Cody Martin 6/18/2021
You can install R here: https://cran.r-project.org/mirrors.html
- Just choose the location closest to you
You can install R studio, a fantastic interactive coding environment for R, here: https://www.rstudio.com/products/rstudio/download/
Then you can open the README.Rmd file from the GitHub repository, and
use the code below.
Need more R help? Checkout my Intro to R document: https://github.com/cody-mar10/intro_R/blob/main/README.md
Scroll to the end to have a copy & paste template to plot code.
First load your packages. Normally, you would need to install any packages you don’t have loaded, but I am forcing the code to install packages for you if they are not already.
pkgs <- c("tidyverse", "ggprism", "ggrepel") # note tidyverse includes ggplot2 and dplyr
# Check if packages are installed
for (p in pkgs) {
if(! p %in% installed.packages()){
install.packages(p, dependencies = TRUE)
}
}
# Load packages
invisible(lapply(pkgs, library, character.only=T))
# Note normally you can just do:
# library(tidyverse) for exampleHere, I read in my wide-formatted lysis curve data. Although I have simple names, you should keep your names useful for coding. Here is an example:
- pRE gp60-63 = BAD!!! Don’t ever use spaces
- pRE_gp60-63 = GOOD Use _ instead of spaces
Be consistent in your sample naming. Furthermore, you first column must be the time column.
file <- "data/simpledata.csv"
data <- read_csv(file)
# change first column name to "Time"
colnames(data)[1] <- "Time"
data| Time | MG1655 | N4 | A | B | C | D | E |
|---|---|---|---|---|---|---|---|
| 0 | 0.223 | 0.239 | 0.237 | 0.231 | 0.228 | 0.235 | 0.231 |
| 30 | 0.523 | 0.488 | 0.489 | 0.505 | 0.510 | 0.506 | 0.504 |
| 35 | 0.620 | 0.558 | 0.441 | 0.505 | 0.550 | 0.503 | 0.510 |
| 40 | 0.685 | 0.612 | 0.212 | 0.354 | 0.517 | 0.332 | 0.379 |
| 45 | 0.772 | 0.683 | 0.110 | 0.222 | 0.428 | 0.176 | 0.244 |
| 50 | 0.844 | 0.718 | 0.089 | 0.137 | 0.371 | 0.118 | 0.174 |
| 55 | 0.888 | 0.789 | 0.067 | 0.111 | 0.354 | 0.098 | 0.136 |
| 60 | 0.968 | 0.815 | 0.068 | 0.105 | 0.346 | 0.095 | 0.126 |
| 70 | 1.390 | 0.859 | 0.051 | 0.091 | 0.331 | 0.108 | 0.111 |
| 80 | 1.544 | 0.962 | 0.053 | 0.068 | 0.151 | 0.086 | 0.074 |
| 93 | 2.010 | 1.224 | 0.048 | 0.070 | 0.125 | 0.064 | 0.084 |
To proceed, we need to take your wide-formatted data and convert it to long-formatted data. This way each row will be a singular observation. This is how computational software likes to work with data since the software can use efficient vectorized operations.
If you only have one variable to test in your lysis curve like different genetic backgrounds, you will only need separately colored lines to plot for the visual difference. This will be a “simple” plot, and we can re-format your data easily.
However, if you have multiple varibles in your lysis curve like genetic background and +/- some chemical like DNP, nalidixic acid, chloramphenicol, etc. then READ CAREFULLY!!!
You should name your sample columns with ODs with all variable levels separated by _ UNDERSCORES ONLY!!!
Example: suppose I have genotype A and B, and +/- DNP, I would name my columns like this:
A_- B_- A_+ B_+
DO NOT, and I repeat DO NOT!!! use _ anywhere else. This is because the long formatting code will separate the column name into separate columns at the _ delimiter, so the above columns would separate into columns like this:
| Genotype | DNP |
|---|---|
| A | - |
| B | - |
| A | + |
| B | + |
I’ve included an example of both simple and multivariable data sets:
data/simpledata.csvdata/complexdata.csv
The gather function from dplyr will convert data from wide to long,
turning all numeric data into a single stacked column.
