Merge pull request #111 from clhunsen/claus-updates

Various minor and major fixes and adjustments Reviewed-by: Thomas Bock <[email protected]>
se-sic · Mar 26, 2018 · f07ee11 · f07ee11
2 parents 73e9caf + e1a339d
commit f07ee11
Show file tree

Hide file tree

Showing 7 changed files with 167 additions and 82 deletions.
diff --git a/NEWS.md b/NEWS.md
@@ -1,6 +1,22 @@
 # codeface-extraction-r - Changelog
 
 
+## Unversioned
+
+### Added
+- ...
+
+### Changed/Improved
+- Vertex and edge types (attribute `"type"`) are now a character string (e.g., "Author" or "Unipartite") (#110, 3ca6ed99cf377200adb94a4b27ed1ea7d3a6981a)
+- Default plotting layout is now `igraph::layout.kamada.kawai` (#109, 909965453c47c26c902612cb0c9aa16a5b56746a)
+- Remove parameter 'color.attr' from 'motifs.search.in.network' (d33f6863aaf05ae1a8acf7f5667784713796b734)
+- Fix and clean-up of both the plotting and the motif modules (3ca6ed99cf377200adb94a4b27ed1ea7d3a6981a, consequence of #110)
+
+### Fixed
+- Probably fixed segfaults during plotting by changing the default layout (see above and issue #109)
+- Fix gray-scale plotting of networks (730cc544edbb30ea3aa89a91e123e74b18a942c6)
+
+
 ## 3.1.1
 
 ### Changed/Improved

diff --git a/showcase.R b/showcase.R
@@ -11,7 +11,7 @@
 ## with this program; if not, write to the Free Software Foundation, Inc.,
 ## 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
 ##
-## Copyright 2016-2017 by Claus Hunsen <[email protected]>
+## Copyright 2016-2018 by Claus Hunsen <[email protected]>
 ## Copyright 2017 by Raphael Nömmer <[email protected]>
 ## Copyright 2017 by Christian Hechtl <[email protected]>
 ## Copyright 2017 by Felix Prasse <[email protected]>
@@ -302,8 +302,8 @@ y = NetworkBuilder$new(project.data = y.data, network.conf = net.conf)
 # plot.print.network(g, labels = TRUE, grayscale = FALSE)
 
 # ## set a layout and print directly
-# lay = matrix(c(  20, 179, 552, 693, 956, 1091, 124, 317, 516, 615, 803, 1038,
-#                 245, 175, 185, 255, 253, 225,   73,   8,  75,   0,  96,   86),
+# lay = matrix(c(  20, 179, 693, 552, 956, 1091, 124, 317, 516, 615, 803, 1038,
+#                 245, 175, 255, 185, 253, 225,   73,   8,  75,   0,  96,   86),
 #              nrow = 12, byrow = FALSE) # for sample graph
 # g = igraph::set.graph.attribute(g, "layout", lay)
 # plot.print.network(g, labels = TRUE, grayscale = FALSE)

diff --git a/tests/test-motifs.R b/tests/test-motifs.R
@@ -32,29 +32,29 @@ test_that("Motifs are found in sample network (remove.duplicates = TRUE).", {
         igraph::V(network)[ c("D3", "D4") ],
         igraph::V(network)[ c("D4", "D5") ]
     )
-    matchings = motifs.search.in.network(network, MOTIFS.LINE, color.attr = "type", remove.duplicates = TRUE)
+    matchings = motifs.search.in.network(network, MOTIFS.LINE, remove.duplicates = TRUE)
     expect_equal(matchings, expected, info = "Line motif")
 
     ## Triangle motif (positive)
     expected = list(
         igraph::V(network)[ c("D1", "D2", "A1") ]
     )
-    matchings = motifs.search.in.network(network, MOTIFS.TRIANGLE.POSITIVE, color.attr = "type", remove.duplicates = TRUE)
+    matchings = motifs.search.in.network(network, MOTIFS.TRIANGLE.POSITIVE, remove.duplicates = TRUE)
     expect_equal(matchings, expected, info = "Triangle motif (positive)")
 
