README.Rmd

---
output: 
    github_document:
        html_preview: false
---

<!-- README.md is generated from README.Rmd. Please edit that file -->

# plot-recreation

<!-- badges: start -->
<!-- badges: end -->
Re-creating plots to learn cool tricks and practice problem-solving.

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE,
                      error = TRUE,
                      fig.path = 'figures/')
```
```{r libs, message=FALSE}
library(cowplot)
library(glue)
library(ggtext)
library(gridtext)
library(here)
library(magick)
library(RColorBrewer)
library(tidyverse)
set.seed(2020)
```

## Exercises

View the plots below and try to write the code that creates them!
There are more exercises here than you can probably get to in the time allotted.
Feel free to jump around and pick ones to work on based on how interested you are in learning how to create the plot.
At the end, we'll each share one solution.

As much as this is an exercise in data viz with R, it's also an exercise in finding solutions in package documentation and general problem-solving.
Google is your friend!
If you get stuck, take a look at the [hints](hints.md) or send Kelly a message in Slack.

- [Exercises](#exercises)
  - [Italicize text](#italicize-text)
  - [Color axis text](#color-axis-text)
  - [Color & bold axis label](#color-bold-axis-label)
  - [Assign colors manually](#assign-colors-manually)
  - [Image on a plot](#image-on-a-plot)
  - [Plot on an image](#plot-on-an-image)
  - [Plots in a grid](#plots-in-a-grid)
  - [One title for two plots](#one-title-for-two-plots)
  - [Inset plots](#inset-plots)
  - [Overlaying plots](#overlaying-plots)
  - [Multi-panel figure](#multi-panel-figure)
- [Solutions](#solutions)
      
### Italicize text

![](figures/KEY_otu_italics-1.png)

Source: [Claus Wilke's talk at rstudio::conf(2020)](https://twitter.com/ClausWilke/status/1222944728443809792)

```{r otu_italics}
otu_data <- tribble(~otu, ~bact, ~value,
                       1, 'Staphylococceae', -0.5,
                       2, 'Moraxella', 0.5,
                       3, 'Streptococcus', 2,
                       4, 'Acinetobacter', 3.0
                    ) %>% mutate(name = glue("*{bact}* (OTU {otu})"))
otu_data %>% 
    ggplot(aes(XXXXXX, XXXXXX)) +
    geom_XXXXXX() + 
    XXXXXX() +
    theme(axis.text.y = XXXXXX)
```

### Color axis text

![](figures/KEY_otu_color-1.png)

Source: [Claus Wilke's talk at rstudio::conf(2020)](https://twitter.com/ClausWilke/status/1222944728443809792)

```{r otu_color}
# make some fake data
otu_data <- tribble(~otu, ~bact, ~value,
                       1, 'Staphylococceae', -0.5,
                       2, 'Moraxella', 0.5,
                       3, 'Streptococcus', 2,
                       4, 'Acinetobacter', 3.0
                    ) %>% 
            mutate(color = brewer.pal(4, "Dark2"),
                   name = glue("<i style='color:{color}'>{bact}</i> (OTU {otu})")
                   )
# plot it
otu_data %>% 
    ggplot(aes(name, value, fill=color)) +
    geom_XXXXXX() + 
    scale_XXXXXX_identity() +
    XXXXXX_flip() +
    XXXXXX(axis.text.y = XXXXXX)
```

### Color & bold axis label

![](figures/KEY_candidemia_color-1.png)

Inspiration: fig. 2C from Jay's recent journal club paper: [Zhai et al 2018 Nature Med](https://www.nature.com/articles/s41591-019-0709-7/figures/2)

```{r candidemia_color}
# make some fake data
time_data <-tibble(time = seq(0,5,0.1),
               yes = sin(time * 0.8) + 0.8,
               no = sin(time * -0.5) + 1
                ) %>% pivot_longer(-time, names_to = "infected", values_to = "value")
head(time_data)

