aritger
Reputation: 11
sankeyNetwork not recognizing customized colourScale
I am creating a plot using sankeyNetwork in R and am having issues with the colourScale option. I created a custom color scale, and the plot that is produced is so close to being correct, but somewhere within the sankeyNetwork function, it's breaking and associating two groups with the wrong color.
Here is the color scale:
colorJS <- "d3.scaleOrdinal() .domain([\" Akatopora spp.\", \"brown fine bryozoan\", \"Bugula neritina\", \"Bugulina californica\", \"Calliarthron tuberculosum\", \"Cauloramphus echinus\", \"Cellaria mandibulata\", \"Celleporaria brunnea\", \"Ceramiales\", \"Cheilostomatida\", \"Chondracanthus spinosus\", \"Chondracanthus spp.\", \"Corallina officinalisvarchilensis\", \"Corallinales\", \"cream encrusting bryozoan\", \"Crisia occidentalis\", \"Cyclostomatida\", \"Diaperoforma californica\", \"Ectoprocta\", \"Gelidiales\", \"Gelidium robustum\", \"Gigartinales\", \"Gloiocladia laciniata\", \"Haywardipora spp.\", \"Jellyella white\", \"Lithothrix aspergillum\", \"Microcladia coulteri\", \"nongeniculate pink\", \"nongeniculate red\", \"other_algae\", \"Parasmittina tubulata\", \"peach encrusting bryozoan\", \"red_fil_alg\", \"red_turf_alg\", \"Rhodophyta\", \"Rhodymenia spp.\", \"Rhodymeniales\", \"Sarcodiotheca furcata\", \"Thalamoporella californica \"]) .range([\" #FF8C00\", \"#FF8C00\", \"#FF8C00\", \"#FF8C00\", \"#DB7093\", \"#FF8C00\", \"#FF8C00\", \"#FF8C00\", \"#DB7093\", \"#FF8C00\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#FF8C00\", \"#FF8C00\", \"#FF8C00\", \"#FF8C00\", \"#FF8C00\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#FF8C00\", \"#FF8C00\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#FFB90F\", \"#FF8C00\", \"#FF8C00\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#DB7093\", \"#FF8C00 \"])"
Take note of the fact that "red_fil_alg" and "red_turf_alg" are associated with the pink color.
And here is the code for the sankeyNetwork:
sankeyNetwork(Links = links, Nodes = nodes,
Source = "source", Target = "target",
Value = "value", NodeID = "name",
fontSize = 20, nodeWidth = 5,
colourScale = colorJS,
LinkGroup="group", NodeGroup = NULL)
Which produces this figure:
If it isn't obvious, the links between "other_algae" and "red_fil_alg" and "red_turf_alg" should be pink, not orange. I attempted to solve the problem by removing the spaces in the names of the nodes, but that didn't work. I checked to make sure the names of the nodes and the links match up using setdiff().
