Issues aligning geom_point and geom_errorbar in R ggplot2
I am trying to create a plot in R ggplot2 where my data (see below) has multiple outcomes. For each outcome there are 5 possible methods that need to be plotted (although some are NA and should leave a blank space where this is the case). Each has an estimate, LCI, UCI and effect_type (can be OR, cont or RD). Where effect_type is OR I want to plot on the log scale, for cont or RD this needs a different scale. So I want 2 axes. Each method should be represented by a different shape in geom_point and each effect_type will also have a different colour.
I have tried to do this in the below code but having a range of issues. The main one being that the error bars do not align with the geom_points. Also some data (points and CIs) are not being plotted.
The plot and dataframe as well as the code are below.
Dataframe (df_filtered):
,Exposure,Outcome,Method,Model,Effect_estimate,LCI,UCI,p_value,units,effect_type,SD
1,Loneliness,Self harm,Obs,logistic,1.86,1.72,2.02,6.33E-53,yes/no,OR,NA
2,Loneliness,Self harm,Within-family,logistic,1.69,1.13,2.54,1.00E-02,yes/no,OR,NA
3,Loneliness,Self harm,Between-family,logistic,1.56,1.04,2.33,3.00E-02,yes/no,OR,NA
4,Loneliness,Self harm,2SMR,NA,NA,NA,NA,NA,NA,NA,NA
5,Loneliness,Suicide attempt,Obs,logistic,1.89,1.7,2.11,8.76E-31,yes/no,OR,NA
6,Loneliness,Suicide attempt,Within-family,logistic,1.85,1.06,3.24,3.00E-02,yes/no,OR,NA
7,Loneliness,Suicide attempt,Between-family,logistic,1.9,1.1,3.31,2.00E-02,yes/no,OR,NA
8,Loneliness,Suicide attempt,2SMR,IVW,2.42,1.46,3.99,5.57E-04,NA,OR,NA
9,Loneliness,Depression diagnosis,Obs,logistic,2.54,2.44,2.63,0.00E+00,yes/no,OR,NA
10,Loneliness,Depression diagnosis,Within-family,logistic,2.25,1.89,2.67,8.17E-20,yes/no,OR,NA
11,Loneliness,Depression diagnosis,Between-family,logistic,2.66,2.23,3.17,4.64E-28,yes/no,OR,NA
12,Loneliness,Depression diagnosis,2SMR,IVW,2.13,1.42,3.2,2.74E-04,NA,OR,NA
13,Loneliness,Anxiety diagnosis,Obs,logistic,1.9,1.82,1.98,1.73E-180,yes/no,OR,NA
14,Loneliness,Anxiety diagnosis,Within-family,logistic,1.69,1.39,2.06,1.93E-07,yes/no,OR,NA
15,Loneliness,Anxiety diagnosis,Between-family,logistic,2.01,1.65,2.46,5.56E-12,yes/no,OR,NA
16,Loneliness,Anxiety diagnosis,2SMR,IVW,1.91,0.84,4.36,1.20E-01,NA,OR,NA
17,Loneliness,Self harm,1SMR,ivreg,0.31,0.08,0.54,7.51E-03,yes/no,RD,NA
18,Loneliness,Suicide attempt,1SMR,ivreg,0.15,-0.00743,0.31,6.00E-02,yes/no,RD,NA
19,Loneliness,Depression diagnosis,1SMR,ivreg,0.35,0.15,0.55,5.18E-04,yes/no,RD,NA
20,Loneliness,Anxiety diagnosis,1SMR,ivreg,0.13,-0.04,0.29,1.30E-01,yes/no,RD,NA
21,Loneliness,Depression trait,Obs,linear,0.5815217,0.5570652,0.6032609,0.00E+00,score ranges from 0 to 27,cont,3.68
22,Loneliness,Depression trait,Within-family,linear,0.4809783,0.3831522,0.576087,1.39E-22,score ranges from 0 to 27,cont,3.68