# plot it
time_data %>% ggplot(aes(x = XXXXXX, y = XXXXXX, color = infected)) +
    geom_XXXXXX() +
    ylab("<b style='color:#00BFC4'>Candidemia</b> vs. <b style='color:#F8766D'>no Candidemia</b> ITS1 diversity") +
    theme_XXXXXX() +
    theme(legend.position = XXXXXX,
          axis.title.y = XXXXXX)
```

### Assign colors manually

![](figures/KEY_color_manual-1.png)

Adapted from: https://github.com/SchlossLab/compositional_data_analysis

```{r color_manual}
palette = brewer.pal(n = 4, name = "Paired")
colors = c(A = palette[[2]],
           B = palette[[3]])
otu_table <- tibble(
    sample = c("1", "2", "3", "1", "2", "3"),
    otu = c("A", "A", "A", "B", "B", "B"),
    abun = c(5, 15, 6, 15, 5, 2)
) %>% mutate(color = colors[otu])

# plot total counts (aboslute abundace)
otu_table %>%
    ggplot(aes(x = sample, y = XXXXXX, fill = XXXXXX)) +
    geom_XXXXXX() +
    scale_XXXXXX_XXXXXX() +
    XXXXXX("Total bacteria") +
    theme_classic()
```

### Image on a plot

![](figures/KEY_img_on_plot-1.png)

Adapted from: https://github.com/SchlossLab/compositional_data_analysis

```{r img_on_plot}
otu_table <- tibble(
    sample = c("1", "2", "3", "1", "2", "3"),
    otu = c("A", "A", "A", "B", "B", "B"),
    abun = c(5, 15, 6, 15, 5, 2)
)
# from https://github.com/mothur/logo
logo_file <- here("figures", "mothur_RGB.png")

# plot total counts (aboslute abundace)
plot_total <-  otu_table %>%
    ggplot(aes(x = XXXXXX, y = XXXXXX, fill = XXXXXX)) +
    geom_XXXXXX() +
    XXXXXX("Total bacteria") +
    theme_classic()


ggXXXX() + 
  draw_XXXXXX(plot_total) +
  draw_XXXXXX(logo_file, x = 1, y = 1, 
             hjust = 1.2, vjust = 1.2, 
             width = 0.3, height = 0.3)
```

### Plot on an image

_With great power, comes great responsibility. Use at your own risk!_ 

![](figures/KEY_plot_on_img-1.png)

Adapted from: https://wilkelab.org/cowplot/articles/drawing_with_on_plots.html#combining-plots-and-images

```{r plot_on_img}
rides <- read_csv('https://raw.githubusercontent.com/kelly-sovacool/strava/master/data/processed/activities.csv', col_types = cols()) %>% 
    filter(distance_mi > 1, as.character(type) == 'Ride', year >= 2019) %>% 
    select(distance_mi, start_date)

# source https://www.pinclipart.com/picdir/big/196-1969309_free-bicycle-bike-cartoon-no-background-clipart.png
bike_img <- here('figures', 'bike.png') %>%
    image_read() %>% 
    image_colorize(70, "white")

# make the plot
rides_plot <- rides %>%
  ggplot(aes(XXXXXX, XXXXXX)) +
  geom_point(color='blue', alpha=0.8, size=2) +
    theme_XXXXXX()

XXXXXX() + 
  draw_XXXXXX(bike_img) + 
  draw_XXXXXX(rides_plot)
```

### Plots in a grid

![](figures/KEY_plot_grid-1.png)

Adapted from: https://github.com/SchlossLab/OptiFitAnalysis/tree/master/exploratory/2020#performance-as-measured-by-runtime

```{r plot_grid}
benchmarks_fit <-read_tsv('https://raw.githubusercontent.com/SchlossLab/OptiFitAnalysis/master/subworkflows/2_fit_reference_db/results/benchmarks.tsv?token=AEHR6TPPUNNM245DZ7TUP2C7ADK5E', col_types = cols())
benchmarks_clust <- read_tsv('https://raw.githubusercontent.com/SchlossLab/OptiFitAnalysis/master/subworkflows/1_prep_samples/results/benchmarks.tsv?token=AEHR6TNCIGRVN3B5GEVRVNC7ADLAS', col_types = cols())

