Reordering bars in a group with ggplot and adjusting groups by width

I have made the following bar plot in ggplot2 with these data:

Plot: Significant bacterial families by comparison

head(data):
                            log2FoldChange                  Rank5      comparison
New.CleanUp.ReferenceOTU80        7.356791 f__Desulfovibrionaceae control vs high
29323423                         10.645582 f__Desulfovibrionaceae control vs high
New.CleanUp.ReferenceOTU121       7.128041 f__Desulfovibrionaceae control vs high
4288072                         -15.270046   f__Hyphomicrobiaceae control vs scar
1658672                           8.017656   f__Hyphomicrobiaceae control vs scar
9335464                         -13.997070       f__Neisseriaceae control vs scar

And this command:

ggplot(data = order.merge, 
       aes(x = comparison, y = log2FoldChange,
           group = reorder(row.names(order.merge, log2FoldChange)))) + 
  geom_bar(stat = "identity", aes(fill = Rank5), position=position_dodge()) + 
  ylab("Log2 Fold Change") + 
  coord_flip()

Currently, the Rank5 names are ordered alphabetically within the comparison category. However, the plot does not order the bars by Rank name. For instance, in the figure for the group control vs. NH4, the three bars of Rhodobacteraceae are not next to each other. Secondly, control vs NH4 has 5 fat bars while control vs high has many skinny bars.

1) How can I reorder the bars by rank?

2) Can I adjust the size of each comparison so that all bars are equal width?

Thanks!

Reproducible example:

dput(order.merge[20:40, ])

structure(list(log2FoldChange = c(-12.6763917924836, 18.4867204578785, 
5.87185962697414, 5.66255934712844, 9.90661771655056, 11.7009664985686, 
-15.8045780775922, -17.9176399760384, -11.7160496642962, -16.8548291527408, 
12.0062377518766, 9.27678750597363, -16.3219881932917, -7.35815162561094, 
-20.4898406326553, -21.4584754940589, 6.54572004269336, 11.8012634251803, 
10.2032049438921, -12.5771068977973, 7.01101590101165), Rank5 = structure(c(11L, 
14L, 17L, 18L, 19L, NA, 3L, 7L, 7L, 7L, NA, 2L, 4L, 4L, 4L, 7L, 
7L, 7L, 7L, 7L, 7L), .Names = c("4445466", "1010113", "321533", 
"New.CleanUp.ReferenceOTU123", "800197", "369965", "428807", 
"3834498", "904675", "548736", "369965", "2932342", "321405", 
"1085703", "3633321", "4421174", "4313781", "146037", "248590", 
"4420764", "1123147"), .Label = c("f__Cyanobacteriaceae", "f__Desulfovibrionaceae", 
"f__Hyphomicrobiaceae", "f__Moraxellaceae", "f__Neisseriaceae", 
"f__Pseudomonadaceae", "f__Rhodobacteraceae", "f__Sphingomonadaceae", 
"f__", "f__Alcanivoracaceae", "f__Streptococcaceae", "f__Staphylococcaceae", 
"f__Vibrionaceae", "f__Enterobacteriaceae", "f__Pirellulaceae", 
"f__Xanthomonadaceae", "f__[Amoebophilaceae]", "f__Endozoicimonaceae", 
"f__Hyphomonadaceae"), class = "factor"), comparison = c("control vs high", 
"control vs high", "control vs high", "control vs high", "control vs high", 
"control vs high", "control vs NH4", "control vs NH4", "control vs NH4", 
"control vs NH4", "control vs NH4", "control vs NO3", "control vs NO3", 
"control vs NO3", "control vs NO3", "control vs NO3", "control vs NO3", 
"control vs NO3", "control vs NO3", "control vs NO3", "control vs NO3"
)), .Names = c("log2FoldChange", "Rank5", "comparison"), row.names = c("44454662", 
"10101131", "321533", "New.CleanUp.ReferenceOTU123", "800197", 
"3699652", "4288073", "3834498", "9046752", "5487361", "3699653", 
"29323424", "3214051", "10857031", "36333212", "44211743", "43137811", 
"1460372", "248590", "44207641", "11231471"), class = "data.frame")

Solution

To order the bars by rank you'll need to update your call to reorder() to use Rank5 (see below).

To get bars of the same width, install the development version of ggplot2 and use position_dodge(preserve = "single").

devtools::install_github("tidyverse/ggplot2")
library("ggplot2")

ggplot(
  data = order.merge,
  aes(
    x = comparison,
    y = log2FoldChange,
    # reorder by Rank5 instead of log2FoldChange
    group = reorder(row.names(order.merge), as.numeric(Rank5))
  )) + 
  geom_bar(
    stat = "identity",
    aes(fill = Rank5),
    # preserve individual width of bars
    position = position_dodge(preserve = "single")
  ) + 
  ylab("Log2 Fold Change") + 
  coord_flip()