23,Loneliness,Depression trait,Between-family,linear,0.6956522,0.6032609,0.7880435,1.20E-48,score ranges from 0 to 27,cont,3.68
24,Loneliness,Depression trait,2SMR,NA,NA,NA,NA,NA,NA,NA,3.68
25,Loneliness,Anxiety trait,Obs,linear,0.4705882,0.4470588,0.4941176,0.00E+00,score ranges from 0 to 21,cont,3.4
26,Loneliness,Anxiety trait,Within-family,linear,0.3735294,0.2764706,0.4676471,2.51E-14,score ranges from 0 to 21,cont,3.4
27,Loneliness,Anxiety trait,Between-family,linear,0.5735294,0.4823529,0.6676471,2.61E-33,score ranges from 0 to 21,cont,3.4
28,Loneliness,Anxiety trait,2SMR,NA,NA,NA,NA,NA,NA,NA,3.4
29,Loneliness,Positive affect,Obs,linear,-0.7027027,-0.7162162,-0.6891892,0.00E+00,rating ranges from 1 to 6,cont,0.74
30,Loneliness,Positive affect,Within-family,linear,-0.5945946,-0.6756757,-0.527027,9.56E-52,rating ranges from 1 to 6,cont,0.74
31,Loneliness,Positive affect,Between-family,linear,-0.7567568,-0.8378378,-0.6891892,5.58E-85,rating ranges from 1 to 6,cont,0.74
32,Loneliness,Positive affect,2SMR,NA,NA,NA,NA,NA,NA,NA,0.74
33,Loneliness,Meaning in Life,Obs,linear,-0.5180723,-0.5421687,-0.4939759,0.00E+00,rating ranges from 1 to 5,cont,0.83
34,Loneliness,Meaning in Life,Within-family,linear,-0.4096386,-0.4939759,-0.313253,2.67E-19,rating ranges from 1 to 5,cont,0.83
35,Loneliness,Meaning in Life,Between-family,linear,-0.5903614,-0.6746988,-0.5060241,4.59E-40,rating ranges from 1 to 5,cont,0.83
36,Loneliness,Meaning in Life,2SMR,NA,NA,NA,NA,NA,NA,NA,0.83
37,Loneliness,Wellbeing spectrum,Obs,NA,NA,NA,NA,NA,NA,NA,NA
38,Loneliness,Wellbeing spectrum,Within-family,NA,NA,NA,NA,NA,NA,NA,NA
39,Loneliness,Wellbeing spectrum,Between-family,NA,NA,NA,NA,NA,NA,NA,NA
40,Loneliness,Wellbeing spectrum,2SMR,IVW,-0.28,-0.32,-0.23,4.55E-33,NA,cont,NA
41,Loneliness,Positive affect,Obs,NA,NA,NA,NA,NA,NA,NA,0.74
42,Loneliness,Positive affect,Within-family,NA,NA,NA,NA,NA,NA,NA,0.74
43,Loneliness,Positive affect,Between-family,NA,NA,NA,NA,NA,NA,NA,0.74
44,Loneliness,Positive affect,2SMR,IVW,-0.472973,-0.6216216,-0.3243243,3.37E-10,NA,cont,0.74
45,Loneliness,Life satisfaction,Obs,NA,NA,NA,NA,NA,NA,NA,NA
46,Loneliness,Life satisfaction,Within-family,NA,NA,NA,NA,NA,NA,NA,NA
47,Loneliness,Life satisfaction,Between-family,NA,NA,NA,NA,NA,NA,NA,NA
48,Loneliness,Life satisfaction,2SMR,IVW,-0.47,-0.69,-0.24,4.28E-05,NA,cont,NA
49,Loneliness,Depression trait,1SMR,ivreg,3.5326087,2.0706522,4.9918478,2.16E-06,score ranges from 0 to 27,cont,3.68
50,Loneliness,Anxiety trait,1SMR,ivreg,2.2970588,1.1411765,3.4529412,9.81E-05,score ranges from 0 to 21,cont,3.4
51,Loneliness,Positive affect,1SMR,ivreg,-2.3783784,-3.2432432,-1.5135135,7.98E-08,rating ranges from 1 to 6,cont,0.74
52,Loneliness,Meaning in Life,1SMR,ivreg,-1.1084337,-2.0963855,-0.1204819,3.00E-02,rating ranges from 1 to 5,cont,0.83