     ## Triangle motif (negative)
     expected = list(
         igraph::V(network)[ c("D1", "D2", "A1") ],
         igraph::V(network)[ c("D5", "D6", "A6") ]
     )
-    matchings = motifs.search.in.network(network, MOTIFS.TRIANGLE.NEGATIVE, color.attr = "type", remove.duplicates = TRUE)
+    matchings = motifs.search.in.network(network, MOTIFS.TRIANGLE.NEGATIVE, remove.duplicates = TRUE)
     expect_equal(matchings, expected, info = "Triangle motif (negative)")
 
     ## Square motif (positive)
     expected = list(
         igraph::V(network)[ c("D4", "D5", "A5", "A6") ]
     )
-    matchings = motifs.search.in.network(network, MOTIFS.SQUARE.POSITIVE, color.attr = "type", remove.duplicates = TRUE)
+    matchings = motifs.search.in.network(network, MOTIFS.SQUARE.POSITIVE, remove.duplicates = TRUE)
     expect_equal(matchings, expected, info = "Square motif (positive)")
 
     ## Square motif (negative)
@@ -64,7 +64,7 @@ test_that("Motifs are found in sample network (remove.duplicates = TRUE).", {
         igraph::V(network)[ c("D4", "D5", "A5", "A6") ],
         igraph::V(network)[ c("D4", "D6", "A5", "A6") ]
     )
-    matchings = motifs.search.in.network(network, MOTIFS.SQUARE.NEGATIVE, color.attr = "type", remove.duplicates = TRUE)
+    matchings = motifs.search.in.network(network, MOTIFS.SQUARE.NEGATIVE, remove.duplicates = TRUE)
     expect_equal(matchings, expected, info = "Square motif (negative)")
 
 })
@@ -86,15 +86,15 @@ test_that("Motifs are found in sample network (remove.duplicates = FALSE).", {
         igraph::V(network)[ c("D3", "D4") ], # unordered: D4, D3
         igraph::V(network)[ c("D4", "D5") ]  # unordered: D5, D4
     )
-    matchings = motifs.search.in.network(network, MOTIFS.LINE, color.attr = "type", remove.duplicates = FALSE)
+    matchings = motifs.search.in.network(network, MOTIFS.LINE, remove.duplicates = FALSE)
     expect_equal(matchings, expected, info = "Line motif")
 
     ## Triangle motif (positive)
     expected = list(
         igraph::V(network)[ c("D1", "D2", "A1") ],
         igraph::V(network)[ c("D1", "D2", "A1") ]  # unordered: D2, D1, A1
     )
-    matchings = motifs.search.in.network(network, MOTIFS.TRIANGLE.POSITIVE, color.attr = "type", remove.duplicates = FALSE)
+    matchings = motifs.search.in.network(network, MOTIFS.TRIANGLE.POSITIVE, remove.duplicates = FALSE)
     expect_equal(matchings, expected, info = "Triangle motif (positive)")
 
     ## Triangle motif (negative)
@@ -104,15 +104,15 @@ test_that("Motifs are found in sample network (remove.duplicates = FALSE).", {
         igraph::V(network)[ c("D5", "D6", "A6") ],
         igraph::V(network)[ c("D5", "D6", "A6") ]  # unordered: D6, D5, A1
     )
-    matchings = motifs.search.in.network(network, MOTIFS.TRIANGLE.NEGATIVE, color.attr = "type", remove.duplicates = FALSE)
+    matchings = motifs.search.in.network(network, MOTIFS.TRIANGLE.NEGATIVE, remove.duplicates = FALSE)
     expect_equal(matchings, expected, info = "Triangle motif (negative)")
 
     ## Square motif (positive)
     expected = list(
         igraph::V(network)[ c("D4", "D5", "A5", "A6") ],
         igraph::V(network)[ c("D4", "D5", "A5", "A6") ]  # unordered: D5, D4, A5, A6
     )
-    matchings = motifs.search.in.network(network, MOTIFS.SQUARE.POSITIVE, color.attr = "type", remove.duplicates = FALSE)
+    matchings = motifs.search.in.network(network, MOTIFS.SQUARE.POSITIVE, remove.duplicates = FALSE)
     expect_equal(matchings, expected, info = "Square motif (positive)")
 
