Gephi forums

Filtering, Ranking & Interacting with the graph • Exploring a multigraph in time

2019-03-02T12:07:57+01:00

Dear friends,
I have built a multigraph with networkx and exported it to GEXF.

The nodes are authors of scientific papers and each edge connects to a coauthor for a given publication.

So for example I could have A and B coauthoring three papers in 2019, 2018 and 2017 and this would appear as two nodes and three edges between them.

The graph is medium sized, approx 10.000 authors and 90.000 edges.

What would you suggest to explore the graph time wise? Is there a way to have a slider from 1990 to 2019 and while sliding it only show edges for that year?

I can do this with filters but need a way for end users that do not know how to use Gephi to do the same.

Also another question.

What would be the easiest way of "summing" all edges into one with the weight being the added number of edges?

Thank you very much.

Statistics:Posted by rjalex — 02 Mar 2019 12:07

Filtering, Ranking & Interacting with the graph • Re: Multiple edge weight filters and average edge weight per node

2018-06-06T09:11:02+01:00

Hi,
I think you could use the operator filters to implement some AND conditions for example.

Statistics:Posted by eduramiba — 06 Jun 2018 09:11

Filtering, Ranking & Interacting with the graph • Multiple edge weight filters and average edge weight per node

2018-06-05T23:17:02+01:00

Hey Gephi community,

I've been having lots of fun and some frustrations putting my correlation matrix into Gephi.

My main issue is that I want to show strong negative and positive edges, but filter out the middle near 0 (i.e. show only -0.50 to -0.25 and 0.25 to 0.50). If I apply two edge weight filters, the subsequent filter can only be applied in the range of the initial filter. Any idea how to solve this?

My secondary issue isn't as important, but I would love to change the node size/color to the average correlation value of all weighted edges to that specific node (not weighted degree). If this isn't built in, can I supply a specific value myself to each node for color/size?

Thanks in advance! Giving a talk at a conference in a few weeks and this looks much better than the bar graphs I have to show the means.

Statistics:Posted by DeDwNL — 05 Jun 2018 23:17

Filtering, Ranking & Interacting with the graph • Re: How to remove appearances?

2018-04-18T17:41:58+01:00

I think you can simply apply the 'Unique' sizing/coloring mode and it will reset all.

Statistics:Posted by eduramiba — 18 Apr 2018 17:41

Filtering, Ranking & Interacting with the graph • How to remove appearances?

2018-04-18T17:38:59+01:00

Hello,

Newbie here. I have applied some "appearances" on the nodes of my graph. For example, the labels of the nodes get bigger given the value of a certain metric. Is it possible to remove/reset the appearance? So every label is the same size. The same question applies to all other appearance settings.

Thank you,

Statistics:Posted by vcaballero — 18 Apr 2018 17:38

Filtering, Ranking & Interacting with the graph • Re: Merging multiple edges - not duplicates

2018-03-22T10:08:33+01:00

If you merge the nodes, you choose an strategy to decide what value to keep for each attribute, and there are a few (join values, first, last...).

Statistics:Posted by eduramiba — 22 Mar 2018 10:08

Filtering, Ranking & Interacting with the graph • Re: Edges table won't display all columns

2018-03-22T10:07:34+01:00

Hi,

Currently the table is limited to 20 columns at once, but you can choose them with the top-right lightbulb column.
We will consider adding an horizontal scroll.

Statistics:Posted by eduramiba — 22 Mar 2018 10:07

Filtering, Ranking & Interacting with the graph • Merging multiple edges - not duplicates

2018-03-21T20:26:55+01:00

I have two nodes with three edges between them, all representing a different interaction. If I merge the nodes, and then colorize based on an attribute that was unique to each edge pre-merge, how does Gephi decide which of those attributes to preserve and therefore colorize by?

Statistics:Posted by mbrowning — 21 Mar 2018 20:26

Filtering, Ranking & Interacting with the graph • Edges table won't display all columns

2018-03-21T19:28:43+01:00

The edges table in the Data Laboratory only shows some of my fields and I don't see a way to scroll to the right to view all of them. Even worse, when I click on the buttons at the bottom like "Clear column" or "Fill column with value", I cannot select the columns that do not appear in the table. They are definitely there, but it looks like there's some sort of limit on what the table will show.

Any advice? I need to fill one column with another's value but it's currently impossible.

Statistics:Posted by mbrowning — 21 Mar 2018 19:28

Filtering, Ranking & Interacting with the graph • Resolved: Why are some of my edges really fat and some really thin when the weights are all set to 1.0?

2018-03-13T16:36:59+01:00

I have resolved the problem previously described.

It turned out to be a GEPHI bug, at least I think so. I found by accident that when I exited GEPHI, due to another anomaly, re-executed and restored the saved session, all edges were displayed with the same weights. Realizing that I had kept the GEPHI session (v 0.9.1) up for over a week on my desktop, my conclusion is that GEPHI sessions do not age well and accumulate hiccups like the one I described. My lesson is to save my workspace and exit GEPHI at the end of each workday.