Code:
library(ggplot2)
library(dplyr)
library(gridExtra)
library(cowplot)
df_filtered <- read.csv("data.csv", header=T)
custom_colors <- scale_color_manual(
values = c("HR" = "#97D9E3", "OR" = "#A59BEE", "RD" = "#FDB633", "cont" = "#F6A4B7")
)
custom_shapes <- c("Obs" = 0, "Between-family" = 1, "Within-family" = 2, "1SMR" = 10, "2SMR" = 11)
# Create the combined plot
p <- ggplot(df_filtered, aes(x = Outcome)) +
geom_hline(yintercept = 1, color = "#646363", linetype = "dashed") +
# Points for HR and OR
geom_point(data = df_filtered %>% filter(effect_type %in% c("OR")),
aes(y = Effect_estimate, shape = Method, color = effect_type),
size = 3, position = position_dodge(width = 0.5)) +
# Confidence intervals for HR and OR
geom_errorbar(data = df_filtered %>% filter(effect_type %in% c("OR")),
aes(ymin = LCI, ymax = UCI, color = effect_type),
width = 0.2, position = position_dodge(width = 0.5), size = 0.5) +
# Points for cont and RD on the second y-axis
geom_point(data = df_filtered %>% filter(effect_type %in% c("cont", "RD")),
aes(y = Effect_estimate, shape = Method, color = effect_type),
size = 3, position = position_dodge(width = 0.5)) +
# Confidence intervals for cont and RD on the second y-axis
geom_errorbar(data = df_filtered %>% filter(effect_type %in% c("cont", "RD")),
aes(ymin = LCI, ymax = UCI, color = effect_type),
width = 0.2, position = position_dodge(width = 0.5), size = 0.5) +
# Customize the theme
theme_minimal(base_size = 15) +
theme(panel.background = element_rect(fill = "gray90", color = NA), # Gray background
panel.grid.major = element_line(color = "white"), # White major grid lines
panel.grid.minor = element_line(color = "white", linewidth = 0.5),
legend.position = "right",
axis.text.x = element_text(angle = 45, hjust = 1)) +
labs(y = "Effect Estimate",
x = "Outcome") +
scale_fill_manual(values = custom_colors) + # Use custom colors for effect types
scale_shape_manual(values = custom_shapes) +
# Secondary y-axis for cont and RD
scale_y_continuous(trans = 'log10',
sec.axis = sec_axis(~ ., name = "Secondary Effect Estimate (cont, RD)"))
# Print the plot
print(p)
Plot currently output:
Solution
Dodging and other position adjustments work based on the group aesthetic.
From ?aes_group_order: For most applications the grouping is set implicitly by mapping one or more discrete variables to x, y, colour, fill, alpha, shape, size, and/or linetype.
So your points geoms are dodged based on Method, but your error bars geoms don't "see" that variable, so they are not dodged.
You can manually assign using group = in aes(), like below. In your example, each Outcome only has one effect_type, so it is arguably redundant, but it might be prudent to include effect_type in the group mapping, since it is one of the grouping variables for the point layers.