     ## Square motif (negative)
@@ -126,7 +126,7 @@ test_that("Motifs are found in sample network (remove.duplicates = FALSE).", {
         igraph::V(network)[ c("D4", "D5", "A5", "A6") ], # unordered: D5, D4, A5, A6
         igraph::V(network)[ c("D4", "D6", "A5", "A6") ]  # unordered: D6, D4, A5, A6
     )
-    matchings = motifs.search.in.network(network, MOTIFS.SQUARE.NEGATIVE, color.attr = "type", remove.duplicates = FALSE)
+    matchings = motifs.search.in.network(network, MOTIFS.SQUARE.NEGATIVE, remove.duplicates = FALSE)
     expect_equal(matchings, expected, info = "Square motif (negative)")
 
 })

diff --git a/util-core-peripheral.R b/util-core-peripheral.R
@@ -996,13 +996,13 @@ get.author.class = function(author.data.frame, calc.base.name, result.limit = NU
             sapply(author.cumsum, function(x) isTRUE(all.equal(x, author.class.threshold)))
     ))
 
-    ## classify developers according to threshold
+    ## classify authors according to threshold
     core.classification = rep(FALSE, nrow(author.data))
     core.classification[1:author.class.threshold.idx] = TRUE
 
     ## With no activity/collaboration occurring, all authors are classified as peripheral.
     if (author.class.threshold == 0) {
-        logging::logwarn("No collaboration/activity occured, thus, all developer's classification is set to peripheral.")
+        logging::logwarn("No collaboration/activity occured, thus, all authors' classification is set to peripheral.")
         core.classification = rep(FALSE, length(core.classification))
         # ## old code: if we found no core author (should not happen anymore)
         # } else if (!any(core.classification)) {

diff --git a/util-motifs.R b/util-motifs.R
@@ -66,38 +66,81 @@ MOTIFS.SQUARE.NEGATIVE = igraph::make_empty_graph(directed = FALSE) +
           type = c(TYPE.EDGES.INTER, TYPE.EDGES.INTER, TYPE.EDGES.INTRA))
 
 
+## / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /
+## Mapping of vertex and edge types to numerics ----------------------------
+
+## Map vertex and edge types to numerics
+MOTIF.TYPE.MAPPING = as.data.frame(rbind(
+    c(character = TYPE.AUTHOR, numeric = 1),
+    c(character = TYPE.ARTIFACT, numeric = 2),
+    c(character = TYPE.EDGES.INTRA, numeric = 3),
+    c(character = TYPE.EDGES.INTER, numeric = 4)
+))
+
+
+#' Get numeric vertex types from the given \code{network}.
+#'
+#' For this, the actual vertex types are mapped to numerics using the pre-defined
+#' mapping \code{MOTIF.TYPE.MAPPING}.
+#'
+#' @param network the network from which to get the vertex types
+#' @param index the subset of vertices to consider [default: igraph::V(network)]
+#'
+#' @return the vector of numeric vertex types for the (subset of) vertices in the network
+#'
+#' @seealso \code{MOTIF.TYPE.MAPPING}
+get.vertex.types.as.numeric = function(network, index = igraph::V(network)) {
+
+    ## get the vertex attribute as factor
+    attr.factor = factor(igraph::get.vertex.attribute(network, "type", index))
+
+    ## replace factor levels with corresponding numerics
+    levels(attr.factor) = sapply(levels(attr.factor), function(f) {
+        ## get the numeric from MOTIF.TYPE.MAPPING
+        num = subset(MOTIF.TYPE.MAPPING, character == f, numeric, drop = TRUE)
+        return(num)
+    })
+    ## re-create vertex-attribute vector with the numerics
+    attr.numeric = as.numeric(levels(attr.factor))[attr.factor]
+
+    return(attr.numeric)
+}
+
+
 ## / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /
 ## Motif-identification functions ------------------------------------------
 
