Re: [igraph] Motifs in igraph

[Szabolcs Horvát]

You can use subgraph_isomorphisms with the lad method and induced=TRUE

http://igraph.org/r/doc/subgraph_isomorphisms.html

It will tell you where exactly the motif (i.e. induced subgraph)
appears in the graph.
On Fri, 19 Oct 2018 at 01:15, Stuart Kininmonth
 wrote:
>
>
>
> Hi Igraphers’,
>
>
>
> Has anybody developed code in R or python that can find motifs of various 
> configurations (from 3 and 4 node motifs) that also keeps a list of the node 
> identities?
>
> This would then enable me to see the relationship of the node to the other 
> nodes via the shared motifs.
>
>
>
> With thanks
>
>
>
> Stuart
>
>
>
>
>
> Stuart Kininmonth PhD
>
> Deputy Head of School
> Senior Lecturer
>
>   :Coral Reef Ecology and Management
>   :Marine Spatial Planning
>
>   :Marine Geology
> School of Marine Studies
> Faculty of Science, Technology & Environment
> The University of South Pacific
> Laucala Bay Road, Suva, Fiji Islands
> Tel: (+679) 32 32943  Ext.: 32943
>
> stuart.kininmo...@usp.ac.fj
>
> M: +679 8660103
>
>
>
>
>
>
>
>
>
> ___
> igraph-help mailing list
> igraph-help@nongnu.org
> https://lists.nongnu.org/mailman/listinfo/igraph-help

___
igraph-help mailing list
igraph-help@nongnu.org
https://lists.nongnu.org/mailman/listinfo/igraph-help

Re: [igraph] motifs

[Manuel J. Zetina-Rejón]

Hi Tamas,

Thank you, I got it.

Manuel
=
Dr. Manuel J.  Zetina-Rejón
Instituto Politécnico Nacional
Centro Interdisciplinario de Ciencias Marinas
Lab. de Dinámica y Manejo de Ecosistemas Acuáticos
Av. Instituto Politénico Nacional S/N Col. Playa Palo de Sta. Rita La Paz, BCS, 
México
CP 23096
Tel. +52 (612) 1234658  Ext. 82452 
Fax. +52(612) 1225322
http://www.cicimar.ipn.mx 

Visualízame
http://vizualize.me/mjzetina

Google Scholar
http://scholar.google.com.mx/citations?user=Whcz9wUJ&hl=es 


ResearchGate
https://www.researchgate.net/profile/Manuel_Zetina-Rejon 

=

> El 07/11/2016, a las 04:42, Tamas Nepusz  escribió:
> 
> Hi and sorry for the late reply, this email has fallen through the
> cracks somehow. (I was on the road in China for a few weeks with
> limited internet access).
> 
>> On the other hand, motifs() returns a vector with the number of occurences
>> of each motif in the graph ordered by their isomorphism class. How do I know
>> which isomorphism class are present in my graph. I mean, a possible result
>> of motifs(graph, size =3) could be something like this:
>>> NA NA 32 18
>> 
>> For me, this means that there are 32 motifs of x-isomorphic class and other
>> 18 motifs of y-isomorphic class. But, which classes?
> The i-th element of the vector corresponds to the motif with
> isomorphism class i. So, you can indeed use
> graph_from_isomorphism_class() to figure out how the motif looks like.
> The only catch is that, unlike everything else in R, isomorphism
> classes are zero-based, so the two NAs belong to isomorphism classes 0
> (empty graph) and 1 (two vertices connected, and a third isolated
> vertex), and then you have 32 occurrences of isomorphism class 2 (V
> shape) and 18 occurrences of isomorphism class 3 (full triangle).
> 
> T.
> 
> ___
> igraph-help mailing list
> igraph-help@nongnu.org
> https://lists.nongnu.org/mailman/listinfo/igraph-help

___
igraph-help mailing list
igraph-help@nongnu.org
https://lists.nongnu.org/mailman/listinfo/igraph-help

Re: [igraph] motifs

[Tamas Nepusz]

Hi and sorry for the late reply, this email has fallen through the
cracks somehow. (I was on the road in China for a few weeks with
limited internet access).

> On the other hand, motifs() returns a vector with the number of occurences
> of each motif in the graph ordered by their isomorphism class. How do I know
> which isomorphism class are present in my graph. I mean, a possible result
> of motifs(graph, size =3) could be something like this:
>> NA NA 32 18
>
> For me, this means that there are 32 motifs of x-isomorphic class and other
> 18 motifs of y-isomorphic class. But, which classes?
The i-th element of the vector corresponds to the motif with
isomorphism class i. So, you can indeed use
graph_from_isomorphism_class() to figure out how the motif looks like.
The only catch is that, unlike everything else in R, isomorphism
classes are zero-based, so the two NAs belong to isomorphism classes 0
(empty graph) and 1 (two vertices connected, and a third isolated
vertex), and then you have 32 occurrences of isomorphism class 2 (V
shape) and 18 occurrences of isomorphism class 3 (full triangle).

T.

