---
title: "Practical R: Packages"
author: Abhijit Dasgupta

---

```{r setup, include=F, child = here::here('slides/templates/setup.Rmd')}
```


```{r, include=FALSE}
get_description <- function(pkgs){
  tibble(Package = pkgs) %>% 
    mutate(Description = map_chr(Package, ~packageDescription(.)$Title)) %>% 
    knitr::kable() %>% 
    kableExtra::kable_styling()
}
```



---
class: middle, center, inverse

# What are packages in R?

---

## Packages

Packages are collections of functions, and sometimes data, that are usually unified for 
a common purpose

.saltinline[If _functions_ are recipes, then _packages_ are recipe books]

--

If you want to cook from a recipe, you first have to grab the recipe book from your shelf

--

.heatinline[Similarly, if you want to use a function from a package, you first have to grab or activate
the package in _your current R session_ ]

This is done using the `library` function 

For example, 

```{r, eval=FALSE}
library(tidyverse)
library(janitor)
```

---

## Packages

There is another way to access functions from packages, if you're really only going to 
use one function from it. 

The general form for this is .heatinline[`<package>::<function>`] (note the __two__ colons)

For example, if you just want to use the `clean_names` function from the **janitor** package, you can 
do so by 

```{r, eval=F}
janitor::clean_names(dataset)
```

where `dataset` is the name of the data.frame whose column names you want to clean.

---

## Important operational notes

.pull-left[
.acid[Install packages **once per computer**]

Never install packages inside a R Markdown file
]
.pull-right[
.heat[Activate a package **once per R session**]

]
--


.footnote[The **pacman** package and the `pacman::p_load` function saves you a bunch of trouble by 
installing a package only if it doesn't exist on your computer and then activating the packaage. This one function removes a lot of the operational issues in installing and loading packages in R.]


---
class: middle,center,inverse

# Where are the packages?

---

## CRAN

CRAN is the Comprehensive R Archive Network, a network of mirrored repositories containing R packages.

Today, it really doesn't matter which of the repositories you use. 

In RStudio, the default repository is **Global (CDN) - RStudio** which is a version in the cloud that typically works the fastest. 

![:scale 50%](../img/pkg1.png)

---

## CRAN

You can install packages from CRAN using the following means:

.pull-left[
`install.packages("<package name>")`

Or, if you want to be explicit, or are not using RStudio, 

`install.packages("<package name>", repos = "<repository URL>")`

]
.pull-right[
Using the RStudio _Packages_ panel

(see next slide)

You can find packages using CRAN [Task Views](https://cran.r-project.org/web/views/)

---
background-image: url(../img/pkg2.png)
background-size: contain

---

## GitHub

GitHub is where many R packages reside during development. 

To install a package directly from GitHub, you need the **remotes** package, and 
then you can use

```{r, eval=F}
remotes::install_github("<owner>/<repo>")
```

For example, if you want to install the development version of **dplyr**:

```{r, eval=F}
remotes::install_github("tidyverse/dplyr")
```

---

## Bioconductor

The [Bioconductor](https://www.bioconductor.org) is a R organization dedicated to 
bioinformatics. It has its own repository of over 1900 packages

To install Bioconductor packages, you first need to install the **BiocManager** package from CRAN (note the upper and lower case letters). Then you can install packages by 

```{r, eval=F}
BiocManager::install('<package name>')
```

For example, if you want to install the **DESeq2** package that computes differential gene expressions: 

```{r, eval=F}
BiocManager::install('DESeq2')
```

---
## Installing packages, a summary


.pull-left[
### From CRAN
```{r Rintro-10, eval=F}
install.packages("tidyverse")
```

### From Bioconductor
```{r Rintro-11, eval = F}
install.packages("BiocManager") # do once 
BioManager::install('limma')
```

### From GitHub

```{r Rintro-12, eval = F}
install.packages('remotes') # do once
remotes::install_github("rstudio/rmarkdown") 
# usual format is username/packagename
```
]
.pull-right[
> GitHub often hosts development version of packages published on CRAN or Bioconductor

> Both CRAN and Bioconductor have stringent checks to make sure packages can run properly, with no obvious program flaws. There are typically no
  guarantees about analytic or theoretical correctness, but most packages have been crowd-validated and there are several reliable developer groups
  including RStudio
]



---
class: middle,center,inverse

# Packages commonly used 
## An incomplete listing

---

## Data ingestion

```{r, echo=F}
get_description(c('readr','readxl','haven', 'DBI', 'rvest','jsonlite')) 
```

---

## Data munging

```{r, echo=F}
get_description(c('tidyr','dplyr','stringr','lubridate', 'forcats', 'purrr', 'janitor'))
```

---

## Data visualization

```{r, echo=F}
get_description(c('ggplot2','lattice', 'visdat','naniar','htmlwidgets','leaflet','highcharter','plotly'))
```

There is an entire package ecosystem around `ggplot2` that can be seen [here](https://exts.ggplot2.tidyverse.org/). These include specialized plots, 
different themes and colors, animations, etc. 

---

## Statistics

**Data description**

```{r, echo=FALSE}
pacman::p_load(char=c('tableone','table1','stargazer', 'arsenal','gtsummary', 'flextable', 'Hmisc'))
get_description(c('tableone','table1','stargazer', 'arsenal','gtsummary', 'flextable', 'Hmisc'))
```

---

## Statistics

**Analysis**
```{r, echo=F}
pacman::p_load(char = c('stats','survival','infer','rsample','broom', 'finalfit'))
get_description(c('stats','survival','infer','rsample','broom', 'finalfit'))
```


---

## Statistical modeling

```{r, echo=F}
pacman::p_load(char=c('stats','survival','recipes', 'rms', 'broom', 'rsample'))
get_description(c('stats','survival','recipes', 'rms', 'broom', 'rsample'))
```

---

## Machine Learning

```{r, echo=F}
pacman::p_load(char = c('caret','parsnip','yardstick', 'rpart','party','randomForest','baguette', 'kernlab','earth'))
get_description(c('caret','parsnip','yardstick', 'rpart','party','randomForest','baguette', 'kernlab','earth'))
```

---

## Reporting

```{r, echo=F}
pacman::p_load(char = c('rmarkdown','knitr','bookdown','distill', 'rticles','blogdown','flexdashboard', 'shiny', 'officer'))
get_description(c('rmarkdown','knitr','bookdown','distill', 'rticles','blogdown','flexdashboard', 'shiny', 'officer'))
```