Hi again! Thank you for replying that quickly. You helpfully pointed out i could just manipulate my correlation matrix before i carry out my plotting. That was very helpful on its own.
Currently, as i have just started with network analysis, i use rather easy-to-grasp approaches. iGraph seems to be much more powerful than qGraph, but it seems like i need a while to get an idea how to get my graphs out. For example: qgraph(cor(df, method="spearman", use="pairwise.complete.obs"), layout="spring", minimum=0.4, shape="circle", label.cex=0.4, vsize=0.4 , groups=list, legend=TRUE, border.width=0.1, color=c25, filetype="pdf", filename="1", height=80, width=80, legend.cex=5, usePCH=TRUE) gives me a non-circular, spring-outlaid network of spearman rank correlation from my input dataframe, in which i could control node sizes/shapes, could colourize by a vector with group IDs according to my own palette, and which ignored correlations not meeting my treshold of 0.4. I need now to translate this into iGraph, which seems very untrivial to do, for example, i have yet to find out how to control graphical parameters with graph.adjacency. But i hope i can get back to you later once i developed some expertise with iGraph. Thanks again! Tim Message: 2 Date: Tue, 12 Aug 2014 10:53:59 -0400 From: G?bor Cs?rdi<[email protected]> To: Help for igraph users<[email protected]> Subject: Re: [igraph] Using iGraph for correlated abundances Message-ID: <CABtg=kmj9ia6qjs7-pzm8xm9dyxvq605tc04coga5sq9hze...@mail.gmail.com> Content-Type: text/plain; charset=UTF-8 Hi Tim, yes, you can delete any edges from an you want from an igraph object with delete.edges. Maybe you want to set some elements of a matrix zero, then do it like this: M[ condition ] <- 0 E.g. if you have pairwise p values in another matrix, then you can say: M [ p > p_threshold ] <- 0 To say something more specific, you need to tell us more precisely what you are doing. E.g. see http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example for advice on how to ask questions about R and R packages. Gabor On Tue, Aug 12, 2014 at 10:37 AM, Tim Richter-Heitmann <[email protected]> wrote:
Hi there, i have a very basic question. I want to plot my large abundance data according to simple pairwise x,y pearson and/or spearman correlation. I have used qgraph so far, but it cannot filter for non-significant p-values (it can either display significant observations or correlation strengths, but not both). My question is, can iGraph do that by omitting non significant values at all? I just recently delved into network analysis, from what ive seen on the net, the graph.adjacency is the most common way to do so, and ive seen examples for deleting edges not meeting tresholds, but ive not seen the significancy covered yet. Also fine would be coloring according to cor-strength, label width according to significancy level (or vice versa). Any example code is very welcome. Thank you very much! Tim _______________________________________________ igraph-help mailing list [email protected] https://lists.nongnu.org/mailman/listinfo/igraph-help
------------------------------ _______________________________________________ igraph-help mailing list [email protected] https://lists.nongnu.org/mailman/listinfo/igraph-help End of igraph-help Digest, Vol 97, Issue 5 ****************************************** -- Tim Richter-Heitmann (M.Sc.) PhD Candidate International Max-Planck Research School for Marine Microbiology University of Bremen Microbial Ecophysiology Group (AG Friedrich) FB02 - Biologie/Chemie Leobener Straße (NW2 A2130) D-28359 Bremen Tel.: 0049(0)421 218-63062 Fax: 0049(0)421 218-63069 _______________________________________________ igraph-help mailing list [email protected] https://lists.nongnu.org/mailman/listinfo/igraph-help