Statistics:Posted by jonrhill — 13 Mar 2018 16:36

Filtering, Ranking & Interacting with the graph • Why are some of my edges really fat and some really thin when the weights are all set to 1.0?

2018-03-12T23:42:19+01:00

I loaded my dependency data in as directed graph data from a .gv file with the following format of pairwise relations:
digraph {
01245 -> 01251
01246 -> 01251
01249 -> 01251
01245 -> 01252
01246 -> 01252
01249 -> 01252
01293 -> 02029[comment=?]
01296 -> 02029[comment=?]
01778 -> 01838
01782 -> 02102
01799 -> 01782
01800 -> 01782
01784 -> 02007
01793 -> 01833
01806 -> 01099
01292 -> 01825
}

Yet when I get an Overview visualization, some of the edges displayed automatically by GEPHI are really thick (i.e. wide) and some are really thin, even though the weights are all set to 1.0 in the Data Laboratory (which seems to have no effect on the display). I cannot discern any particular pattern for how these edge thicknesses are being set by GEPHI - it is not a function of the number of connections or node density, for example. Any suggestions how I can get the same edges that are all the same width (or thickness) when the weights are all set to 1.0?

Statistics:Posted by jonrhill — 12 Mar 2018 23:42

Filtering, Ranking & Interacting with the graph • Re: How to change the weight of only filtered records

2018-03-06T09:33:31+01:00

Thank you!

Statistics:Posted by rjalex — 06 Mar 2018 09:33

Filtering, Ranking & Interacting with the graph • FILTER "Mutual Degree Range"

2018-02-09T13:53:17+01:00

Hello,

Any idea of what "Mutual Degree Range" ables to do? I think that it gives the nodes whose grade is in the range selected but I am not sure.

Best regards,
Arturo

Statistics:Posted by turoafro — 09 Feb 2018 13:53

Filtering, Ranking & Interacting with the graph • Re: Finding nodes that share more than one edge

2018-02-02T14:16:59+01:00

Very interesting. I'll never stop learning
Thank you very much

Statistics:Posted by rjalex — 02 Feb 2018 14:16

Filtering, Ranking & Interacting with the graph • Re: Finding nodes that share more than one edge

2018-02-02T12:40:57+01:00

Hi,
The multimode networks plugin might be useful: https://seinecle.github.io/gephi-tutori ... ph-en.html

Statistics:Posted by eduramiba — 02 Feb 2018 12:40

Filtering, Ranking & Interacting with the graph • Finding nodes that share more than one edge

2018-02-02T12:26:47+01:00

Dear friends,
I have a bipartite oriented graph modeling a metabolic network with two types of molecules: xenobiotics (drugs and other biological exogenous substance) and their metabolizing enzymes.

Have three types of relations:

An enzyme can metabolize a xenobiotic

A xenobitoc can inhibit an enzyme

A xenobiotic can augment an enzyme

In some (rarish) cases a xenobiotic is both metabolized by an enzyme but at the same time it also inhibits or augments the same enzyme.

This is modeled with two oriented arcs between the two molecules, one expresses the metabolizes relation while the other the inhibits.

Is there a way to filter the graph to only show these doubly linked molecules?

Thank you

Statistics:Posted by rjalex — 02 Feb 2018 12:26

Filtering, Ranking & Interacting with the graph • Re: How to change the weight of only filtered records

2018-02-02T12:19:00+01:00

Doh!

Thank you so much! Worked as a charm!

Statistics:Posted by rjalex — 02 Feb 2018 12:19

Filtering, Ranking & Interacting with the graph • Re: How to change the weight of only filtered records

2018-01-31T22:59:43+01:00

You can select the desired rows, right click, open all of them on edit panel and set the value you want to the any attribute.

Other way is select the desired rows, right click and use the 'Overwrite data to other selected nodes...' feature.

Statistics:Posted by eduramiba — 31 Jan 2018 22:59

Filtering, Ranking & Interacting with the graph • How to change the weight of only filtered records

2018-01-31T19:31:16+01:00

Dear friends,
I have around 30% of my arcs labeled as "metabolizes" in an "arc_type" column I've added to my data laboratory layout.

Currently all edges have a weight of 1.0

I would like to change this to 3.0 but only for the "metabolizes" arcs.

I triggered a filter that shows only the proper records and tried using the "fill columns" button which worked but then when I stopped the filter I found this changes ALL the records even those which are hidden by the filter.

Is there a(nother) way to perform what I need, within Gephi?

Thank you in advance.

Statistics:Posted by rjalex — 31 Jan 2018 19:31

Filtering, Ranking & Interacting with the graph • Re: Interaction with an edge

2018-01-31T16:40:04+01:00

Hi,
At the moment this can only by done in data laboratory.

Statistics:Posted by eduramiba — 31 Jan 2018 16:40