GGPLOT grouped bar plot with facets: multiple aligned labels per bar? - r

Question

(EDIT: The issues raised in this post are solved below)
I am trying to get my bar plot to look a certain way. I hope you can follow me without really explaining what the purpose of this plot is or the context of the data, so I'll dive right in.
I'm using this code
ggplot(ds, aes(x=fraction, y=AUC)) +
geom_bar(aes(fill=factor(demographics, c("adjusted", "not adjusted"))), position=position_dodge(width=0.9), stat="identity") +
facet_grid(~FN, switch="x") +
geom_text(aes(label=round(AUC, 2), fontface="bold", vjust=-0.4), position=position_dodge(width=0.9), size=2.75) +
theme(legend.title=element_blank(), legend.position="bottom",
axis.text.y=element_blank(),
axis.title.x=element_blank(), axis.ticks.x=element_blank(), axis.ticks.y=element_blank(),
panel.background=element_blank())
to produce the following plot
What I want to see, however, is
that the AUC-values on top of the bars are correctly aligned with every single bar (grouped by "demographics" [adjusted/not adjusted]) and not with the middle of the group "fraction" (serum/plasma) as it is right now
that the two bars in the last facet "FN=5" each have only half of their current width (to match the width of the other bars in facets "FN=1" to "FN=4"
that there is another label printed vertically within every bar. This label is stored in the character variable "features" contained in the data set "ds" used to print the plot
ds <- structure(list(ds = structure(c(1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L,
2L, 3L, 3L, 3L, 3L, 4L, 4L, 4L, 4L, 4L, 1L, 1L, 1L, 1L, 2L, 2L,
2L, 2L, 2L, 3L, 3L, 3L, 3L, 4L, 4L, 4L, 4L, 4L), .Label = c("1",
"2", "3", "4"), class = "factor"), FN = structure(c(1L, 2L, 3L,
4L, 1L, 2L, 3L, 4L, 5L, 1L, 2L, 3L, 4L, 1L, 2L, 3L, 4L, 5L, 1L,
2L, 3L, 4L, 1L, 2L, 3L, 4L, 5L, 1L, 2L, 3L, 4L, 1L, 2L, 3L, 4L,
5L), .Label = c("FN=1", "FN=2", "FN=3", "FN=4", "FN=5"), class = "factor"),
fraction = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 1L,
1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L), .Label = c("serum",
"plasma"), class = "factor"), demographics = structure(c(1L,
1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 2L, 2L, 2L,
2L, 2L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 1L, 1L, 1L, 1L,
2L, 2L, 2L, 2L, 2L), .Label = c("not adjusted", "adjusted (age, sex, education, ethnicity, ApoE4)"
), class = "factor"), AUC = c(0.741264171455195, 0.749961174095356,
0.797639384997671, 0.763006678055599, 0.894704146606616,
0.9038670601025, 0.90169280944246, 0.912564062742662, 0.912564062742662,
0.672077922077922, 0.715367965367965, 0.746753246753247,
0.791125541125541, 0.804112554112554, 0.827922077922078,
0.829004329004329, 0.83982683982684, 0.83982683982684, 0.741264171455195,
0.749961174095356, 0.797639384997671, 0.763006678055599,
0.894704146606616, 0.9038670601025, 0.90169280944246, 0.912564062742662,
0.912564062742662, 0.672077922077922, 0.715367965367965,
0.746753246753247, 0.791125541125541, 0.804112554112554,
0.827922077922078, 0.829004329004329, 0.83982683982684, 0.83982683982684
), features = c("p21", "p3 + p21", "p3 + p7 + p19", "p1 + p8 + p21 + p23",
"p8", "p11 + p24", "p14 + p17 + p24", "p11 + p13 + p14 + p17",
"p11 + p13 + p14 + p17", "p13", "p9 + p17", "p9 + p14 + p17",
"p7 + p9 + p17 + p19", "p20", "p9 + p19", "p9 + p17 + p19",
"p4 + p8 + p9 + p17", "p4 + p8 + p9 + p17", "p21", "p3 + p21",
"p3 + p7 + p19", "p1 + p8 + p21 + p23", "p8", "p11 + p24",
"p14 + p17 + p24", "p11 + p13 + p14 + p17", "p11 + p13 + p14 + p17",
"p13", "p9 + p17", "p9 + p14 + p17", "p7 + p9 + p17 + p19",
"p20", "p9 + p19", "p9 + p17 + p19", "p4 + p8 + p9 + p17",
"p4 + p8 + p9 + p17")), .Names = c("ds", "FN", "fraction",
"demographics", "AUC", "features"), row.names = c(NA, -36L), class = "data.frame")
EDIT: Eventually, using the help in the comments section, I used this code
ggplot(ds, aes(x=fraction, y=AUC, fill=factor(demographics, c("adjusted (age, sex, education, ethnicity, ApoE4)", "not adjusted")))) +
geom_bar(position=position_dodge(width=0.9), stat="identity") +
facet_grid(~FN) +
geom_text(aes(label=round(AUC, 2), fontface="bold", vjust=1), position=position_dodge(width=0.9), size=2.75) +
geom_text(aes(y=0.3, label=features, fontface="bold"), color="white", position=position_dodge(width=0.9), angle=90, size=3) +
theme_bw() +
theme(legend.title=element_blank(), legend.position="bottom",
axis.text.y=element_blank(),
axis.title.x=element_blank(), axis.title.y=element_text(face="bold"), axis.ticks.x=element_blank(), axis.ticks.y=element_blank(),
panel.background=element_blank(), panel.grid.minor=element_blank(), panel.grid.major=element_blank())
to produce the results I initially wanted (I changed my mind on halving the width of the columns in facet "FN=5" though), looking like this
To increase readability, I decided to flip the plot using this code
ggplot(ds, aes(x=fraction, y=AUC, fill=factor(demographics, c("not adjusted", "adjusted (age, sex, education, ethnicity, ApoE4)")))) +
geom_bar(position=position_dodge(width=0.9), stat="identity") +
facet_grid(FN~.) +
geom_text(aes(y=AUC, label=round(AUC, 2), fontface="bold"), position=position_dodge(width=0.9), hjust=1.15, size=3.25) +
geom_text(aes(y=0.4, label=features, fontface="bold"), position=position_dodge(width=0.9), color="white", size=3) +
theme_bw() +
theme(legend.title=element_blank(), legend.position="bottom",
axis.title.x=element_text(face="bold"), axis.title.y=element_blank(),
axis.ticks.x=element_blank(), axis.ticks.y=element_blank(),
panel.background=element_blank(), panel.grid.minor=element_blank(), panel.grid.major=element_blank(),
strip.text.y = element_text(angle=0)) + coord_flip()
which results in a horizontal bar plot now looking like this (colors and order of legend keys are not quite right yet)