wideToLong <- function(data, variables=c()) {
# check if any columns have the _ delimiter specifying multiple varibles
var_check <- grepl("_", colnames(data))
# if any columns have the _ delimiter
if (TRUE %in% var_check && length(variables) != 0) {
data_long <- data %>%
gather(key="Sample", value = "OD", -Time) %>%
separate(Sample, sep="_", remove=F, into=variables)
} else {
data_long <- data %>%
gather(key="Sample", value = "OD", -Time)
}
return(data_long)
}
data_long <- wideToLong(data=data)
data_long %>% head(10)| Time | Sample | OD |
|---|---|---|
| 0 | MG1655 | 0.223 |
| 30 | MG1655 | 0.523 |
| 35 | MG1655 | 0.620 |
| 40 | MG1655 | 0.685 |
| 45 | MG1655 | 0.772 |
| 50 | MG1655 | 0.844 |
| 55 | MG1655 | 0.888 |
| 60 | MG1655 | 0.968 |
| 70 | MG1655 | 1.390 |
| 80 | MG1655 | 1.544 |
I’ve made extensive use of the ggplot2, ggprism, and ggrepel
packages to make the best plots you will ever set your gaze on. This
code is mostly fullproof. If you only have simple plots, it should be
fine.
Here is how I plot my graphs using my customizations:
# define custom offset to move line labels away from axis
offset <- max(data_long$Time)*0.035
# ggprism has default colors to use, but I want to reorder them
cols = ggprism_data$colour_palettes$colors[c(6,1:5,7:20)]
# make ggplot object
simpleplot <- function(data_long) {
# set minor ticks based on max OD
# y-axis needs to look different if OD values rise above 1 due to log scale
if (max(data_long$OD, na.rm = T) > 1) {
max_yax = 10
y_minor = c(rep(1:9, 3)*(10^rep(-2:0, each=9)), 10) # minor ticks
} else {
max_yax = 1
y_minor = c(rep(1:9, 2)*(10^rep(c(-2, -1), each=9)), 1) # minor ticks
}
# makeplot
g <- data_long %>%
ggplot(aes(x=Time, y=OD)) +
geom_line(aes(color=Sample), size=1.25) +
geom_point(aes(shape=Sample), color="black", size=3.5) +
geom_text_repel(data=subset(data_long, Time == max(data_long$Time)), # labels next to lines
aes(label=Sample,
color=Sample,
x=Inf, # put label off plot
y=OD), # put label at same height as last data point
direction="y",
xlim=c(max(data_long$Time)+offset, Inf), # offset labels
min.segment.length=Inf, # won't draw lines
hjust=0, # left justify
size=5,
fontface="bold") +
scale_shape_prism(palette = "complete") + # change prism bullet shape palette
scale_color_manual(values=cols) +
scale_y_log10(limit=c(0.01,max_yax), # put y on log10 scale
minor_breaks=y_minor,
guide=guide_prism_minor(),
expand=c(0,0)) +
scale_x_continuous(minor_breaks=seq(0,max(data_long$Time),by=10),
guide=guide_prism_minor(),
expand=c(0,0)) +
labs(x="Time (min)",
y="A550") +
theme_prism(border=T) + # theme like prism plot
coord_cartesian(clip="off") +
theme(aspect.ratio=1/1,
legend.position = "none",
plot.margin=unit(c(1,5,1,1), "lines"))
return(g)
}
simpleplot(data_long)
Let’s suppose you did a lysis curve testing more than just one variable
like in data/complexdata.csv. Let’s first take a look at the data:
file <- "data/complexdata.csv"
data <- read_csv(file)
colnames(data)[1] <- "Time"
data| Time | pRE_-N4_None | gp60-63_-N4_None | pRE_+N4_t=20 | gp60-63_+N4_t=20 | pRE_+N4_t=25 | gp60-63_+N4_t=25 | pRE_+N4_t=30 | gp60-63_+N4_t=30 |
|---|---|---|---|---|---|---|---|---|
| 0 | 0.200 | 0.201 | 0.219 | 0.208 | 0.217 | 0.204 | 0.219 | 0.204 |
| 15 | 0.337 | 0.307 | 0.331 | 0.295 | 0.335 | 0.314 | 0.335 | 0.312 |
| 20 | 0.366 | 0.326 | 0.363 | 0.377 | 0.365 | 0.357 | 0.374 | 0.363 |
| 25 | 0.391 | 0.374 | 0.420 | 0.378 | 0.415 | 0.406 | 0.403 | 0.396 |
| 30 | 0.440 | 0.421 | 0.467 | 0.414 | 0.465 | 0.393 | 0.449 | 0.444 |
| 35 | 0.501 | 0.395 | 0.538 | 0.374 | 0.519 | 0.338 | 0.511 | 0.268 |
| 40 | 0.552 | 0.334 | 0.572 | 0.247 | 0.588 | 0.191 | 0.571 | 0.193 |
| 45 | 0.607 | 0.105 | 0.616 | 0.159 | 0.624 | 0.094 | 0.614 | 0.063 |
| 50 | 0.662 | 0.056 | 0.692 | 0.125 | 0.683 | 0.072 | 0.690 | 0.056 |
| 55 | 0.725 | 0.050 | 0.730 | 0.128 | 0.766 | 0.070 | 0.779 | 0.049 |
| 60 | 0.816 | 0.061 | 0.813 | 0.126 | 0.860 | 0.075 | 0.792 | 0.059 |
If you take a look at the function that converts the data from wide to
long, you will notice you can input variable names. This tells the code
to split your sample columns into n columns for each variable you have
using the separate function from dplyr.