Here are the links and the nodes, if they are useful:
> links
source target value group
1 18 9 1.000000 Ectoprocta
2 18 9 1.500000 Ectoprocta
3 18 9 15.900000 Ectoprocta
4 18 9 1.000000 Ectoprocta
5 18 9 1.000000 Ectoprocta
6 18 9 1.363636 Ectoprocta
7 18 9 1.357143 Ectoprocta
8 18 9 3.025641 Ectoprocta
9 18 9 3.333333 Ectoprocta
10 18 9 1.666667 Ectoprocta
11 18 9 4.934783 Ectoprocta
12 18 9 4.500000 Ectoprocta
13 18 16 1.125000 Ectoprocta
14 18 16 1.250000 Ectoprocta
15 18 16 2.818182 Ectoprocta
16 34 8 1.000000 Rhodophyta
17 34 13 1.000000 Rhodophyta
18 34 13 8.333333 Rhodophyta
19 34 13 1.000000 Rhodophyta
20 34 13 24.655172 Rhodophyta
21 34 13 8.243902 Rhodophyta
22 34 19 1.000000 Rhodophyta
23 34 21 2.000000 Rhodophyta
24 34 21 3.000000 Rhodophyta
25 34 21 2.000000 Rhodophyta
26 34 29 2.555556 Rhodophyta
27 34 29 5.921569 Rhodophyta
28 34 36 1.769231 Rhodophyta
29 34 36 4.187500 Rhodophyta
30 9 0 1.000000 Cheilostomatida
31 9 2 1.500000 Cheilostomatida
32 9 3 15.900000 Cheilostomatida
33 9 5 1.000000 Cheilostomatida
34 9 6 1.000000 Cheilostomatida
35 9 7 1.363636 Cheilostomatida
36 9 14 1.357143 Cheilostomatida
37 9 23 3.025641 Cheilostomatida
38 9 24 3.333333 Cheilostomatida
39 9 30 1.666667 Cheilostomatida
40 9 31 4.934783 Cheilostomatida
41 9 38 4.500000 Cheilostomatida
42 16 1 1.125000 Cyclostomatida
43 16 15 1.250000 Cyclostomatida
44 16 17 2.818182 Cyclostomatida
45 8 26 1.000000 Ceramiales
46 13 4 1.000000 Corallinales
47 13 12 8.333333 Corallinales
48 13 25 1.000000 Corallinales
49 13 27 24.655172 Corallinales
50 13 28 8.243902 Corallinales
51 19 20 1.000000 Gelidiales
52 21 10 2.000000 Gigartinales
53 21 11 3.000000 Gigartinales
54 21 37 2.000000 Gigartinales
55 29 32 2.555556 other_algae
56 29 33 5.921569 other_algae
57 36 22 1.769231 Rhodymeniales
58 36 35 4.187500 Rhodymeniales
> nodes
name
1 Akatopora spp.
2 brown fine bryozoan
3 Bugula neritina
4 Bugulina californica
5 Calliarthron tuberculosum
6 Cauloramphus echinus
7 Cellaria mandibulata
8 Celleporaria brunnea
9 Ceramiales
10 Cheilostomatida
11 Chondracanthus spinosus
12 Chondracanthus spp.
13 Corallina officinalisvarchilensis
14 Corallinales
15 cream encrusting bryozoan
16 Crisia occidentalis
17 Cyclostomatida
18 Diaperoforma californica
19 Ectoprocta
20 Gelidiales
21 Gelidium robustum
22 Gigartinales
23 Gloiocladia laciniata
24 Haywardipora spp.
25 Jellyella white
26 Lithothrix aspergillum
27 Microcladia coulteri
28 nongeniculate pink
29 nongeniculate red
30 other_algae
31 Parasmittina tubulata
32 peach encrusting bryozoan
33 red_fil_alg
34 red_turf_alg
35 Rhodophyta
36 Rhodymenia spp.
37 Rhodymeniales
38 Sarcodiotheca furcata
39 Thalamoporella californica
Upvotes: 1
Views: 260
Answers (3)
hnnh
Reputation: 21
I was facing a similar issue. Replacing all white spaces in the node names through non-breaking white spaces solved the problem.
Upvotes: 0
CJ Yetman
Reputation: 8848
There's a mistake in your list of colors. You're explicitly setting the color for "other_algae" to "#FFB90F". If you fix that, it will have the right color...