 #' Search for the given \code{motif} in the given \code{network}.
 #'
+#' The vertex attribute *"type"* is used to match the vertices in the \code{motif} with
+#' the ones in the given \code{network}. This means basically that a vertex with
+#' a certain value for its attribute "type" in the \code{motif} is only matched with
+#' vertices in \code{network} that have the same attribute values.
+#'
 #' @param network The network in which the given \code{motif} is searched
 #' @param motif The motif to search for in the given \code{network}
-#' @param color.attr The the vertex and edge attribute that is used to match the vertices
-#'                   and the edges in the \code{motif} with the ones in the given \code{network}.
-#'                   This means that a vertex A with the attribute declared in \code{color.attr}
-#'                   in the \code{motif} is only matched with vertices with the same attribute in
-#'                   the network. The same holds for edges.
 #' @param remove.duplicates If \code{remove.duplicates == TRUE}, any duplicate matched motifs are
 #'                          removed. This logical value basically resembles the idea of respecting
 #'                          the order within a matched motif or not.
 #'
-#' @return A list of vertex sequences denoting the matched motifs, ordered by \code{color.attr}
-#'         and "name" vertex attributes
+#' @return A list of vertex sequences denoting the matched motifs, ordered by "type" and "name"
+#'         vertex attributes
+#'
+#' @seealso \code{igraph::subgraph_isomorphisms} (method "vf2")
 #'
 #' @examples
-#' motifs.search.in.network(get.sample.network(), MOTIFS.TRIANGLE.NEGATIVE, color.attr = "type", remove.duplicates = TRUE)
-motifs.search.in.network = function(network, motif, color.attr = "type", remove.duplicates = TRUE) {
+#' motifs.search.in.network(get.sample.network(), MOTIFS.TRIANGLE.NEGATIVE, remove.duplicates = TRUE)
+motifs.search.in.network = function(network, motif, remove.duplicates = TRUE) {
 
     ## find motif in network
     vs = igraph::subgraph_isomorphisms(target = network, pattern = motif, method = "vf2",
-                                       vertex.color1 = igraph::get.vertex.attribute(network, color.attr),
-                                       vertex.color2 = igraph::get.vertex.attribute(motif, color.attr))
+                                       vertex.color1 = get.vertex.types.as.numeric(network),
+                                       vertex.color2 = get.vertex.types.as.numeric(motif))
 
     ## normalize found vertex sequences (sort vertices)
     vs.cleaned = lapply(vs, function(seq) {
         ## get types and names of vertices
-        types = igraph::get.vertex.attribute(network, color.attr, index = seq)
+        types = get.vertex.types.as.numeric(network, index = seq)
         names = igraph::get.vertex.attribute(network, "name", index = seq)
 
         ## sort vertex sequence by types and names
@@ -160,7 +203,6 @@ motifs.count = function(network, motifs, remove.duplicates = TRUE, raw.data = FA
         motifs.search.in.network(
             network = network,
             motif = motif,
-            color.attr = "type",
             remove.duplicates = remove.duplicates
         )
     })

diff --git a/util-networks.R b/util-networks.R
@@ -31,13 +31,13 @@ requireNamespace("igraph") # networks
 ## / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /
 ## Vertex and edge types ---------------------------------------------------
 
-## node types
-TYPE.AUTHOR = 1
-TYPE.ARTIFACT = 2
+## vertex types
+TYPE.AUTHOR = "Author"
+TYPE.ARTIFACT = "Artifact"
 
-# edge types
-TYPE.EDGES.INTRA = 3
-TYPE.EDGES.INTER = 4
+## edge types
+TYPE.EDGES.INTRA = "Unipartite"
+TYPE.EDGES.INTER = "Bipartite"
 
 
 ## / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /
@@ -790,20 +790,20 @@ construct.network.from.list = function(list, network.conf, directed = FALSE) {
             ## get vertex data
             nodes = unique(set[, 1])
 
-            ## break if there is no developer
+            ## break if there is no author
             if (length(nodes) < 1) {
                 return(NULL)
             }
 
-            ## if there is only one developer, just create the node, but no edges
+            ## if there is only one author, just create the node, but no edges
             if (length(nodes) == 1) {
                 edges = data.frame()
                 attr(edges, "nodes.processed") = nodes # store set of processed nodes
                 return(edges)
             }
 
             ## get combinations
-            combinations = combn(nodes, 2) # all unique pairs of developers
+            combinations = combn(nodes, 2) # all unique pairs of authors
 
             ## construct edge list
             edges = apply(combinations, 2, function(comb) {
@@ -911,7 +911,7 @@ add.edges.for.bipartite.relation = function(net, net1.to.net2, network.conf) {
     vertex.sequence.for.edges = parallel::mcmapply(function(d, a.df) {
         a = a.df[["data.vertices"]]
         new.edges = lapply(a, function(vert) {
-            igraph::V(net)[d, vert] # get two vertices from source network:  c(developer, artifact)
+            igraph::V(net)[d, vert] # get two vertices from source network:  c(author, artifact)
         })
         return(new.edges)
     }, names(net1.to.net2), net1.to.net2)