Ordering the x-axis in an R graph - r

Question

I have a data.frame that looks like:
gvs order labels
1 -2.3321916 1 Adygei
2 -1.4996229 5 Basque
3 1.7958170 15 French
4 2.5543214 19 Italian
5 -2.7758460 33 Orcadian
6 -1.9659984 39 Russian
7 2.1239768 41 Sardinian
8 -1.8515908 47 Tuscan
9 -1.5597359 6 Bedouin
10 -1.2534511 14 Druze
11 -0.1625003 31 Mozabite
12 -1.0265275 35 Palestinian
13 -0.8519079 2 Balochi
14 -2.4279528 8 Brahui
15 -3.1717421 9 Burusho
16 -0.9258497 17 Hazara
17 -1.2207974 21 Kalash
18 -1.0325107 24 Makrani
19 -3.2102686 37 Pathan
20 -0.9377928 43 Sindhi
21 -1.7657017 48 Uygurf
22 -0.5058627 10 Cambodian
23 -0.7819299 12 Dai
24 -1.4095947 13 Daur
25 2.2810477 16 Han
26 -0.9007551 18 Hezhen
27 2.6614486 20 Japanese
28 -0.9441980 23 Lahu
29 -0.7237586 29 Miao
30 -0.9452944 30 Mongola
31 -1.2035258 32 Naxi
32 -0.7703779 34 Oroqen
33 -3.0895998 42 She
34 -0.7037952 45 Tu
35 -1.9311354 46 Tujia
36 -0.5423822 49 Xibo
37 -1.6244801 50 Yakut
38 -0.9049735 51 Yi
39 -2.6491331 11 Colombian
40 2.3706977 22 Karitiana
41 -2.7590587 26 Maya
42 -0.9614190 38 Pima
43 -1.6961014 44 Surui
44 -0.8449225 28 Melanesian
45 -1.1163019 36 Papuan
46 -0.9298674 3 BantuKenya
47 -2.8859587 4 BantuSouthAfrica
48 -1.4494841 7 BiakaPygmy
49 -0.7381369 25 Mandenka
50 -0.5644325 27 MbutiPygmy
51 -0.9195156 40 San
52 2.0949378 52 Yoruba
I would like to graph the column gvs along the x-axis in the order of the column order, and then have the label for each point along the x-axis to be from the column labels. Does anyone know how this is done? I want the graph to look like a less colorful version of the graphs in figure-5 in this paper http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1004412

Answer 1

Based on your comments, it looks like (1) labels doesn't correspond to gvs and order, and (2) if I sort the first two columns by order, the data frame will be ordered properly. Please let me know if this is not correct.
Sort first two columns by order, leaving third column alone:
df[,c("gvs","order")] = df[order(df$order), c("gvs","order")]
Set the ordering of labels based on the current ordering of labels in the sample data frame:
df$labels = factor(df$labels, levels=df$labels)
Add a grouping variable for region. I did this by creating a new group each time the alphabetic ordering of labels went "backwards". The regions are just numbers here, but you can give them descriptive names if you want to use them:
df$group = c(0, cumsum(diff(match(substr(df$labels,1,1), LETTERS)) < 0))
Add fake p-values (since point size was based on p-value in the graph you linked to):
set.seed(595)
df$p.value = runif(nrow(df), 0, 0.5)
Plot the data, including a different color for each regional group, point-size based on p-value, and black borders around points with p < 0.05. geom_line add the regional means:
library(dplyr)
ggplot(df, aes(labels, gvs, size=p.value, fill=factor(group))) +
geom_line(data=df %>% group_by(group) %>% mutate(gvs=mean(gvs)),
aes(group=group, colour=factor(group)), size=0.8,alpha=0.5) +
geom_point(pch=21, stroke=1, aes(color=p.value<0.05)) +
theme_bw() +
theme(axis.text.x=element_text(angle=-90, hjust=0, vjust=0.5),
panel.grid.major=element_blank(),
panel.grid.minor=element_blank()) +
scale_size_continuous(name="p values", limits=c(0, 0.5), breaks=seq(0,1,0.1), range=c(4,1)) +
scale_color_manual(values=c(hcl(seq(15,375,length.out=8),100,65)[1:7],NA,"black")) +
labs(x="Language", fill="Region") +
guides(colour=FALSE,
size=guide_legend(reverse=TRUE, override.aes=list(color=NA,fill="grey50")),
fill=guide_legend(reverse=TRUE, override.aes=list(color=NA, size=3)))

Answer 2

Read data frame:
df <- data.frame(gvs = c(-2.3321916, -1.4996229, 1.795817, 2.5543214, -2.775846, -1.9659984,
2.1239768, -1.8515908, -1.5597359, -1.2534511, -0.1625003, -1.0265275,
-0.8519079, -2.4279528, -3.1717421, -0.9258497, -1.2207974, -1.0325107,
-3.2102686, -0.9377928, -1.7657017, -0.5058627, -0.7819299, -1.4095947,
2.2810477, -0.9007551, 2.6614486, -0.944198, -0.7237586, -0.9452944,
-1.2035258, -0.7703779, -3.0895998, -0.7037952, -1.9311354, -0.5423822,
-1.6244801, -0.9049735, -2.6491331, 2.3706977, -2.7590587, -0.961419,
-1.6961014, -0.8449225, -1.1163019, -0.9298674, -2.8859587, -1.4494841,
-0.7381369, -0.5644325, -0.9195156, 2.0949378),
order = c(1L, 5L, 15L, 19L, 33L, 39L, 41L, 47L, 6L, 14L, 31L, 35L, 2L,
8L, 9L, 17L, 21L, 24L, 37L, 43L, 48L, 10L, 12L, 13L, 16L, 18L,
20L, 23L, 29L, 30L, 32L, 34L, 42L, 45L, 46L, 49L, 50L, 51L, 11L,
22L, 26L, 38L, 44L, 28L, 36L, 3L, 4L, 7L, 25L, 27L, 40L, 52L),
labels = c("Adygei", "Basque", "French", "Italian", "Orcadian", "Russian",
"Sardinian", "Tuscan", "Bedouin", "Druze", "Mozabite", "Palestinian",
"Balochi", "Brahui", "Burusho", "Hazara", "Kalash", "Makrani",
"Pathan", "Sindhi", "Uygurf", "Cambodian", "Dai", "Daur", "Han",
"Hezhen", "Japanese", "Lahu", "Miao", "Mongola", "Naxi", "Oroqen",
"She", "Tu", "Tujia", "Xibo", "Yakut", "Yi", "Colombian", "Karitiana",
"Maya", "Pima", "Surui", "Melanesian", "Papuan", "BantuKenya",
"BantuSouthAfrica", "BiakaPygmy", "Mandenka", "MbutiPygmy", "San",
"Yoruba"))
Order data
df.ordered <- df[ order(df$order) , ]
And some simple (ugly) sample plotting which you can surely improve upon (maybe with ggplot)
plot(df.ordered$gvs, pch = 19)
axis(1, at=1:52, labels=df.ordered$labels, las=2)

Answer 3

Another option that doesn't rely on the sorting of the dataframe is to use the limits parameter of a discrete scale (which as a side benefit can allow you do do more arbitrary ordering when plotting.)
df <-read.csv(/path/to/file/df.csv')
xorder <-df[order(df$order),'labels']
ggplot(df, aes(x=labels, y=gvs, size=gvs)) +
geom_point() +
scale_x_discrete(limits=xorder)+
theme(axis.text.x=element_text(angle=90))