Data & Workflows

04

# Data & Workflows

## R Markdown for Pharma

### Alison Hill, Thomas Mock &#183; RStudio

#### [https://rmd4pharma.netlify.app//](https://https://rmd4pharma.netlify.app//)

---

#

# Go here and log in (free):

http://rstd.io/rmd4pharma-cloud

---
class: middle, center, inverse

1

# Parameters

---
class: left, top

### Knit with Parameters

````
---
title: "Penguins"
output: html_document
params:
  species: Adelie
---

```{r setup, include = FALSE}
library(tidyverse) 
smaller <- palmerpenguins::penguins %>% 
 filter(species == params$species,
 !is.na(body_mass_g))
```

# Penguins
`r nrow(palmerpenguins::penguins) - nrow(smaller)` penguins are classified as `r params$species`.

```{r, echo = FALSE, message = FALSE}
smaller %>%
  ggplot(aes(x = body_mass_g, y = flipper_length_mm)) +
  geom_point() +
  geom_smooth(method = "lm") +
  labs(title = params$species)
```

````
]

---

![](images/penguins.png)

---
class: middle, center, inverse

2

# Command Line

---
class: left, top

### `rmarkdown::render()`

```r
library(rmarkdown) # needs to be loaded first
render("rmarkdown-report.Rmd")
```

Now that was easy! But why would you want to use `render` rather than just clicking `knit`?

How about running a batch of RMDs?  
+ a report  
+ a safety document  
+ a slide deck  
+ or a full website

Run for each drug of interest *all at once*!

---

<div class="my-footer">https://rstd.io/rmd4pharma-cloud</div>

---
class: top, left

# Render loops

```r
# the setup
unique_species <- unique(palmerpenguins::penguins$species)
unique_species
#> [1] Adelie Gentoo Chinstrap
#> Levels: Adelie Chinstrap Gentoo
```

```r
# the loop
for(species in unique_species){
  rmarkdown::render(
    "penguin-report.Rmd",
    params = list(species = species)
  )
}
```

#### psst...the wrong way - but what's wrong?

* What's the name of the output report?

---
class: top, left

### We need two things

For the `rmarkdown::render()` function to work the way we want:  
1. The list of species -> `params` 
2. The name of each report -> `output_file`

We could do this with a loop, but let's start **purrr**ing...

---
class: middle, center

> "Of course *someone* has to write **loops**.  
> It doesn't have to be *you*.  
> `-` Jenny Bryan

---
class: middle, center

![](https://raw.githubusercontent.com/allisonhorst/stats-illustrations/master/rstats-artwork/map_frosting.png)

---
class: middle, center

![](https://raw.githubusercontent.com/allisonhorst/stats-illustrations/master/rstats-artwork/map2_cupcakes.png)

---
class: top, left

### Rendering multiple reports

```r
# the setup
unique_species <- unique(palmerpenguins::penguins$species)

# the function
penguin_render <- function(species){
 rmarkdown::render(
 "penguin-report.Rmd",
 output_file = paste0(species, "-report.html"),
 params = list(species = species)
 )
}

# create the reports
purrr::walk(unique_species, penguin_render)
```

---
class: your-turn

# Your turn

Use this format to create separate site reports for each of the 13 sites in our `mockdata`. Note that there are two main files:  
- `project/analysis/R/parameterized-rmd.R` -- **Skeleton code**  
- `project/analysis/reports/site-report.Rmd` -- **Actual report**

```r
# the setup
unique_species <- unique(palmerpenguins::penguins$species)

# the function
penguin_render <- function(species){
 rmarkdown::render(
 "penguin-report.Rmd",
 output_file = paste0(species, "-report.html"),
 params = list(species = species)
 )
}

# create the reports
purrr::walk(unique_species, penguin_render)
```

---
class: middle, center, inverse

3

# Reference external R code

---
class: top, left

### Entire .R files

`source()` will allow you to read in entire .R files, and is commonly used to *source* external R functions.

### Chunks of .R files

Adding `## ----` and a name to read chunks by name with `knitr::read_chunk()`

```r
## ---- smaller-penguins
smaller <- penguins %>%
 filter(
 species == "Adelie",
 !is.na(body_mass_g)
 )
## ---- plot-penguins
smaller %>%
 ggplot(aes(body_mass_g)) +
 geom_histogram(binwidth = 100)
```

---
class: top, left

````
---
title: "Penguins"
date: 2020-08-11
output: html_document
---
```{r setup, include = FALSE}
library(tidyverse)
library(palmerpenguins)
```
```{r, child=c(“adelie-report.Rmd”)}
```

````

]
--

````
---
output: html_document
---
```{r, echo = FALSE}
smaller <- penguins %>%
 filter(species == "Adelie",
 !is.na(body_mass_g))
```
# Penguins
`r nrow(penguins) - nrow(smaller)` penguins 
are classified as `r params$species`.
```{r, echo = FALSE}
smaller %>%
 ggplot(aes(body_mass_g)) +
 geom_histogram(binwidth = 100)
```

````
]

---

## Conditional Child Documents

Add specific chunks if conditions met!

````
---
title: "Drug Safety and Response"
output: html_document
---
```{r setup, include = FALSE}
library(tidyverse)
drug_data <- read_csv("drug-data.csv)

pass_fail <- summarize(drug_data, mean = mean(toxicity_param))
```

```{r, child=if (pass_fail >= 1.5) “safe-drug-addendum.Rmd”}
```

# Dose Response
```{r}
ggplot(drug_data, aes(x = drug_cocn, y = drug_response)) +
 geom_point() +
 geom_smooth()
```

````

---
class: left, top

### Links

+ [Parameterized Reports](https://bookdown.org/yihui/rmarkdown/parameterized-reports.html)  
+ [Parameterized Lesson](https://rmarkdown.rstudio.com/lesson-6.html)  
+ [RMarkdown Cookbook](https://bookdown.org/yihui/rmarkdown-cookbook/)  
+ [RMarkdown Definitive Guide](https://bookdown.org/yihui/rmarkdown/)

---
class: inverse, middle, center

# Thank you!