Search code examples
rggplot2ggraph

ggraph edges are connecting wrong?

I am working on generating a hierarchical edge plot where the edge's color/transparency/thickness varies by the column (pvalue) in my connect dataframe, however the color/transparency/thickness of the edges in the plot I generated don't always map to the values in column (pvalue). For example, subgroup1 and subgroup4 should have the strongest thickest connection (pvalue is E-280), when in fact they don't, rather the connection between subgroup3 and subgroup4 looks to be strongest.

enter image description here

This data generates a reproducible example:

> dput(vertices)
structure(list(name = structure(c(3L, 1L, 2L, 4L, 5L, 6L, 7L), .Label = c("gp1", 
"gp2", "origin", "subgroup1", "subgroup2", "subgroup3", "subgroup4"
), class = "factor"), id = c(NA, NA, NA, 1L, 2L, 3L, 4L), angle = c(NA, 
NA, NA, 0, -90, 0, -90), hjust = c(NA, NA, NA, 1, 1, 1, 1)), row.names = c(NA, 
-7L), class = "data.frame")

> dput(hierarchy)
structure(list(from = structure(c(3L, 3L, 1L, 1L, 2L, 2L), .Label = c("gp1", 
"gp2", "origin"), class = "factor"), to = structure(1:6, .Label = c("gp1", 
"gp2", "subgroup1", "subgroup2", "subgroup3", "subgroup4"), class = "factor")), class = "data.frame", row.names = c(NA, 
-6L))

> dput(connect)
structure(list(from = structure(c(1L, 1L, 2L, 3L, 1L, 2L, 3L, 
1L), .Label = c("subgroup1", "subgroup2", "subgroup3"), class = "factor"), 
    to = structure(c(1L, 2L, 2L, 1L, 3L, 3L, 3L, 3L), .Label = c("subgroup2", 
    "subgroup3", "subgroup4"), class = "factor"), pvalue = c(1.68e-204, 
    1.59e-121, 9.32e-73, 9.32e-73, 1.59e-21, 9.32e-50, 9.32e-40, 
    9.32e-280)), class = "data.frame", row.names = c(NA, -8L))

and this is the code I used to make this example plot:

from <- match( connect$from, vertices$name)
to <- match( connect$to, vertices$name)
col <- connect$pvalue


#Let's add information concerning the label we are going to add: angle, horizontal adjustement and potential flip
#calculate the ANGLE of the labels
vertices$id <- NA
myleaves <- which(is.na( match(vertices$name, hierarchy$from) ))
nleaves <- length(myleaves)
vertices$id[ myleaves ] <- seq(1:nleaves)
vertices$angle <- 90 - 360 * vertices$id / nleaves
# calculate the alignment of labels: right or left
# If I am on the left part of the plot, my labels have currently an angle < -90
vertices$hjust <- ifelse( vertices$id < 41, 1, 0)
# flip angle BY to make them readable
vertices$angle <- ifelse(vertices$angle < -90, vertices$angle+180, vertices$angle)

mygraph <- graph_from_data_frame( hierarchy, vertices=vertices )

ggraph(mygraph, layout = 'dendrogram', circular = TRUE) + 
  geom_node_point(aes(filter = leaf, x = x*1.05, y=y*1.05), size = 2, alpha = 0.8) +
  geom_conn_bundle(data = get_con(from = from, to = to, col = col), aes(colour=col, alpha = col, width = col))  +
  geom_node_text(aes(x = x*1.1, y=y*1.1, filter = leaf, label=name, angle = angle, hjust=hjust), size=3.5, alpha=0.6) +scale_edge_color_continuous(trans = "log",low="red", high="yellow")+ scale_edge_alpha_continuous(trans = "log",range = c(1, 0.1)) +scale_edge_width_continuous(trans = "log", range = c(4, 1))+
  theme_void()

I think there is wrong mapping somewhere but I can't figure out where. Thank you so much for your input!

Solution

  • I believe there is a bug in this library. Rearranging the input data by the column of choice (pvalue in my case) in an ascending order helped but did not solve the issue.

    connect_new <- arrange(connect, pvalue)

    and I found the solution in a github issue submitted by another user. The subgroups within each group need to be ordered alphabetically in the hierarchy and vertices file. In addition, in the connect dataframe, the subgroups need to be ordered following the same order in the hierarchy and vertices file. Thanks to zhuxr11