plot_box_time <- function(df) {
  df %>%
    group_by(dataset, method) %>%
    ggplot(aes(x = method, y = s, color = dataset)) +
    geom_XXXXXX() +
    scale_XXXXXX_XXXXXX() +
    facet_XXXXXX("ref") +
    ylim(0,2100) +
    labs(y = 'seconds') +
    theme_XXXXXX() +
    XXXXXX(axis.title.x = element_blank())
}

time_plot_fit <- benchmarks_fit %>% 
  plot_box_time() +
  XXXXXX(axis.text.y = XXXXXX(),
        axis.title.y = XXXXXX())

time_plot_clust <- benchmarks_clust %>% 
  plot_box_time() +
  theme(legend.position = XXXXXX)

plot_XXXXXX(time_plot_clust, time_plot_fit, XXXXXX = "h", rel_widths = c(1,3))
```

### One title for two plots

![](figures/KEY_one_title_grid-1.png)

Adapted from: https://wilkelab.org/cowplot/articles/plot_grid.html#joint-plot-titles

```{r one_title_grid}
mph_per_kph <- 0.621371
rides <- read_csv('https://raw.githubusercontent.com/kelly-sovacool/strava/master/data/processed/activities.csv', col_types = cols()) %>% 
    filter(distance_mi > 1, as.character(type) == 'Ride') %>% 
    mutate(average_speed_mph = average_speed * mph_per_kph) %>% 
    select(distance_mi, average_speed_mph, moving_time_hrs, start_date)

# make a plot grid consisting of two panels
p1 <- ggplot(rides, aes(x = XXXXXX, y = distance_mi)) + 
  geom_XXXXXX(colour = "blue") + 
  theme_XXXXXX(12) + 
  XXXXXX_grid(minor = 'none')

p2 <- ggplot(rides, aes(x = moving_time_hrs, y = XXXXXX)) + 
  geom_XXXXXX(colour = "green") + 
  theme_XXXXXX(12) + 
  background_XXXXXX(minor = 'none') +
    theme(axis.title.y = element_XXXXXX())

plot_row <- plot_XXXXXX(p1, p2)

# now add the title
title <- ggXXXX() + 
  draw_XXXXXX(
    "Distance travelled with average speed and time spent moving",
    fontface = 'bold',
    x = 0,
    hjust = 0
  ) +
  XXXXXX(
    # add margin on the left of the drawing canvas,
    # so title is aligned with left edge of first plot
    plot.margin = margin(0, 0, 0, 7)
  )
plot_XXXXXX(XXXXXX, plot_row, ncol = XXXXXX,
  # rel_heights values control vertical title margins
  rel_heights = c(0.1, 1)
)
```

### Inset plots

![](figures/KEY_inset-1.png)

Adapted from: https://wilkelab.org/cowplot/articles/drawing_with_on_plots.html#making-inset-plots

```{r inset}
rides <- read_csv('https://raw.githubusercontent.com/kelly-sovacool/strava/master/data/processed/activities.csv', col_types = cols()) %>% 
    filter(distance_mi > 1, as.character(type) == 'Ride') %>% 
    select(distance_mi, start_date, year)

plot_point <- rides %>% ggplot(aes(start_date, XXXXXX)) +
  XXXXXX(color='red') +
  theme_minimal_grid(12)

inset <- rides %>%  
    group_by(year) %>% 
    summarize(sum_distance_mi=sum(distance_mi)) %>% 
    ggplot(aes(XXXXXX, sum_distance_mi)) + 
  geom_bar(stat = "Identity", XXXXXX = "skyblue2", alpha = 0.7) + 
    labs(y='Total Distance (mi)') +
  XXXXXX_XXXXXX_continuous(expand = expansion(mult = c(0, 0.05))) +
  theme_XXXXXX_hgrid(11)

ggdraw(plot_point + theme_XXXXXX_open(12)) +
  draw_XXXXXX(inset, x = 0.1, y = .45, width = .5, height = .5) +
  draw_XXXXXX_XXXXXX(
    c("A", "B"),
    c(0, 0.1),
    c(1, 0.95),
    size = 12
  )
```

### Overlaying plots

![](figures/KEY_overlay-1.png)

Adapted from: https://wilkelab.org/cowplot/articles/aligning_plots.html

