Query TCGA API for exome capture BED files -- thanks to Sehyun Oh

tcga_capture.R

#!/usr/bin/env Rscript

library(GenomicDataCommons)
library(dplyr)

manifest = GenomicDataCommons::files() %>%
  GenomicDataCommons::filter(~ cases.project.project_id == "TCGA-LUAD" &
           experimental_strategy == "WXS" &
           data_format == "BAM") %>%
  GenomicDataCommons::manifest()

manifest = manifest %>% 
	slice(1:10) %>%
	select(id, filename)

res = lapply(manifest$id, function(uuid) {
  #con = curl::curl(paste0("https://api.gdc.cancer.gov/files/", uuid, "?pretty=true&fields=analysis.metadata.read_groups.target_capture_kit_target_region,analysis.metadata.read_groups.target_capture_kit_name,analysis.metadata.read_groups.target_capture_kit_vendor,analysis.metadata.read_groups.target_capture_kit_catalog_number"))
  con = curl::curl(paste0("https://api.gdc.cancer.gov/files/", uuid, "?pretty=true&fields=analysis.metadata.read_groups.target_capture_kit_target_region"))
  x = jsonlite::fromJSON(con)
  return(x)
})

y = lapply(res, function(x) unique(x$data$analysis$metadata$read_groups))
y = do.call(rbind, y)
manifest = cbind(manifest, y)
print(manifest)