ggraph网络图:指定节点坐标

时间:2017-02-05 04:51:33

标签: r networking cartesian-coordinates ggraph ggforce

尝试使用ggraph包绘制网络并指定每个节点的坐标。虽然我可以使用igraph包实现此功能 - 但我无法使用ggraph包执行此操作。

# reproducible example to generate a random graph
library(igraph)
g1 <- erdos.renyi.game(20, 1/2)
plot(g1)

# function to produce coordinates for each node in order of the node
# degree (number of links per node)
coord <- function(g){   
    n.nod <- length(V(g))
    mat.c <- matrix(0, nrow = n.nod, ncol = 2)
    deg <- degree(g)
    uniq.deg <- unique(deg)
    min.d <- min(deg); max.d<- max(deg)
    spa <- 10/(max.d - min.d) # how much to increment 0 to 10 space along y-axis

    divi.y.1 <- seq(from = min.d, to=max.d, by = 1)
    divi.y.2 <- seq(from = 0, to=10, by = spa)  # both have same length

    ind.x <- c(); ind.x[1] = 0
    for(x in 2:n.nod){
      ind.x[x] <- ind.x[x-1] + 0.1
      if(ind.x[x] >= 10){
        ind.x[x] = ind.x[x-1] - 0.1
      }  
    } 
 d1 <- data.frame(divi.y.1, divi.y.2)

 # plotting space of grid is (0,0), (0, 10), (10, 10), (10, 0)
 for(i in 1:n.nod){
    # y-axis in order
    inD <-  which(d1$divi.y.1 == deg[i])
    mat.c[i, 2] <- d1[inD,2]
 } 
 mat.c[, 1] <- ind.x
 return(data.frame(mat.c))
}

这是绘制igraph对象的“老式”方式:

# plot igraph object - the old fashion way
x11()
plot(g1, layout = coord(g1), rescale = T, frame=T,
 vertex.frame.color = "black", edge.color = "lightsteelblue",
 edge.width = 1, vertex.label.cex = 1,
 vertex.label.color = "black", 
 main = "Nodes in order of degree: top have more links, bottom fewer links")

ggraph文档的链接是here。另请参阅ggraph的GitHub存储库以获取软件包安装说明(需要&gt; R.3.3版本)。下面是工作的ggraph图(但我没有指定每个节点的坐标):

library(ggraph)
V(g1)$NaMe <- seq(1:20)

x11()
   ggraph(g1, 'igraph', algorithm = 'kk') +  
   geom_edge_link(colour = "black", alpha = 0.8, show.legend = F) + 
   geom_node_label(aes(label = NaMe)) + ggtitle("ggraph plot: How to allocate coordinate for each node?") + 
   ggforce::theme_no_axes()

这是我尝试使用每个节点的指定坐标制作ggraph图。在ggraph()中类似的示例和早期尝试将节点的坐标传递到绘图上之后,我尝试了以下操作:

g <- make_ring(10) + make_full_graph(5)
coords <- layout_(g, as_star())
plot(g, layout = coords)
# try to replicate this example:
coords2 <- add_layout_(g1, coord(g1))

还尝试使用this function。这很困难,因为文档中没有示例。

Lay <- layout_igraph_manual(g1, coord(g1))
Lay <- layout_igraph_igraph(g1, coord(g1))

x11()
ggraph(g1, 'igraph', algorithm = 'kk' ) + add_layout_(Lay) +
# layout_igraph_circlepack() + 
geom_edge_link(colour = "black", alpha = 0.8, show.legend = F) + 
geom_node_label(aes(label = NaMe)) + ggtitle("ggraph plot: Cannot allocate    coordinate for each node") + 
ggforce::theme_no_axes()

3 个答案:

答案 0 :(得分:2)

我认为作者无意直接致电layout_igraph_manual。如果将布局命名为“manual”,则可以传递手动布局顶点所需的参数。如果我理解正确的话,这就让我了解你的目标:

coords <- coord(g1)
names(coords) <- c("x","y")
coords <- as.data.frame(coords)

ggraph(g1, layout = "manual", circular = FALSE, node.positions = coords)+
  geom_edge_link(colour = "black", alpha = 0.8, show.legend = F) + 
  geom_node_label(aes(label = NaMe)) + ggtitle("ggraph plot: How to allocate coordinate for each node?") + 
  ggforce::theme_no_axes()

根据文档,您的手动布局是一个数据框并且包含xy列非常重要。我已经对coord()函数的输出执行了此操作。您可能只想更新它以返回该表单中的数据。

答案 1 :(得分:1)

使用ggraph时,通常不必考虑节点坐标 - 它们是根据所选布局自动提供的。如果您想事先自己计算布局,只需使用手动布局 - 查看layout_igraph_manual的文档

答案 2 :(得分:1)

谢谢@ThomasP85(哇!ggraph的开发人员)。在这个例子中,我想指定每个节点的坐标。我已经尝试了layout_igraph_manuallayout_add_layout_和其他许多人,似乎没有用。

这是什么工作

co2<- data.frame(coord(g1))  # my mistake above, the output of coord() should be a matrix
colnames(co2)<- c("x", "y")
Lay<- createLayout(g1, layout = "nicely") # try any layout
Lay$x<- co2$x  # <== now overwrite their coordinates
Lay$y<- co2$y


x11()  # <==== this is ggraph version of the igraph plot above 
ggraph(graph=g1, data=Lay) + 
geom_edge_link(colour = "black", alpha = 0.5, show.legend = F) + 
geom_node_label(aes(label = NaMe)) + ggtitle("Nodes with higher degree at top, lower degree nodes at the bottom") + 
ggforce::theme_no_axes()

@ ThomasP85如果您可以使用layout_igraph_manual提供上述代码的工作示例,我们将不胜感激。