```{r overlay}
# make up some non-sensical data
alpha_data <- tibble(sample = as.factor(1:10),
                     otu_count = runif(10, min = 0, max = 100),
                     alpha_div = rnorm(10, mean = 2, sd = 0.5)
                     )

plot_col <- alpha_data %>%
    ggplot(aes(sample, otu_count)) +
    XXXXXX_col(fill = "#6297E770") +
    XXXXXX_XXXXXX_continuous(expand = expansion(mult = c(0, 0.05)),
                             position = "right") +
    scale_x_discrete() +
    XXXXXX_minimal_hgrid(11, rel_small = 1) +
    XXXXXX(
        panel.grid.major = element_XXXXXX(color = "#6297E770"),
        axis.line.x = element_XXXXXX(),
        axis.text.x = element_XXXXXX(),
        axis.title.x = element_XXXXXX(),
        axis.ticks = element_XXXXXX(),
        axis.ticks.length = grid::unit(0, "pt"),
        axis.text.y = element_XXXXXX(color = "#6297E7"),
        axis.title.y = element_XXXXXX(color = "#6297E7")
    )

plot_point <- alpha_data %>% 
    ggplot(aes(sample, alpha_div)) + 
  XXXXXX_XXXXXX(size = 3, color = "#D5442D") + 
  XXXXXX_XXXXXX_continuous(limits = c(0, 4.5)) +
  theme_half_open(11, rel_small = 1) +
    scale_XXXXXX_discrete() +
  theme(
    axis.ticks.y = element_XXXXXX(color = "#BB2D05"),
    axis.text.y = element_XXXXXX(color = "#BB2D05"),
    axis.title.y = element_XXXXXX(color = "#BB2D05"),
    axis.line.y = element_XXXXXX(color = "#BB2D05")
  )

aligned_plots <- XXXXXX_plots(plot_col, plot_point, align="hv", axis="tblr")
overlaid_plots <- XXXXXX(aligned_plots[[1]]) + 
    XXXXXX_plot(aligned_plots[[2]])
overlaid_plots
```

### Multi-panel figure

![](figures/KEY_multi-panel-1.png)

Source: https://wilkelab.org/cowplot/articles/shared_legends.html

```{r multi-panel}
# plot 1
p1 <- ggplot(iris, aes(Sepal.Length, Sepal.Width, color = Species)) + 
  geom_XXXXXX() + 
  stat_smooth(method = "lm") +
  facet_XXXXXX(. ~ Species) +
  theme_half_open(12) +
  background_grid(major = 'y', minor = "none") + 
  panel_border() + 
  XXXXXX(legend.position = "none")

# plot 2
p2 <- ggplot(iris, aes(Sepal.Length, fill = Species)) +
  XXXXXX_density(alpha = .7) + 
  scale_XXXXXX_XXXXXX(expand = expansion(mult = c(0, 0.05))) +
  theme_XXXXXX_open(12) +
  XXXXXX(legend.justification = "top")
p2a <- p2 + XXXXXX(legend.position = "none")

# plot 3
p3 <- ggplot(iris, aes(Sepal.Width, fill = Species)) +
  geom_density(alpha = .7) + 
  XXXXXX_XXXXXX_continuous(expand = c(0, 0)) +
  theme_half_XXXXXX(12) +
  XXXXXX(legend.position = "none")

legend <- XXXXXX_XXXXXX(p2)

# align all plots vertically
plots <- XXXXXX_XXXXXX(p1, p2a, p3, align = 'v', axis = 'l')

# put together the bottom row and then everything
bottom_row <- plot_XXXXXX(
  plots[[2]], plots[[3]], legend,
  labels = c("B", "C"),
  rel_widths = c(1, 1, .3),
  nrow = 1
)
plot_XXXXXX(plots[[1]], bottom_row, labels = c("A"), ncol = 1)
```

## Solutions

A link to the source or inspiration is given under each code chunk. 
See the [key](https://github.com/SchlossLab/plot-recreation/tree/key) branch for the exact solutions.

Many of these are taken directly from another source, in some cases with slight modifications. The original creators' copyrights still apply; all are OSI-approved licenses.