___
igraph-help mailing list
igraph-help@nongnu.org
https://lists.nongnu.org/mailman/listinfo/igraph-help

Re: [igraph] motifs

[Manuel Zetina-Rejon]

Hi Tamas!

Thank you very much for your reply. Actually, it makes sense to me on the use 
of connected subgraphs as motifs, in the sense that an isolated vertex doesn’t 
play a important role in the subgraph structure.

On the other hand, motifs() returns a vector with the number of occurences of 
each motif in the graph ordered by their isomorphism class. How do I know which 
isomorphism class are present in my graph. I mean, a possible result of 
motifs(graph, size =3) could be something like this:
> NA NA 32 18

For me, this means that there are 32 motifs of x-isomorphic class and other 18 
motifs of y-isomorphic class. But, which classes? I’ve playing with something 
like this to explore how all possible isomorphic classes (n = 3) look like:

iso <- vector(mode = "list", length = 3)
iso.class <- 1:length(iso)

for(i in 1:length(iso)){
  iso[[i]] <- graph_from_isomorphism_class(3, i, directed = FALSE)
iso[[i]]$name <- paste("Class", iso.class[i])
}

par(mfrow = c(1,3))
for(i in 1:length(iso)){
plot(iso[[i]], layout = layout_in_circle(iso[[i]]),  main = iso[[i]]$name)
}

But How do I know which of them are present in the original graph?

Thank you very much!

Manuel

> El 14/10/2016, a las 02:31, Tamas Nepusz  escribió:
> 
> Hi!
> 
> I haven't read the papers once again, but in my opinion a disconnected
> motif doesn't really make sense. Consider a disconnected motif that
> consists of a fully connected triangle and an additional isolated
> vertex, and then take a graph that contains one triangle and one
> million isolated vertices. Does that really mean that this "motif"
> appears one million times in the graph? Is that a significant finding?
> If I added an additional one million totally unrelated vertices to the
> graph, does that make the motif appear twice as frequently?
> 
> Anyway, if you want to search for disconnected patterns in a graph,
> you can still use count_subgaph_isomorphisms() with method="lad" and
> induced=TRUE; see:
> 
> http://igraph.org/r/doc/count_subgraph_isomorphisms.html
> 
> It will be much slower, though -- searching for connected motifs is
> much easier if the average degree of a vertex is low.
> 
> T.
> 
> 
> On Fri, Oct 14, 2016 at 8:59 AM, Manuel Zetina-Rejon  
> wrote:
>> Hi Guys!
>> 
>> This is probably a basic question, but I don’t find the clear criteria or 
>> reference, why in igraph help, you mention that unconnected subgraphs (of x 
>> isomorphic class) are not considered motifs? For that reason, motifs() is NA 
>> for unconnected subgraphs. It is also not clear if you mean strongly or 
>> weakly connected subgraphs
>> 
>> According to Milo et al. (2002) and Shen-Orr et al. (2002) motifs are not 
>> necessarily connected, even in directed graphs.
>> 
>> Thank you for your opinions
>> 
>> 
>> Manuel
>> ___
>> igraph-help mailing list
>> igraph-help@nongnu.org
>> https://lists.nongnu.org/mailman/listinfo/igraph-help
> 
> ___
> igraph-help mailing list
> igraph-help@nongnu.org
> https://lists.nongnu.org/mailman/listinfo/igraph-help

___
igraph-help mailing list
igraph-help@nongnu.org
https://lists.nongnu.org/mailman/listinfo/igraph-help

Re: [igraph] motifs

[Tamas Nepusz]

Hi!

I haven't read the papers once again, but in my opinion a disconnected
motif doesn't really make sense. Consider a disconnected motif that
consists of a fully connected triangle and an additional isolated
vertex, and then take a graph that contains one triangle and one
million isolated vertices. Does that really mean that this "motif"
appears one million times in the graph? Is that a significant finding?
If I added an additional one million totally unrelated vertices to the
graph, does that make the motif appear twice as frequently?

Anyway, if you want to search for disconnected patterns in a graph,
you can still use count_subgaph_isomorphisms() with method="lad" and
induced=TRUE; see:

http://igraph.org/r/doc/count_subgraph_isomorphisms.html

It will be much slower, though -- searching for connected motifs is
much easier if the average degree of a vertex is low.

T.


On Fri, Oct 14, 2016 at 8:59 AM, Manuel Zetina-Rejon  wrote:
> Hi Guys!
>
> This is probably a basic question, but I don’t find the clear criteria or 
> reference, why in igraph help, you mention that unconnected subgraphs (of x 
> isomorphic class) are not considered motifs? For that reason, motifs() is NA 
> for unconnected subgraphs. It is also not clear if you mean strongly or 
> weakly connected subgraphs
>
> According to Milo et al. (2002) and Shen-Orr et al. (2002) motifs are not 
> necessarily connected, even in directed graphs.
>
> Thank you for your opinions
>
>
> Manuel
> ___
> igraph-help mailing list
> igraph-help@nongnu.org
> https://lists.nongnu.org/mailman/listinfo/igraph-help

___
igraph-help mailing list
igraph-help@nongnu.org
https://lists.nongnu.org/mailman/listinfo/igraph-help