geom_errorbar(data = df_filtered %>% filter(effect_type %in% c("OR")),
aes(ymin = LCI, ymax = UCI, color = effect_type,
group = paste(effect_type, Method)),
width = 0.2, position = position_dodge(width = 0.5), size = 0.5) +
I typically use group = paste(GROUP_VAR1, GROUP_VAR2) or interaction(GROUP_VAR1, GROUP_VAR2)
df_filtered <- data.frame(
stringsAsFactors = FALSE,
Exposure = c("Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness","Loneliness","Loneliness",
"Loneliness"),
Outcome = c("Self harm","Self harm","Self harm",
"Self harm","Suicide attempt",
"Suicide attempt","Suicide attempt","Suicide attempt",
"Depression diagnosis",
"Depression diagnosis","Depression diagnosis",
"Depression diagnosis","Anxiety diagnosis",
"Anxiety diagnosis","Anxiety diagnosis",
"Anxiety diagnosis","Self harm",
"Suicide attempt","Depression diagnosis",
"Anxiety diagnosis","Depression trait",
"Depression trait","Depression trait","Depression trait",
"Anxiety trait","Anxiety trait",
"Anxiety trait","Anxiety trait",
"Positive affect","Positive affect","Positive affect",
"Positive affect","Meaning in Life",
"Meaning in Life","Meaning in Life",
"Meaning in Life","Wellbeing spectrum",
"Wellbeing spectrum","Wellbeing spectrum",
"Wellbeing spectrum","Positive affect",
"Positive affect","Positive affect",
"Positive affect","Life satisfaction",
"Life satisfaction","Life satisfaction",
"Life satisfaction","Depression trait",
"Anxiety trait","Positive affect","Meaning in Life"),
Method = c("Obs","Within-family","Between-family",
"2SMR","Obs","Within-family",
"Between-family","2SMR","Obs","Within-family",
"Between-family","2SMR","Obs",
"Within-family","Between-family","2SMR","1SMR",
"1SMR","1SMR","1SMR","Obs",
"Within-family","Between-family","2SMR","Obs",
"Within-family","Between-family","2SMR","Obs",
"Within-family","Between-family",
"2SMR","Obs","Within-family",
"Between-family","2SMR","Obs","Within-family",
"Between-family","2SMR","Obs","Within-family",
"Between-family","2SMR","Obs",
"Within-family","Between-family","2SMR","1SMR",
"1SMR","1SMR","1SMR"),
Model = c("logistic","logistic","logistic",NA,
"logistic","logistic","logistic","IVW",
"logistic","logistic","logistic","IVW",
"logistic","logistic","logistic","IVW",
"ivreg","ivreg","ivreg","ivreg",
"linear","linear","linear",NA,"linear",
"linear","linear",NA,"linear","linear",
"linear",NA,"linear","linear","linear",
NA,NA,NA,NA,"IVW",NA,NA,NA,"IVW",
NA,NA,NA,"IVW","ivreg","ivreg",
"ivreg","ivreg"),
Effect_estimate = c(1.86,1.69,1.56,NA,1.89,1.85,1.9,2.42,
2.54,2.25,2.66,2.13,1.9,1.69,2.01,
1.91,0.31,0.15,0.35,0.13,0.5815217,
0.4809783,0.6956522,NA,0.4705882,
0.3735294,0.5735294,NA,-0.7027027,
-0.5945946,-0.7567568,NA,-0.5180723,-0.4096386,
-0.5903614,NA,NA,NA,NA,-0.28,NA,NA,
NA,-0.472973,NA,NA,NA,-0.47,
3.5326087,2.2970588,-2.3783784,-1.1084337),
LCI = c(1.72,1.13,1.04,NA,1.7,1.06,1.1,1.46,
2.44,1.89,2.23,1.42,1.82,1.39,1.65,
0.84,0.08,-0.00743,0.15,-0.04,
0.5570652,0.3831522,0.6032609,NA,0.4470588,
0.2764706,0.4823529,NA,-0.7162162,
-0.6756757,-0.8378378,NA,-0.5421687,
-0.4939759,-0.6746988,NA,NA,NA,NA,-0.32,
NA,NA,NA,-0.6216216,NA,NA,NA,-0.69,
2.0706522,1.1411765,-3.2432432,
-2.0963855),
UCI = c(2.02,2.54,2.33,NA,2.11,3.24,3.31,
3.99,2.63,2.67,3.17,3.2,1.98,2.06,2.46,
4.36,0.54,0.31,0.55,0.29,0.6032609,
0.576087,0.7880435,NA,0.4941176,
0.4676471,0.6676471,NA,-0.6891892,-0.527027,
-0.6891892,NA,-0.4939759,-0.313253,
-0.5060241,NA,NA,NA,NA,-0.23,NA,NA,
NA,-0.3243243,NA,NA,NA,-0.24,
4.9918478,3.4529412,-1.5135135,-0.1204819),