links <- read.table(header = TRUE, text = "
source target value group
18 9 1.000000 Ectoprocta
18 9 1.500000 Ectoprocta
18 9 15.900000 Ectoprocta
18 9 1.000000 Ectoprocta
18 9 1.000000 Ectoprocta
18 9 1.363636 Ectoprocta
18 9 1.357143 Ectoprocta
18 9 3.025641 Ectoprocta
18 9 3.333333 Ectoprocta
18 9 1.666667 Ectoprocta
18 9 4.934783 Ectoprocta
18 9 4.500000 Ectoprocta
18 16 1.125000 Ectoprocta
18 16 1.250000 Ectoprocta
18 16 2.818182 Ectoprocta
34 8 1.000000 Rhodophyta
34 13 1.000000 Rhodophyta
34 13 8.333333 Rhodophyta
34 13 1.000000 Rhodophyta
34 13 24.655172 Rhodophyta
34 13 8.243902 Rhodophyta
34 19 1.000000 Rhodophyta
34 21 2.000000 Rhodophyta
34 21 3.000000 Rhodophyta
34 21 2.000000 Rhodophyta
34 29 2.555556 Rhodophyta
34 29 5.921569 Rhodophyta
34 36 1.769231 Rhodophyta
34 36 4.187500 Rhodophyta
9 0 1.000000 Cheilostomatida
9 2 1.500000 Cheilostomatida
9 3 15.900000 Cheilostomatida
9 5 1.000000 Cheilostomatida
9 6 1.000000 Cheilostomatida
9 7 1.363636 Cheilostomatida
9 14 1.357143 Cheilostomatida
9 23 3.025641 Cheilostomatida
9 24 3.333333 Cheilostomatida
9 30 1.666667 Cheilostomatida
9 31 4.934783 Cheilostomatida
9 38 4.500000 Cheilostomatida
16 1 1.125000 Cyclostomatida
16 15 1.250000 Cyclostomatida
16 17 2.818182 Cyclostomatida
8 26 1.000000 Ceramiales
13 4 1.000000 Corallinales
13 12 8.333333 Corallinales
13 25 1.000000 Corallinales
13 27 24.655172 Corallinales
13 28 8.243902 Corallinales
19 20 1.000000 Gelidiales
21 10 2.000000 Gigartinales
21 11 3.000000 Gigartinales
21 37 2.000000 Gigartinales
29 32 2.555556 other_algae
29 33 5.921569 other_algae
36 22 1.769231 Rhodymeniales
36 35 4.187500 Rhodymeniales
")
nodes <- read.csv(header = TRUE, text ="
name
Akatopora spp.
brown fine bryozoan
Bugula neritina
Bugulina californica
Calliarthron tuberculosum
Cauloramphus echinus
Cellaria mandibulata
Celleporaria brunnea
Ceramiales
Cheilostomatida
Chondracanthus spinosus
Chondracanthus spp.
Corallina officinalisvarchilensis
Corallinales
cream encrusting bryozoan
Crisia occidentalis
Cyclostomatida
Diaperoforma californica
Ectoprocta
Gelidiales
Gelidium robustum
Gigartinales
Gloiocladia laciniata
Haywardipora spp.
Jellyella white
Lithothrix aspergillum
Microcladia coulteri
nongeniculate pink
nongeniculate red
other_algae
Parasmittina tubulata
peach encrusting bryozoan
red_fil_alg
red_turf_alg
Rhodophyta
Rhodymenia spp.
