Sequencing-related orgs' GitHub repos and licensing

gh-orgs-licenses.R

# ------------------------------------------------------------
# GitHub org license audit + stacked bar chart (R / tidyverse)
# ------------------------------------------------------------
# Prereqs:
# install.packages(c("gh","dplyr","purrr","tidyr","stringr","ggplot2","forcats"))
# Auth: Sys.setenv(GITHUB_PAT = "<your fine-grained or classic PAT>")  # optional but recommended

library(gh)
library(dplyr)
library(purrr)
library(tidyr)
library(stringr)
library(ggplot2)
library(forcats)

# ---------------------------
# 1) INPUT: org IDs to check
# ---------------------------
orgs <- c(
  "sbg",
  "bionanogenomics",
  "BioSkryb",
  "BGIResearch",
  "nebiolabs",
  "GenapsysInc",
  "fulcrumgenomics",
  "ATCC-Bioinformatics",
  "seqeralabs",
  "dnanexus",
  "latchbio",
  "Elembio",
  "thermofisher",
  "Ultimagen",
  "Singular-Genomics",
  "Nanostring-Biostats",
  "MGI-tech-bioinformatics",
  "becls",
  "insightsengineering",
  "10xgenomics",
  "pacificbiosciences",
  "nanoporetech",
  "illumina"
)

# Feel free to replace with your own list.

# ----------------------------------------------
# 2) Helper: classify license into three buckets
# ----------------------------------------------
classify_license <- function(spdx_id, license_name) {
  # Normalize
  spdx_id <- toupper(trimws(coalesce(spdx_id, "")))
  lname <- tolower(trimws(coalesce(license_name, "")))

  # Buckets (edit to taste):
  open_spdx <- c(
    "MIT",
    "BSD-2-CLAUSE",
    "BSD-3-CLAUSE",
    "APACHE-2.0",
    "ISC",
    "UNLICENSE",
    "GPL-2.0",
    "GPL-3.0",
    "AGPL-3.0",
    "LGPL-2.1",
    "LGPL-3.0",
    "MPL-2.0",
    "EPL-2.0",
    "Unlicense",
    "Zlib",
    "CC-BY-4.0",
    "CC-BY-SA-4.0",
    "CC0-1.0",
    "Zlib"
  )

  restrictive_spdx <- c(
    "CC-BY-NC-4.0",
    "CC-BY-NC-SA-4.0",
    "PROPRIETARY",
    "NON-SPDX LICENSE"
  )
  # Some repos use non-SPDX names; simple string heuristics as a fallback:
  is_open_name <- function(x) {
    str_detect(x, "(mit|bsd|apache|gpl|lgpl|agpl|mpl|isc|unlicense|zlib)")
  }
  is_restrict_name <- function(x) {
    str_detect(x, "(non-?commercial|proprietary|no\\s*license)")
  }

  if (spdx_id %in% open_spdx || is_open_name(lname)) {
    "open"
  } else if (spdx_id %in% restrictive_spdx || is_restrict_name(lname)) {
    "restrictive"
  } else if (spdx_id == "" && lname == "") {
    "restrictive/unknown"
  } else {
    # Unknown / ambiguous licenses default to "none" (adjust if you prefer a separate "unknown")
    "restrictive/unknown"
  }
}

# -----------------------------------------------------
# 3) Fetch repos for an org (safe wrapper + pagination)
# -----------------------------------------------------
get_org_repos <- function(org) {
  # gh() handles pagination with .limit = Inf; returns public repos
  res <- tryCatch(
    gh(
      "/orgs/{org}/repos",
      org = org,
      .limit = Inf,
      per_page = 100,
      type = "public"
    ),
    error = function(e) NULL
  )
  if (is.null(res) || length(res) == 0) {
    return(tibble(
      org = org,
      repo = character(),
      spdx_id = character(),
      license_name = character()
    ))
  }

  tibble(
    org = org,
    repo = map_chr(res, ~ .x$name %||% NA_character_),
    spdx_id = map_chr(res, ~ .x$license$spdx_id %||% NA_character_),
    license_name = map_chr(res, ~ .x$license$name %||% NA_character_)
  )
}

# --------------------------------------
# 4) Pull everything + classify licenses
# --------------------------------------
repo_df <- map_dfr(orgs, get_org_repos)
repo_df <-
  repo_df |>
  mutate(
    license_class = pmap_chr(
      list(spdx_id, license_name),
      ~ classify_license(..1, ..2)
    )
  )

# If an org has zero public repos or fetch failed, ensure it still shows up with 0s
counts_df <- repo_df %>%
  mutate(
    license_class = factor(
      license_class,
      levels = c("restrictive/unknown", "open")
    )
  ) %>%
  count(org, license_class, name = "n", .drop = FALSE) %>%
  group_by(org) %>%
  complete(license_class, fill = list(n = 0)) %>%
  ungroup()

totals_df <- counts_df %>%
  group_by(org) %>%
  summarise(total_repos = sum(n), .groups = "drop") %>%
  arrange(desc(total_repos))

# ----------------------------------------
# 5) Plot: stacked bars by license classes
# ----------------------------------------
# Order orgs by total repos (descending) for readability
counts_df <- counts_df %>%
  left_join(totals_df, by = "org") %>%
  mutate(org = fct_reorder(org, total_repos))

p <- ggplot(counts_df, aes(x = org, y = n, fill = license_class)) +
  geom_col() +
  coord_flip() +
  labs(
    title = "GitHub repositories by organization and license class",
    x = "Organization",
    y = "Number of repositories",
    fill = "License class"
  ) +
  theme_minimal(base_size = 12)

print(p)

# Export to clipboard for datawrapper
counts_df |>
  select(-total_repos) |>
  pivot_wider(
    names_from = license_class,
    values_from = n,
    values_fill = 0
  ) |>
  clipr::write_clip()