p_value = c(6.33e-53,0.01,0.03,NA,8.76e-31,0.03,
0.02,0.000557,0,8.17e-20,4.64e-28,
0.000274,1.73e-180,1.93e-07,5.56e-12,0.12,
0.00751,0.06,0.000518,0.13,0,
1.39e-22,1.2e-48,NA,0,2.51e-14,2.61e-33,NA,
0,9.56e-52,5.58e-85,NA,0,2.67e-19,
4.59e-40,NA,NA,NA,NA,4.55e-33,NA,NA,
NA,3.37e-10,NA,NA,NA,4.28e-05,
2.16e-06,9.81e-05,7.98e-08,0.03),
units = c("yes/no","yes/no","yes/no",NA,"yes/no",
"yes/no","yes/no",NA,"yes/no",
"yes/no","yes/no",NA,"yes/no","yes/no",
"yes/no",NA,"yes/no","yes/no","yes/no",
"yes/no","score ranges from 0 to 27",
"score ranges from 0 to 27",
"score ranges from 0 to 27",NA,"score ranges from 0 to 21",
"score ranges from 0 to 21",
"score ranges from 0 to 21",NA,
"rating ranges from 1 to 6","rating ranges from 1 to 6",
"rating ranges from 1 to 6",NA,
"rating ranges from 1 to 5","rating ranges from 1 to 5",
"rating ranges from 1 to 5",NA,NA,NA,
NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,
"score ranges from 0 to 27",
"score ranges from 0 to 21","rating ranges from 1 to 6",
"rating ranges from 1 to 5"),
effect_type = c("OR","OR","OR",NA,"OR","OR","OR",
"OR","OR","OR","OR","OR","OR","OR",
"OR","OR","RD","RD","RD","RD","cont",
"cont","cont",NA,"cont","cont","cont",
NA,"cont","cont","cont",NA,"cont",
"cont","cont",NA,NA,NA,NA,"cont",NA,
NA,NA,"cont",NA,NA,NA,"cont",
"cont","cont","cont","cont"),
SD = c(NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,NA,
NA,NA,NA,NA,NA,NA,NA,NA,NA,3.68,
3.68,3.68,3.68,3.4,3.4,3.4,3.4,
0.74,0.74,0.74,0.74,0.83,0.83,0.83,
0.83,NA,NA,NA,NA,0.74,0.74,0.74,0.74,
NA,NA,NA,NA,3.68,3.4,0.74,0.83)
)
- Retrieving expected data.frame for testthat expectation
- How to use an anonymous function with pipe operator?
- R passing RSelenium driver environment as function argument
- Keep file names merging list with lapply?
- replace strings that contains one another
- Stop parsing out zeros after decimals in ggplot2's annotate
- Plot data.tree coloring and labelling by level
- Extracting the date and time out of a date and time format
- How to reshape a dataframe into wide format with specified column pairs
- Show partial segments removed after limit scales for `geom_pointrange`
- Control ggplot2 legend look without affecting the plot
- Concatenate strings over all rows in a single column of a data frame
- Is there some way to keep variable names from.SD+.SDcols together with non .SD variable names in data.table?
- How to create mutually dependent checkbox columns in a DT table?
- A problem with an stacked, grouped and faceted graph y gg plot
- Installing R on Linux: configure: error: libcurl >= 7.28.0 library and headers are required with support for https
- Create a column conditional on multiple other character and numerical columns
- data.frame to array (2 columns)
- Changing options() in a function environment without changing options() in global environment in R?
- pivot(), group_by() and summarise() with nested data
- Why legend is not showing up in plot layout using patchwork in R?
- in dplyr::mutate, dplyr::starts_with works for .before but not .after?
- Finding optimal cut-off points for dividing a variable into equal groups considering ties in which quantiles do not work
- How to add a legend to hline?
- R Shiny Suppress slider handle for sliderInput until click
- brmultinom brglm2 vifs and dredge dont work
- R, how to get current time miliseconds, without extra information ? (Sys.time(), format("%X"))
- Installing multiple versions of R
- Command to see 'R' path that RStudio is using
- Fast pairwise longest common substring from start