Rhodymeniales
Sarcodiotheca furcata
Thalamoporella californica
")
colors <- read.csv(header = TRUE, strip.white = TRUE, text = "
domain , range
Akatopora spp. , #FF8C00
brown fine bryozoan , #FF8C00
Bugula neritina , #FF8C00
Bugulina californica , #FF8C00
Calliarthron tuberculosum , #DB7093
Cauloramphus echinus , #FF8C00
Cellaria mandibulata , #FF8C00
Celleporaria brunnea , #FF8C00
Ceramiales , #DB7093
Cheilostomatida , #FF8C00
Chondracanthus spinosus , #DB7093
Chondracanthus spp. , #DB7093
Corallina officinalisvarchilensis , #DB7093
Corallinales , #DB7093
cream encrusting bryozoan , #FF8C00
Crisia occidentalis , #FF8C00
Cyclostomatida , #FF8C00
Diaperoforma californica , #FF8C00
Ectoprocta , #FF8C00
Gelidiales , #DB7093
Gelidium robustum , #DB7093
Gigartinales , #DB7093
Gloiocladia laciniata , #DB7093
Haywardipora spp. , #FF8C00
Jellyella white , #FF8C00
Lithothrix aspergillum , #DB7093
Microcladia coulteri , #DB7093
nongeniculate pink , #DB7093
nongeniculate red , #DB7093
other_algae , #DB7093
Parasmittina tubulata , #FF8C00
peach encrusting bryozoan , #FF8C00
red_fil_alg , #DB7093
red_turf_alg , #DB7093
Rhodophyta , #DB7093
Rhodymenia spp. , #DB7093
Rhodymeniales , #DB7093
Sarcodiotheca furcata , #DB7093
Thalamoporella californica , #FF8C00
")
library(networkD3)
colorJS <- paste0("d3.scaleOrdinal().domain([\"", paste(colors$domain, collapse = "\", \""), "\"]).range([\"", paste(colors$range, collapse = "\", \""), "\"])")
sankeyNetwork(Links = links, Nodes = nodes,
Source = "source", Target = "target",
Value = "value", NodeID = "name",
fontSize = 20, nodeWidth = 5,
colourScale = colorJS,
LinkGroup="group", NodeGroup = NULL)
Upvotes: 0
Momemor
Reputation: 83
I have tested several changes to the code to narrow down the glitch (see below). None of them solve the problem, so the most likely cause is still a bug in the library, more specifically on how sankeyNetwork() runs through the ordinal scale of domains and ranges.
I suggest you flag this issue on the developer's space.
https://github.com/christophergandrud/networkD3/issues?page=2&q=is%3Aissue+is%3Aopen
As a matter of fact, there is a similar issue reported: https://github.com/christophergandrud/networkD3/issues/242 The solution to that one was removing spaces in the strings passed on to D3 as domain. However, I have tried that one already without success:
- replaced spaces and underscores with "-"
- removed the dots after the "spp"s
- introduced a zero-based index in the nodes df to rule out this issue
library(networkD3)
library(tidyverse)
nodes <- data.frame(names = c("Akatopora-spp","brown-fine-bryozoan","Bugula-neritina","Bugulina-californica","Calliarthron-tuberculosum","Cauloramphus-echinus","Cellaria-mandibulata","Celleporaria-brunnea","Ceramiales","Cheilostomatida","Chondracanthus-spinosus","Chondracanthus-spp","Corallina-officinalisvarchilensis","Corallinales","cream-encrusting-bryozoan","Crisia-occidentalis","Cyclostomatida","Diaperoforma-californica","Ectoprocta","Gelidiales","Gelidium-robustum","Gigartinales","Gloiocladia-laciniata","Haywardipora-spp","Jellyella-white","Lithothrix-aspergillum","Microcladia-coulteri","nongeniculate-pink","nongeniculate-red","Other-algae","Parasmittina-tubulata","peach-encrusting-bryozoan","red-fil-algae","red-turf-algae","Rhodophyta","Rhodymenia-spp","Rhodymeniales","Sarcodiotheca-furcata","Thalamoporella-californica"))