wideToLong <- function(data, variables=c()) {
# check if any columns have the _ delimiter specifying multiple varibles
var_check <- grepl("_", colnames(data))
# if any columns have the _ delimiter
if (TRUE %in% var_check && length(variables) != 0) {
data_long <- data %>%
gather(key="Sample", value = "OD", -Time) %>%
separate(Sample, sep="_", remove=F, into=variables)
} else {
data_long <- data %>%
gather(key="Sample", value = "OD", -Time)
}
return(data_long)
}In this dataset, the variables are genotype, N4 addition, and Time of
addition, so I code the variable names as Genotype, N4_add, and
Time_add.
# define your variables
var <- c("Genotype", "N4_add", "Time_add")
data_long <- wideToLong(data, var)
data_long %>% head(10)| Time | Sample | Genotype | N4_add | Time_add | OD |
|---|---|---|---|---|---|
| 0 | pRE_-N4_None | pRE | -N4 | None | 0.200 |
| 15 | pRE_-N4_None | pRE | -N4 | None | 0.337 |
| 20 | pRE_-N4_None | pRE | -N4 | None | 0.366 |
| 25 | pRE_-N4_None | pRE | -N4 | None | 0.391 |
| 30 | pRE_-N4_None | pRE | -N4 | None | 0.440 |
| 35 | pRE_-N4_None | pRE | -N4 | None | 0.501 |
| 40 | pRE_-N4_None | pRE | -N4 | None | 0.552 |
| 45 | pRE_-N4_None | pRE | -N4 | None | 0.607 |
| 50 | pRE_-N4_None | pRE | -N4 | None | 0.662 |
| 55 | pRE_-N4_None | pRE | -N4 | None | 0.725 |
See how there are extra columns in our long-formatted data frame based on the variables we input! We can use these extra columns to better distinguish all our curves in the lysis curve.
Currently, you are limited to only 3 different variables. Since there are only 9 possible spots in the shaker bath, I don’t think it is possible to have more than 3 different total conditions. You can modify the code below if you somehow have 4 or more different experimental conditions tested.
# ggprism has default colors to use, but I want to reorder them
cols = ggprism_data$colour_palettes$colors[c(6,1:5,7:20)]
complexplot <- function(data_long, variables) {
if (max(data_long$OD, na.rm = T) > 1) {
max_yax = 10
y_minor = c(rep(1:9, 3)*(10^rep(-2:0, each=9)), 10) # minor ticks
} else {
max_yax = 1
y_minor = c(rep(1:9, 2)*(10^rep(c(-2, -1), each=9)), 1) # minor ticks
}
variables = rep(variables, 2)
g <- data_long %>%
ggplot(aes_string(x="Time", y="OD",
color=variables[1], linetype=variables[2],
shape=variables[3]
)) +
geom_line(size=1.25) +
geom_point(color="black", fill="black", size=3.5) +
scale_shape_prism(palette = "complete") + # change prism bullet shape palette
scale_color_manual(values=cols) +
scale_y_log10(limit=c(0.01,max_yax), # put y on log10 scale
minor_breaks=y_minor,
guide=guide_prism_minor(),
expand=c(0,0)) +
scale_x_continuous(breaks=seq(0,max(data_long$Time),by=10),
guide=guide_prism_minor(),
expand=c(0,0)) +
