Re: [igraph] subgraph by passing list of nodes that grabs also nodes pointing to that list

[Jose Quesada]

Gábor Csárdi wrote:
> Jose,
>
> On Thu, May 14, 2009 at 8:28 AM, Jose Quesada <address@hidden> wrote:
>   
>> Gábor Csárdi wrote:
>>
>> Thanks Gabor,
>>     
>>> there are many ways to do this, perhaps the simplest is to use
>>> neighborhood() to collect the neighboring vertices and then call
>>> subgraph().
>>>
>>> vv <- neighborhood(g, order=1, nodes=mynodes, mode="in")
>>>
>>>       
>> hmm, this doesn't work on my network; it returns exactly the same list,
>> mynodes, with no additional ones.
>> I created the network by reading from a MatrixMarket file into a
>> CsparseMatrix object with readMM, then constructing the graph from that
>> matrix.
>>
>> I'm attaching a toy example, and code to reproduce the problem:
>>
>> library(igraph)
>> library(Matrix)
>>
>> # Toy: just last few movies
>> dataMtxFile = "movies2actors-10-idsOnly.mtx"
>> # ma for movie actors
>> matoy = as(readMM(paste(datapath, dataMtxFile, sep="")), "CsparseMatrix")
>> tg = graph.incidence(matoy, directed=T, mode="out")
>> mynodes =  1377253
>> vv <- neighborhood(tg, order=1, nodes=mynodes, mode="in")
>> vv
>> #> vv
>> #[[1]]
>> #[1] 1377253 <- lists only the original node; misses five inbound
>> connections.
>> vv <- neighborhood(tg, order=1, nodes=mynodes, mode="all") #same for out
>> vv # this lists all nodes but the ones that are actually connected to
>> 1377253
>> # [[1]]
>> #  [1] 1377253 1413423 1457015 1464520 1517383 1641295 1704978 1741976
>> 1744269 1744274 1744315 1745028 1745031 1756801 1783566
>> # [16] 1803739 1836911 1836953 1836954 1836964 1836967 1844560 1848929
>> 1880207 1946456 2132881 2203538 2278148 2282809 2324406
>> # [31] 2409592 2409595 2410120 2410900 2681118 2946503 2946504 2946655
>> 2974742
>>
>> What am I doing wrong?
>>     
>
> It just seems that vertex 1377253 has no edges pointing to it, only
> edges pointing to other vertices from it. If you use 'mode="out"' in
> graph.incidence(), then all edges will point from 'type 1' vertices to
> 'type 2' vertices.
>
> Do you really want to use a directed graph?
>
>   
Yes I do. But in any case, I've tried all combinations of in, out and
all in both graph.incidence and neighborhood and I either get the lone
node or the list.
One particular thing about the list: the ids there are _not_  actor nor
movies in this toy subgraph. Something weird is going on here. I'm not
sure how to debug this problem...

I'm having a lot of trouble to do what seemingly are very simple tasks.
This is surprising. Is there anything like head() that works for igraph
objects? I'm using
write.table(ids, "clipboard", sep='\t') and pasting into a spreadsheet
to get something more visual. I'm sure there are better ways to operate
with igraph objects...

Thanks!


> G.
>
>   
>> Thanks!
>> -Jose
>>     
>>> subgraph(g, unique(unlist(vv)))
>>>
>>> Best,
>>> Gabor
>>>
>>>       
>> --
>> Jose Quesada, PhD.
>> Max Planck Institute,
>> Center for Adaptive Behavior and Cognition -ABC-,
>> Lentzeallee 94, office 224, 14195 Berlin
>> http://www.josequesada.name/
>>
>>
>> _______________________________________________
>> igraph-help mailing list
>> address@hidden
>> http://lists.nongnu.org/mailman/listinfo/igraph-help
>>
>>
>>     
>
>
>
>   


-- 
Jose Quesada, PhD.
Max Planck Institute,
Center for Adaptive Behavior and Cognition -ABC-, 
Lentzeallee 94, office 224, 14195 Berlin
http://www.josequesada.name/