links <- data.frame(matrix(c(1,18,9,1.000000,"Ectoprocta",2,18,9,1.500000,"Ectoprocta",3,18,9,15.900000,"Ectoprocta",4,18,9,1.000000,"Ectoprocta",5,18,9,1.000000,"Ectoprocta",6,18,9,1.363636,"Ectoprocta",7,18,9,1.357143,"Ectoprocta",8,18,9,3.025641,"Ectoprocta",9,18,9,3.333333,"Ectoprocta",10,18,9,1.666667,"Ectoprocta",11,18,9,4.934783,"Ectoprocta",12,18,9,4.500000,"Ectoprocta",13,18,16,1.125000,"Ectoprocta",14,18,16,1.250000,"Ectoprocta",15,18,16,2.818182,"Ectoprocta",16,34,8,1.000000,"Rhodophyta",17,34,13,1.000000,"Rhodophyta",18,34,13,8.333333,"Rhodophyta",19,34,13,1.000000,"Rhodophyta",20,34,13,24.655172,"Rhodophyta",21,34,13,8.243902,"Rhodophyta",22,34,19,1.000000,"Rhodophyta",23,34,21,2.000000,"Rhodophyta",24,34,21,3.000000,"Rhodophyta",25,34,21,2.000000,"Rhodophyta",26,34,29,2.555556,"Rhodophyta",27,34,29,5.921569,"Rhodophyta",28,34,36,1.769231,"Rhodophyta",29,34,36,4.187500,"Rhodophyta",30,9,0,1.000000,"Cheilostomatida",31,9,2,1.500000,"Cheilostomatida",32,9,3,15.900000,"Cheilostomatida",33,9,5,1.000000,"Cheilostomatida",34,9,6,1.000000,"Cheilostomatida",35,9,7,1.363636,"Cheilostomatida",36,9,14,1.357143,"Cheilostomatida",37,9,23,3.025641,"Cheilostomatida",38,9,24,3.333333,"Cheilostomatida",39,9,30,1.666667,"Cheilostomatida",40,9,31,4.934783,"Cheilostomatida",41,9,38,4.500000,"Cheilostomatida",42,16,1,1.125000,"Cyclostomatida",43,16,15,1.250000,"Cyclostomatida",44,16,17,2.818182,"Cyclostomatida",45,8,26,1.000000,"Ceramiales",46,13,4,1.000000,"Corallinales",47,13,12,8.333333,"Corallinales",48,13,25,1.000000,"Corallinales",49,13,27,24.655172,"Corallinales",50,13,28,8.243902,"Corallinales",51,19,20,1.000000,"Gelidiales",52,21,10,2.000000,"Gigartinales",53,21,11,3.000000,"Gigartinales",54,21,37,2.000000,"Gigartinales",55,29,32,2.555556,"other_algae",56,29,33,5.921569,"other_algae",57,36,22,1.769231,"Rhodymeniales",58,36,35,4.187500,"Rhodymeniales"), ncol = 5, byrow = TRUE), stringsAsFactors = FALSE) %>%
rename(ID = X1, source = X2, target = X3, value = X4, group = X5) %>%
mutate(value = as.numeric(value)) # input links df
links[links$group == "other-algae",]$group <- "Other-algae"
nodes <- nodes %>%
rowid_to_column(., var = "ID") %>% # introduce index
mutate(ID = ID-1) # zero-based index
colorJS <- "d3.scaleOrdinal() .domain([\"Akatopora-spp\",\"brown-fine-bryozoan\",\"Bugula-neritina\",\"Bugulina-californica\",\"Calliarthron-tuberculosum\",\"Cauloramphus-echinus\",\"Cellaria-mandibulata\",\"Celleporaria-brunnea\",\"Ceramiales\",\"Cheilostomatida\",\"Chondracanthus-spinosus\",\"Chondracanthus-spp\",\"Corallina-officinalisvarchilensis\",\"Corallinales\",\"cream-encrusting-bryozoan\",\"Crisia-occidentalis\",\"Cyclostomatida\",\"Diaperoforma-californica\",\"Ectoprocta\",\"Gelidiales\",\"Gelidium-robustum\",\"Gigartinales\",\"Gloiocladia-laciniata\",\"Haywardipora-spp\",\"Jellyella-white\",\"Lithothrix-aspergillum\",\"Microcladia-coulteri\",\"nongeniculate-pink\",\"nongeniculate-red\",\"Other-algae\",\"Parasmittina-tubulata\",\"peach-encrusting-bryozoan\",\"red-fil-algae\",\"red-turf-algae\",\"Rhodophyta\",\"Rhodymenia-spp\",\"Rhodymeniales\",\"Sarcodiotheca-furcata\",\"Thalamoporella-californica\"]) .range([\" #FF8C00\",\"#FF8C00\",\"#FF8C00\",\"#FF8C00\",\"#DB7093\",\"#FF8C00\",\"#FF8C00\",\"#FF8C00\",\"#DB7093\",\"#FF8C00\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#FF8C00\",\"#FF8C00\",\"#FF8C00\",\"#FF8C00\",\"#FF8C00\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#FF8C00\",\"#FF8C00\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#FFB90F\",\"#FF8C00\",\"#FF8C00\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#DB7093\",\"#FF8C00\"])"