labs(x="Time (min)",
y="A550",
color=variables[1],
linetype=variables[2],
shape=variables[3]
) +
theme_prism(border=T) + # theme like prism plot
coord_cartesian(clip="off") +
theme(aspect.ratio=1/1,
legend.title = element_text(),
plot.margin=unit(c(1,5,1,1), "lines"))
return(g)
}
complexplot(data_long, var)
To save your plots as png images, you can just use this simple code:
save = paste0(strsplit(basename(file), ".csv")[[1]], ".png")
png(save, width=7.5, height=7.5, units="in", res=200)
## MAKE PLOT IN HERE
## IE do this
simpleplot(data_long)
dev.off()This dataset has two different experimental variables: strain and
addition of N4. Everything works exactly as described, but just to show
the generality of my complexplot code:
file <- "data/complexdata2.csv"
data <- read_csv(file)
colnames(data)[1] <- "Time"
data| Time | pRE_-N4 | gp60-63_-N4 | gp63-T65I_-N4 | gp63-T71A_-N4 | pRE_+N4 | gp60-63_+N4 | gp63-T65I_+N4 | gp63-T71A_+N4 |
|---|---|---|---|---|---|---|---|---|
| 0 | 0.208 | 0.176 | 0.179 | 0.167 | 0.190 | 0.167 | 0.176 | 0.168 |
| 15 | 0.300 | 0.267 | 0.260 | 0.251 | 0.299 | 0.262 | 0.268 | 0.250 |
| 20 | 0.326 | 0.302 | 0.279 | 0.265 | 0.326 | 0.296 | 0.272 | 0.291 |
| 25 | 0.380 | 0.321 | 0.224 | 0.305 | 0.383 | 0.337 | 0.178 | 0.318 |
| 30 | 0.441 | 0.382 | 0.102 | 0.263 | 0.475 | 0.380 | 0.080 | 0.268 |
| 35 | 0.485 | 0.390 | 0.038 | 0.181 | 0.505 | 0.368 | 0.040 | 0.195 |
| 40 | 0.550 | 0.273 | 0.030 | 0.066 | 0.566 | 0.308 | 0.044 | 0.089 |
| 45 | 0.571 | 0.111 | 0.032 | 0.036 | 0.574 | 0.233 | 0.032 | 0.052 |
| 50 | 0.671 | 0.078 | 0.031 | 0.033 | 0.637 | 0.212 | 0.031 | 0.039 |
| 55 | 0.702 | 0.081 | 0.027 | 0.036 | 0.711 | 0.216 | 0.035 | 0.036 |
| 60 | 0.745 | 0.076 | 0.037 | 0.037 | 0.777 | 0.223 | 0.040 | 0.037 |
var <- c("Strain", "N4_addition")
data_long <- wideToLong(data, var)
data_long %>% head(10)| Time | Sample | Strain | N4_addition | OD |
|---|---|---|---|---|
| 0 | pRE_-N4 | pRE | -N4 | 0.208 |
| 15 | pRE_-N4 | pRE | -N4 | 0.300 |
| 20 | pRE_-N4 | pRE | -N4 | 0.326 |
| 25 | pRE_-N4 | pRE | -N4 | 0.380 |
| 30 | pRE_-N4 | pRE | -N4 | 0.441 |
| 35 | pRE_-N4 | pRE | -N4 | 0.485 |
| 40 | pRE_-N4 | pRE | -N4 | 0.550 |
| 45 | pRE_-N4 | pRE | -N4 | 0.571 |
| 50 | pRE_-N4 | pRE | -N4 | 0.671 |
| 55 | pRE_-N4 | pRE | -N4 | 0.702 |
complexplot(data_long, var)
pkgs <- c("tidyverse", "ggprism", "ggrepel") # note tidyverse includes ggplot2 and dplyr
# Check if packages are installed
for (p in pkgs) {
if(! p %in% installed.packages()){
install.packages(p, dependencies = TRUE)
}
}
# Load packages
invisible(lapply(pkgs, library, character.only=T))
# Read in WIDE FORMATED data
### USER INPUT - CHANGE THIS LINE ###
file <- "data/simpledata.csv"
data <- read_csv(file)
