coding-in-r

name: coding-in-r description: How to write R well — renv, tidyverse/data.table, project paths, style, scripts, seeds. TRIGGER when authoring or editing .R/.Rmd/.qmd files. For running R on the Yale SOM HPC cluster (Slurm, renv on /gpfs), use running-r instead. related: - programming-and-coding - running-r - accelerating-python - code-review updated: 2026-05-22

Coding in R

Rule: renv, portable paths, script as source of truth. Never rely on the global workspace.

For cluster execution, also load running R.

Defaults

renv for package versions. Commit renv.lock and .Rprofile.
On SOM HPC: module load r; no rig or mise default.
pak for fast installs when available.
styler for formatting.
lintr for linting.
here::here() for paths. No setwd() in tracked scripts.
readr::read_csv() over base read.csv().

install.packages("renv")
renv::init()
renv::install(c("tidyverse", "data.table", "here", "fs", "styler", "lintr"))
renv::snapshot()

Restore with:

renv::restore()

On HPC, restore once before arrays. Do not let many workers install packages concurrently.

Layout

project/
├── renv.lock
├── .Rprofile
├── README.md
├── scripts/       # entry points
├── R/             # shared helpers
├── notebooks/     # .Rmd / .qmd
├── data/raw/      # immutable; usually ignored
├── data/derived/  # rebuildable
└── results/       # figures/tables/logs

Paths

Never commit setwd() or personal absolute paths.

library(here)
counts_path <- here("data", "raw", "counts.csv")
counts <- readr::read_csv(counts_path)

For machine-specific roots, read an environment variable from .Renviron or the shell.

Style

snake_case for variables/functions.
<- for assignment; = for arguments.
Native pipe |> unless %>% materially helps.
2-space indentation; aim for 80-char lines, max 120.
Named arguments: mean(x, na.rm = TRUE), not mean(x, T).
library() at top.
Never attach() or <<-.
set.seed(42) before stochastic work.

Data

Default to tidyverse for clarity. Use data.table when data are large, joins/grouping dominate, or in-place mutation matters. Benchmark before rewriting clear code for speed.

CLI skeleton

#!/usr/bin/env Rscript
suppressPackageStartupMessages({
  library(optparse)
})

option_list <- list(
  make_option(c("-i", "--input"), type = "character"),
  make_option(c("-o", "--output"), type = "character")
)
opt <- parse_args(OptionParser(option_list = option_list))
if (is.null(opt$input) || is.null(opt$output)) {
  stop("--input and --output are required", call. = FALSE)
}

set.seed(42)

Checklist

styler::style_dir(".") run
lintr::lint_dir(".") clean or justified
No setwd() or personal paths
set.seed() where needed
renv::snapshot() if deps changed
Small smoke test run
HPC uses module load r + renv, not rig/mise