Exploring R and Rust in Bioinformatics
Thank you!
About Mossa
- Maintainer & Lead Developer of extendr
- PhD Veterinary Epidemiology
- Postdoc @ Københavns Universitet - University of Copenhagen
About Josiah
- Sr. Product Engineer @ Esri
- Spatial Statistician
- Building R packages for the R-ArcGIS Bridge
What is extendr?
extending R with Rust
What is extendr?
- Include Rust into R packages
- Akin to Rcpp & cpp11
What even is Rust?
- Low-level language
- Modern day equivalent to C++
- Memory safe-er
- Developer friendly
Scientific computing with Rust is growing
extendr is comprised of:
- {rextendr} - usethis-like R package
- extendr-api - the “core” Rust library
- extendr-ffi - bindings to R’s C API
- extendr-engine - R engine for embedding R into Rust
extendr packages
- 16 CRAN packages
- Many, many more on GitHub
![]()
380k downloads and counting
Why Rust?
- Easy to make bindings
Blazingly fast
The compiler is nice
The compiler is nice
Easy (enough) to refactor
Broaden your reach
Quick to pickup
A little skill goes a long way
Example: single
#[extendr] fn gh_encode(x: f64, y: f64, length: usize) -> String { let coord = Coord { x, y }; encode(coord, length).expect("Failed to encode the geohash") }#[extendr] fn gh_encode(x: f64, y: f64, length: usize) -> String { let coord = Coord { x, y }; encode(coord, length).expect("Failed to encode the geohash") }#[extendr] fn gh_encode(x: f64, y: f64, length: usize) -> String { let coord = Coord { x, y }; encode(coord, length).expect("Failed to encode the geohash") }#[extendr] fn gh_encode(x: f64, y: f64, length: usize) -> String { let coord = Coord { x, y }; encode(coord, length).expect("Failed to encode the geohash") }
Example: vectorize
#[extendr]
fn gh_encode(x: &[f64], y: &[f64], length: usize) -> Vec<String> {
x
.into_iter()
.zip(y.into_iter())
.map(|(xi, yi)| {
let coord = Coord { x: xi, y: yi };
encode(coord, length)
.expect("Failed to encode the geohash")
})
.collect::<Vec<_>()
}#[extendr]
fn gh_encode(x: &[f64], y: &[f64], length: usize) -> Vec<String> {
x
.into_iter()
.zip(y.into_iter())
.map(|(xi, yi)| {
let coord = Coord { x: xi, y: yi };
encode(coord, length)
.expect("Failed to encode the geohash")
})
.collect::<Vec<_>()
}#[extendr]
fn gh_encode(x: &[f64], y: &[f64], length: usize) -> Vec<String> {
x
.into_iter()
.zip(y.into_iter())
.map(|(xi, yi)| {
let coord = Coord { x: xi, y: yi };
encode(coord, length)
.expect("Failed to encode the geohash")
})
.collect::<Vec<_>()
}#[extendr]
fn gh_encode(x: &[f64], y: &[f64], length: usize) -> Vec<String> {
x
.into_iter()
.zip(y.into_iter())
.map(|(xi, yi)| {
let coord = Coord { x: xi, y: yi };
encode(coord, length)
.expect("Failed to encode the geohash")
})
.collect::<Vec<_>()
}#[extendr]
fn gh_encode(x: &[f64], y: &[f64], length: usize) -> Vec<String> {
x
.into_iter()
.zip(y.into_iter())
.map(|(xi, yi)| {
let coord = Coord { x: xi, y: yi };
encode(coord, length)
.expect("Failed to encode the geohash")
})
.collect::<Vec<_>()
}#[extendr]
fn gh_encode(x: &[f64], y: &[f64], length: usize) -> Vec<String> {
x
.into_iter()
.zip(y.into_iter())
.map(|(xi, yi)| {
let coord = Coord { x: xi, y: yi };
encode(coord, length)
.expect("Failed to encode the geohash")
})
.collect::<Vec<_>()
}Example: parallelize
#[extendr]
fn gh_encode(x: &[f64], y: &[f64], length: usize) -> Vec<String> {
x
.into_iter()
.zip(y.into_iter())
.par_bridge() // convert into a parallel iterator
.with_min_len(1024) // set minimum parallel chunk length
.map(|(xi, yi)| {
let coord = Coord { x: xi, y: yi };
encode(coord, length)
.expect("Failed to encode the geohash")
})
.collect::<Vec<_>()
}extendr in bioformatics
WebGestaltR:
metabodecon
mspredictr
viewmastR
extendr across-language
Phylo2Vec
Exon
and even more
Exploring R and Rust in Bioinformatics