### USER INPUT - CHANGE THIS LINE ###
### Input your variable names in quotes followed by ,
### like this c("var1", "var2")
### If you only have one variable like strain/genotype,
### you can leave this line UNCHANGED.
var <- c()
# Reformat data into long format
# Rename first column with time to be Time
# first column MUSTTTTT BE TIME
colnames(data)[1] <- "Time"
wideToLong <- function(data, variables=c()) {
# check if any columns have the _ delimiter specifying multiple varibles
var_check <- grepl("_", colnames(data))
if (TRUE %in% var_check && length(variables) != 0) { # if any columns have the _ delimiter
data_long <- data %>%
gather(key="Sample", value = "OD", -Time) %>%
separate(Sample, sep="_", remove=F, into=variables)
} else {
data_long <- data %>%
gather(key="Sample", value = "OD", -Time)
}
return(data_long)
}
data_long <- wideToLong(data=data, var)
# define custom offset to move line labels away from axis
offset <- max(data_long$Time)*0.035
# ggprism has default colors to use, but I want to reorder them
cols = ggprism_data$colour_palettes$colors[c(6,1:5,7:20)]
# make ggplot object
simpleplot <- function(data_long) {
if (max(data_long$OD, na.rm = T) > 1) {
max_yax = 10
y_minor = c(rep(1:9, 3)*(10^rep(-2:0, each=9)), 10) # minor ticks
} else {
max_yax = 1
y_minor = c(rep(1:9, 2)*(10^rep(c(-2, -1), each=9)), 1) # minor ticks
}
g <- data_long %>%
ggplot(aes(x=Time, y=OD)) +
geom_line(aes(color=Sample), size=1.25) +
geom_point(aes(shape=Sample), fill="black", size=3.5) +
geom_text_repel(data=subset(data_long, Time == max(data_long$Time)), # labels next to lines
aes(label=Sample,
color=Sample,
x=Inf, # put label off plot
y=OD), # put label at same height as last data point
direction="y",
xlim=c(max(data_long$Time)+offset, Inf), # offset labels
min.segment.length=Inf, # won't draw lines
hjust=0, # left justify
size=5,
fontface="bold") +
scale_shape_prism(palette = "complete") + # change prism bullet shape palette
scale_color_manual(values=cols) +
scale_y_log10(limit=c(0.01,max_yax), # put y on log10 scale
minor_breaks=y_minor,
guide=guide_prism_minor(),
expand=c(0,0)) +
scale_x_continuous(minor_breaks=seq(0,max(data_long$Time),by=10),
guide=guide_prism_minor(),
expand=c(0,0)) +
labs(x="Time (min)",
y="A550") +
theme_prism(border=T) + # theme like prism plot
coord_cartesian(clip="off") +
theme(aspect.ratio=1/1,
legend.position = "none",
plot.margin=unit(c(1,5,1,1), "lines"))
return(g)
}
complexplot <- function(data_long, variables) {
if (max(data_long$OD, na.rm = T) > 1) {
max_yax = 10
y_minor = c(rep(1:9, 3)*(10^rep(-2:0, each=9)), 10) # minor ticks
} else {
max_yax = 1
y_minor = c(rep(1:9, 2)*(10^rep(c(-2, -1), each=9)), 1) # minor ticks
}
variables = rep(variables, 2)
g <- data_long %>%
ggplot(aes_string(x="Time", y="OD",
color=variables[1], linetype=variables[2],
shape=variables[3]
)) +
geom_line(size=1.25) +
geom_point(color="black", fill="black", size=3.5) +
scale_shape_prism(palette = "complete") + # change prism bullet shape palette
scale_color_manual(values=cols) +
scale_y_log10(limit=c(0.01,max_yax), # put y on log10 scale
minor_breaks=y_minor,
guide=guide_prism_minor(),
expand=c(0,0)) +
scale_x_continuous(breaks=seq(0,max(data_long$Time),by=10),
guide=guide_prism_minor(),
expand=c(0,0)) +
labs(x="Time (min)",
y="A550",
color=variables[1],
linetype=variables[2],
shape=variables[3]
) +
theme_prism(border=T) + # theme like prism plot
coord_cartesian(clip="off") +
theme(aspect.ratio=1/1,
legend.title = element_text(),
plot.margin=unit(c(1,5,1,1), "lines"))
return(g)
}
# save plot as .png
save = paste0(strsplit(basename(file), ".csv")[[1]], ".png")
png(save, width=7.5, height=7.5, units="in", res=200)
if (ncol(data_long) == 3) {
simpleplot(data_long=data_long)
} else {
complexplot(data_long=data_long, variables=var)
}
dev.off()
## or if you want to save as a svg file for making full figures in inkscape
save = paste0(strsplit(basename(file), ".csv")[[1]], ".svg")
svg(save, width=7.5, height=7.5)
if (ncol(data_long) == 3) {
simpleplot(data_long=data_long)
} else {
complexplot(data_long=data_long, variables=var)
}
dev.off()