Plotting a Secondary Y axis when Log Transformation in Primary needed

Question

I m trying to plot a line graph with two axes (see pic). I tried many ways from examples taken from here but nothing worked. The left axis must be Log10 while the secondary (right) normal. Also, I do need to set their limits. All data come from a single dataset (dfwc) and I took two subsets (here plotted as "aa" and "bb", Y-left and Y-right, respectively). Finally, I want to have legend colours by strain (as shown in pic from the left). I hope you can help me out, thank you in advance!

  aa <- ggplot(dfwc, aes(x=time, y=numbers, color=strain, group=strain)) + theme_dp1() + 
              geom_point(aes(color=strain, group=strain), subset(dfwc,strain %in% c("mc", "mut1", "mut2", "mut3")), alpha=.5, size=2) +
              geom_errorbar(aes(ymax=numbers-sd, ymin=numbers+sd), subset(dfwc,strain %in% c("mc", "mut1", "mut2", "mut3")), size=.4, width=.1, show.legend=FALSE) +
              geom_line(aes(color=strain), subset(dfwc,strain %in% c("mc", "mut1", "mut2", "mut3")), size=.6, linetype="dashed") +
            scale_y_log10(breaks=c(0.01, 0.1, 1, 10),
                    labels=scales::trans_format("log10", scales::math_format(10^.x))) +
            annotation_logticks(colour="#8d96a3", sides='l', size = .5, short=unit(1,"mm"), mid=unit(2,"mm"), long=unit(2.5,"mm")) + 
            expand_limits(y=c(0.05,8)) +  #set Y1 limits
            scale_color_manual(name = "Strain", values=colorsg, labels=c("WT", "RsmB", "RsmB:IGR", "RsmB,vfmE"))+
           labs(x=bold("Time")~(hours), y=bold("Bacterial growth")~(OD[600]))  # axes labels
    aa
    #second subset
  bb <- ggplot(dfwc, aes(x=time, y=numbers, color=strain, group=strain)) + theme_dp1() + 
            geom_point(aes(), subset(dfwc,strain %in% c("vmc", "vmut1", "vmut2", "vmut3")), alpha=.4, size=1.5, show.legend=FALSE) +
            geom_errorbar(aes(ymax=numbers-sd, ymin=numbers+sd), subset(dfwc,strain %in% c("vmc", "vmut1", "vmut2", "vmut3")), size=.4, width=.05, show.legend=FALSE) +
            geom_line(aes(), subset(dfwc,strain %in% c("vmc", "vmut1", "vmut2", "vmut3")), size=.6, linetype=1, show.legend=FALSE) +
            scale_y_continuous(sec.axis = sec_axis(~.*1, name = "This is my Right Axis"))
    bb}

I want to have these 2 images in 1 graph

Data transformation (raw data here: https://docs.google.com/spreadsheets/d/1zuNzjTjf_0MRyRoTCEfiP1WqRl3viWHueNXepiA7m9o/edit#gid=0)

data_long <- gather(curve, time, numbers, X0:X12, factor_key=TRUE) #change according to hours numbers
data_long
plot(data_long)
names(data_long)
str(data_long)  #To view variable structure

# Rename factor names (time) from "h0" to "h12" to "0" to "14"  #useful! 
levels(data_long$time)[levels(data_long$time)=="X0"] <- "0"
levels(data_long$time)[levels(data_long$time)=="X2"] <- "2"
levels(data_long$time)[levels(data_long$time)=="X4"] <- "4"
levels(data_long$time)[levels(data_long$time)=="X6"] <- "6"
levels(data_long$time)[levels(data_long$time)=="X8"] <- "8"
levels(data_long$time)[levels(data_long$time)=="X10"] <- "10"
levels(data_long$time)[levels(data_long$time)=="X12"] <- "12"
levels(data_long$time)[levels(data_long$time)=="X14"] <- "14"

#Statistics ! 
dfwc<-summarySE(data_long, measurevar="numbers", groupvars=c("strain", "time"))
dfwc

Hi! Can you provide some data? See "minimal reproducible example". — Magnus Nordmo, May 07 '20 at 08:58
Hi! thanks for your help! I just edited the post (it now includes all data and transformations I did; ready to be copied) :) — dplazas, May 07 '20 at 11:50

score 0 · Answer 1 · answered May 07 '20 at 12:57

I cannot help you with adding a secondary log axis on the left, while having a normal axis on the right. Usually you would to it the other way around. Also the log axis doesnt work well in this case with alot of values below 1. See this example:

library(tidyverse)

del <- read.csv("~/Path/del.csv", stringsAsFactors=FALSE)

del <- del %>%  
  pivot_longer(3:9,names_to = "time") %>% 
  mutate(time =  stringr::str_replace_all(time, pattern = 'X',replacement = 'T'),
         time = fct_relevel(time,'T0','T2','T4','T6','T8','T10','T12'))

del <- del %>% 
  mutate(subject = as.factor(subject),
         strain = as.factor(strain),
         time = as.factor(time))

del %>% 
  filter(!is.na(value)) %>%  
  ggplot(aes(x=time, y=value, color=strain, group=strain)) + 
  geom_point(alpha=.5, size=2) +
  geom_line() +
  scale_y_continuous(sec.axis = sec_axis(~log10(.)))

Note: It's great that you provide data! Please be advised the minimal part. The most effective questions aims at a particular problem with every unnecessary detail omitted. Also, its best of the reproducible example is self-contained. Here, you use data that you do not supply in the ggplot code for example. See reprex

Hey! Ok I see. I have also obtained similar graphs... some time ago, I made what I want by using Plotly [it was simpler because I used the "yaxis2" function...], but after multiple headaches with saving in svg format I gave up. That's the main issue in ggplot, in most of the examples the secondary axis is the one modified, not the left-side axis. Thank you anyway for your help! — dplazas, May 07 '20 at 14:38

Plotting a Secondary Y axis when Log Transformation in Primary needed

1 Answers1

Linked