Merge pull request #13 from umccr/amber-cobalt-share

Share Amber/Cobalt/Signature results

R/query.R

@@ -100,8 +100,16 @@ portaldb_query_workflow <- function(query = NULL) {
 #' @examples
 #' \dontrun{
 #' libids <- shQuote(paste(c("L2400340", "L2400256"), collapse = "|"))
-#' query <- glue("WHERE REGEXP_LIKE(\"library_id\", {libids});")
-#' portaldb_query_limsrow(query)
+#' query1 <- glue("WHERE REGEXP_LIKE(\"library_id\", {libids});")
+#' portaldb_query_limsrow(query1)
+#' sbjids <- paste(c("SBJ04470", "SBJ04487", "SBJ04488"), collapse = "|")
+#' query2 <- glue(
+#'   "WHERE REGEXP_LIKE(\"subject_id\", '{sbjids}') AND \"type\" = 'WGS' ",
+#'   "AND \"phenotype\" = 'tumor' ORDER BY \"subject_id\" DESC;"
+#' )
+#' d <- portaldb_query_limsrow(query2)
+#' # get tumor libids for each sbjid
+#' d |> dplyr::select(subject_id, library_id)
 #' }
 #' @export
 portaldb_query_limsrow <- function(query = NULL) {
@@ -121,7 +129,7 @@ portaldb_query_limsrow <- function(query = NULL) {
 #'
 #' @examples
 #' \dontrun{
-#' libids <- shQuote(paste(c("L2100192"), collapse = "|"))
+#' libids <- shQuote(paste(c("L2100192", "L2100191"), collapse = "|"))
 #' query <- glue("WHERE REGEXP_LIKE(\"rglb\", {libids});")
 #' res <- portaldb_query_fastqlistrow(query)
 #' }

inst/scripts/datashare/datashare.R

@@ -179,14 +179,30 @@ SampleID_tumor <- d_um_tidy[["SampleID_tumor"]]
 LibraryID_normal <- d_um_tidy[["LibraryID_normal"]]
 sbjid_sampid_dir <- glue("{SubjectID}__{SampleID_tumor}")
 umccrise_dir <- file.path(d_um_tidy[["gds_outdir_umccrise"]], sbjid_sampid_dir)
-d_um_urls1 <- dracarys::gds_files_list_filter_relevant(
-  gdsdir = umccrise_dir, token = token_ica,
-  include_url = TRUE, page_size = 500, regexes = umccrise_files
-)
-d_um_urls2 <- dracarys::gds_files_list_filter_relevant(
-  gdsdir = d_um_tidy[["gds_indir_dragen_somatic"]], token = token_ica,
-  include_url = TRUE, page_size = 500, regexes = tn_files
-)
+umccrise_work_dir <- file.path(d_um_tidy[["gds_outdir_umccrise"]], "work", sbjid_sampid_dir)
+amber_dir <- file.path(umccrise_work_dir, "purple/amber")
+cobalt_dir <- file.path(umccrise_work_dir, "purple/cobalt")
+sigs_dir <- file.path(umccrise_dir, "cancer_report_tables/sigs")
+d_um_urls1 <- umccrise_dir |>
+  dracarys::gds_files_list_filter_relevant(
+    token = token_ica, include_url = TRUE, page_size = 500, regexes = umccrise_files
+  )
+d_um_urls_sigs <- sigs_dir |>
+  dracarys::gds_files_list(token = token_ica, include_url = TRUE, page_size = 100) |>
+  mutate(type = "Signatures") |>
+  select("type", "bname", "size", "file_id", "path", "presigned_url")
+d_um_urls_amber <- amber_dir |>
+  dracarys::gds_files_list(token = token_ica, include_url = TRUE, page_size = 100) |>
+  mutate(type = "AMBER") |>
+  select("type", "bname", "size", "file_id", "path", "presigned_url")
+d_um_urls_cobalt <- cobalt_dir |>
+  dracarys::gds_files_list(token = token_ica, include_url = TRUE, page_size = 100) |>
+  mutate(type = "COBALT") |>
+  select("type", "bname", "size", "file_id", "path", "presigned_url")
+d_um_urls2 <- d_um_tidy[["gds_indir_dragen_somatic"]] |>
+  dracarys::gds_files_list_filter_relevant(
+    token = token_ica, include_url = TRUE, page_size = 500, regexes = tn_files
+  )
 fq_list <- d_tn_tidy |>
   select("fastq_tumor", "fastq_normal") |>
   pivot_longer(cols = c("fastq_tumor", "fastq_normal"), names_to = "fastq_tn") |>
@@ -225,6 +241,7 @@ if ((nrow(d_um_urls2) != nrow(tn_files)) | ((nrow(d_um_urls1) != nrow(umccrise_f
 }
 urls_all <- bind_rows(d_um_urls1, d_um_urls2, fq_urls) |>
   arrange(type) |>
+  bind_rows(d_um_urls_amber, d_um_urls_cobalt, d_um_urls_sigs) |>
   mutate(
     sbjid_libid = glue("{SubjectID}__{LibraryID_tumor}"),
     path = sub("gds://", "", .data$path),

inst/scripts/datashare/test.sh

@@ -2,9 +2,9 @@
 
 set -euo pipefail
 
-./datashare.R --subject_id SBJ03144 --library_id_tumor L2301290
-./datashare.R --subject_id SBJ04397 --library_id_tumor L2301291
-./datashare.R --subject_id SBJ04398 --library_id_tumor L2301292
-./datashare.R --subject_id SBJ04399 --library_id_tumor L2301293
-./datashare.R --subject_id SBJ04400 --library_id_tumor L2301294
-./datashare.R --subject_id SBJ04400 --library_id_tumor L2301295
+./datashare.R --subject_id SBJ03144 --library_id_tumor L2301290 --csv_output urls.csv
+./datashare.R --subject_id SBJ04397 --library_id_tumor L2301291 --csv_output urls.csv --append
+./datashare.R --subject_id SBJ04398 --library_id_tumor L2301292 --csv_output urls.csv --append
+./datashare.R --subject_id SBJ04399 --library_id_tumor L2301293 --csv_output urls.csv --append
+./datashare.R --subject_id SBJ04400 --library_id_tumor L2301294 --csv_output urls.csv --append
+./datashare.R --subject_id SBJ04400 --library_id_tumor L2301295 --csv_output urls.csv --append