com.

Visit this group at http://groups.google.com/group/cytoscape-discuss?hl=en.

For more options, visit https://groups.google.com/groups/opt_out.