https://github.com/teunbrand/ggnomics

A small project to add ggplot2 extensions
https://github.com/teunbrand/ggnomics

bioconductor genomics ggplot-extension visualization

Last synced: about 2 months ago
JSON representation

A small project to add ggplot2 extensions

• Host: GitHub
• URL: https://github.com/teunbrand/ggnomics
• Owner: teunbrand
• License: other
• Created: 2019-04-15T14:37:23.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2021-08-31T14:30:56.000Z (over 3 years ago)
• Last Synced: 2024-10-14T17:28:14.079Z (2 months ago)
• Topics: bioconductor, genomics, ggplot-extension, visualization
• Language: R
• Homepage: https://teunbrand.github.io/ggnomics/
• Size: 1.22 MB
• Stars: 80
• Watchers: 6
• Forks: 4
• Open Issues: 6
• Metadata Files:
  • Readme: README.Rmd
  • License: LICENSE

Awesome Lists containing this project

• awesome-r-dataviz - ggnomics - A small project to add ggplot2 extensions. (ggplot / Miscellaneous)

README

        
---

output: github_document

---

```{r, include = FALSE}

ragg_png = function(..., res = 192) {

  ragg::agg_supertransparent(..., res = res, units = "in")

}

knitr::opts_chunk$set(

  collapse = TRUE,

  comment = "#>",

  fig.path = "man/figures/README-",

  out.width = "80%",

  dev = "ragg_png",

  fig.ext = "png"

)

```

# ggnomics 

[![Lifecycle: experimental](https://img.shields.io/badge/lifecycle-experimental-orange.svg)](https://www.tidyverse.org/lifecycle/#experimental)

[![Travis build status](https://travis-ci.org/teunbrand/ggnomics.svg)](https://travis-ci.org/teunbrand/ggnomics)

[![Codecov test coverage](https://codecov.io/gh/teunbrand/ggnomics/branch/BioC/graph/badge.svg)](https://codecov.io/gh/teunbrand/ggnomics)

This packages offers tools to integrate Bioconductor S4 classes and the ggplot2 plotting system. 

## Installation

You can install this experimental branch from [GitHub](https://github.com/) with:

``` r

# install.packages("devtools")

devtools::install_github("teunbrand/ggnomics")

```

## Example

The main idea is that you would be able to plot S4 Vector classes as you would any other vector, without having to convert to base R or tidyverse data-structures first. Below is an example of how the `GRanges` class inside a `DataFrame` is handled. 

```{r, genomic_scale}

suppressPackageStartupMessages({

  library(ggnomics)

  library(GenomicRanges)

})

df <- DataFrame(

  x = GRanges(c("chr1:100-200:+", "chr1:150-300:-", "chr2:200-300:*")),

  y = 1:3

)

g <- ggplot(df, aes(x, y, fill = strand(x))) +

  geom_tile(width = 0, height = 0.8)

g

```

## Explanation

To showcase what happens in the example above notice a few things. Most noticeable is that the `GRanges` class is used as a valid position variable. You can also notice the new axis guide, that appropriately indicates the `seqnames` slot of the input.

Because `geom_tile()` re-parametrises the data as a rectangle with `xmin` and `xmax` values, the `GRanges` object respectively uses it's start- and end-positions depending on the context of the aesthetic's name. This opens up a wide range of geoms to be compatible with ranged classes.

The `aes(..., fill = strand(x))` statement uses ggplot2's non-standard evaluation as you would expect with `base::data.frame`s. Furthermore, `strand(x)` evaluates to a `factor-Rle`, which is now a valid, mappable, discrete class. 

The data retains the S4 classes throughout the vast majority of plot building, allowing specialised geoms and stats to take advantage of this.

```{r, layer_data}

(ld <- layer_data(g))

```

The trick here is that the S4 classes are protected from incompatible S3 `data.frame` operations, such as `rbind()`, by being wrapped in their own `vctrs`-class. They still contain the beloved S4 classes though.

```{r}

class(ld$x)

class(Nightfall(ld$x))

```

## Footnotes

This package is still very much a work in progress. Since the Bioconductor ecosystem is rich with specialised infrastructure and classes, I haven't managed to try and test every class. However, most things should work with `Rle`, `IRanges` and `GRanges` classes. 

Furthermore, not all ggplot2 features lend themselves well to S4 vectors yet: I've yet to devise a polar coordinate system and non-identity scale transformations. Also not all position adjustments or layer stats are expected to work perfectly.

I'd also like to mention that this is not the only genomics-oriented ggplot package out there. Take a look at the [ggbio](https://www.bioconductor.org/packages/release/bioc/html/ggbio.html) package for instance. In my view, the main difference is that this package is not going to produce your pretty plots for you, but by extending ggplot2, this package allows the flexibility required to build your own genomics plots.