sankeyNetwork(Links = links, Nodes = nodes,
Source = "source", Target = "target",
Value = "value", NodeID = "names",
fontSize = 20, nodeWidth = 5,
colourScale = colorJS,
LinkGroup="group", NodeGroup = NULL)
Output:
Oddly enough, the links in question turn out dark orange -same as the top category- and not light orange as in your original output:
I also tried a standard D3 colour scale colourScale = JS("d3.scaleOrdinal(d3.schemeCategory20)") and it seems to run fine through your original node categories:
Furthermore, the node-link structure seems to be correct, as seen when building a simple network (the problematic nodes are highlighted):
simpleNetwork(links, Source = "source",
Target = "target",
fontSize = 14,
fontFamily = "serif",
linkColour = "#666",
nodeColour = "#69b3a2",
opacity = 0.9,
zoom = T)
Finally I tried to run the domain through a three-colour scale to watch the behaviour. The "Ceramiales" category is assigned the same colour as "Other-algae" so the issue is not directly related to the string in the problematic category.
colorJS2 <- "d3.scaleOrdinal() .domain([\"Akatopora spp.\", \"brown fine bryozoan\", \"Bugula neritina\", \"Bugulina californica\", \"Calliarthron tuberculosum\", \"Cauloramphus echinus\", \"Cellaria mandibulata\", \"Celleporaria brunnea\", \"Ceramiales\", \"Cheilostomatida\", \"Chondracanthus spinosus\", \"Chondracanthus spp.\", \"Corallina officinalisvarchilensis\", \"Corallinales\", \"cream encrusting bryozoan\", \"Crisia occidentalis\", \"Cyclostomatida\", \"Diaperoforma californica\", \"Ectoprocta\", \"Gelidiales\", \"Gelidium robustum\", \"Gigartinales\", \"Gloiocladia laciniata\", \"Haywardipora spp.\", \"Jellyella white\", \"Lithothrix aspergillum\", \"Microcladia coulteri\", \"nongeniculate pink\", \"nongeniculate red\", \"other_algae\", \"Parasmittina tubulata\", \"peach encrusting bryozoan\", \"red_fil_alg\", \"red_turf_alg\", \"Rhodophyta\", \"Rhodymenia spp.\", \"Rhodymeniales\", \"Sarcodiotheca furcata\", \"Thalamoporella californica\"]) .range([\" #222222\", \"#DB7093\", \"#FFB90F\"])"
sankeyNetwork(Links = links, Nodes = nodes,
Source = "source", Target = "target",
Value = "value", NodeID = "names",
fontSize = 20, nodeWidth = 5,
colourScale = colorJS2,
LinkGroup="group", NodeGroup = NULL)
Output:
Hence my guess about a glitch in the way the function runs through the js domain and range strings. Good luck.
Upvotes: 0
Related Questions
- Modify networkD3 sankey plot with user-defined colors
- R networkD3 sankey - change the colors for all links to blue
- Adding a color legend with JavaScript to a networkD3 sankeyNetwork() in R
- how to fix incorrect color for nodes in networkD3 (sankeyNetwork)?
- R Specify colors in sankeyNetwork
- change the color of links in networkD3 sankeyNetwork using R
- networkD3: How to assign color in sankey chart to the node without output
- Sankey Diagram with networkD3 package will not plot
- How to color groups in networkD3's sankeyNetwork?
- Own colour range for Sankey Diagram with networkD3 package in R