Merge pull request #151 from umccr/cwl_launch

Add CWL launcher scripts

inst/scripts/cwl_launch/funcs.R

@@ -0,0 +1,63 @@
+query_workflow_rnasum <- function(sbjids) {
+  wfrnms <- glue("umccr__automated__rnasum__{sbjids}") |>
+    paste(collapse = "|")
+  q1 <- glue(
+    "WHERE \"type_name\" = 'rnasum' AND REGEXP_LIKE(\"wfr_name\", '{wfrnms}') ",
+    "ORDER BY \"start\" DESC;"
+  )
+  rportal::portaldb_query_workflow(q1)
+}
+
+json_obj <- function(rnasum_version, workdir, outdir, arriba_pdf, arriba_tsv, dataset,
+                     dragen_fusions, dragen_mapping_metrics,
+                     manta_tsv, pcgr_tiers_tsv, purple_gene_tsv,
+                     salmon, sample_name, subject_id, library_id) {
+  report_dir <- glue("{subject_id}_{library_id}")
+  fc <- function(f) {
+    list(
+      class = "File",
+      path = f
+    )
+  }
+  list(
+    "Name" = glue("rnasum_{rnasum_version}_{report_dir}"),
+    "Input" = list(
+      "arriba_pdf" = fc(arriba_pdf),
+      "arriba_tsv" = fc(arriba_tsv),
+      "dataset" = dataset,
+      "dragen_fusions" = fc(dragen_fusions),
+      "dragen_mapping_metrics" = fc(dragen_mapping_metrics),
+      "manta_tsv" = fc(manta_tsv),
+      "pcgr_tiers_tsv" = fc(pcgr_tiers_tsv),
+      "purple_gene_tsv" = fc(purple_gene_tsv),
+      "report_dir" = report_dir,
+      "salmon" = fc(salmon),
+      "sample_name" = sample_name,
+      "subject_id" = subject_id
+    ),
+    "engineParameters" = list(
+      "workDirectory" = workdir,
+      "outputDirectory" = outdir,
+      "overrides" = list(
+        "#main" = list(
+          "requirements" = list(
+            "DockerRequirement" = list(
+              "dockerPull" = glue("ghcr.io/umccr/rnasum:{rnasum_version}")
+            )
+          )
+        )
+      )
+    )
+  )
+}
+
+launch_rnasum <- function(json_body, token, wfl_id = "wfl.87e07ae6b46645a181e04813de535216", wfl_vname = "0.5.0--304c3ac") {
+  token <- dracarys::ica_token_validate(token)
+  base_url <- "https://aps2.platform.illumina.com/v1"
+  httr::POST(
+    url = glue("{base_url}/workflows/{wfl_id}/versions/{wfl_vname}:launch"),
+    config = httr::add_headers(Authorization = glue::glue("Bearer {token}")),
+    body = json_body,
+    encode = "json"
+  )
+}

inst/scripts/cwl_launch/launch_dev.R

@@ -0,0 +1,114 @@
+#--- Launch rnasum on a set of sbjs ---#
+# 1. Grab metadata from prod portaldb for rnasum workflows that have succeeded
+#   - this will give us paths to input gds dirs (umccrise, dragen, arriba)
+# 2. Fish out the specific files required for rnasum from those input dirs, and
+#    generate their presigned urls
+# 3. Use the wfl.87e07ae6b46645a181e04813de535216 dev workflow to launch version
+#    0.5.0--304c3ac of the rnasum workflow.
+#   - this will take in a json object with the required inputs and
+#     engineParameters that override the docker container used.
+
+{
+  require(dplyr)
+  require(glue, include.only = "glue")
+  require(here, include.only = "here")
+  # devtools::install_github("umccr/rportal")
+  # devtools::install_github("umccr/dracarys")
+  require(rportal, include.only = "awsvault_profile")
+  require(dracarys, include.only = "ica_token_validate")
+  require(tibble, include.only = "tribble")
+  require(tidyr, include.only = "pivot_wider")
+  require(stringr, include.only = "str_replace")
+}
+
+# keep long functions in separate file
+source(here("inst/scripts/cwl_launch/funcs.R"))
+# connect to aws
+rportal::awsvault_profile("upro")
+# ica token (prod for presigned urls, dev for running)
+token_prod <- dracarys::ica_token_validate(Sys.getenv("ICA_ACCESS_TOKEN_PRO"))
+token_dev <- dracarys::ica_token_validate(Sys.getenv("ICA_ACCESS_TOKEN_DEV"))
+# subjects from prod you want inputs from
+sbjids <- c(
+  "SBJ02999",
+  "SBJ03038",
+  "SBJ03039",
+  "SBJ04660",
+  "SBJ04661",
+  "SBJ04662",
+  "SBJ04774",
+  "SBJ04775",
+  "SBJ04776",
+  "SBJ04777"
+)
+# grab rnasum workflow metadata for given sbjids
+pmeta_raw <- query_workflow_rnasum(sbjids)
+# grab top run of each sbj
+pmeta_tidy <- pmeta_raw |>
+  rportal::meta_rnasum() |>
+  filter(end_status == "Succeeded", rnasum_dataset == "PANCAN") |>
+  slice_head(n = 1, by = "SubjectID")
+
+# which files to fish for in the gds dirs
+rnasum_file_regex <- tibble::tribble(
+  ~regex, ~fun,
+  "quant\\.genes\\.sf$", "DragenWtsQuantGenesSfFile",
+  "fusion_candidates\\.final$", "DragenWtsFusionsFinalFile",
+  "fusions\\.pdf$", "ArribaFusionsPdfFile",
+  "fusions\\.tsv$", "ArribaFusionsTsvFile",
+  "somatic\\.pcgr\\.snvs_indels\\.tiers\\.tsv$", "PcgrTiersTsvFile",
+  "purple\\.cnv\\.gene\\.tsv$", "PurpleCnvGeneTsvFile",
+  "manta\\.tsv$", "MantaTsvFile",
+  "mapping_metrics\\.csv$", "MapMetricsFile"
+)
+
+# melt gds_indir columns to get a single column with paths to gds directories
+# of interest, and fish out files of interest from each of them and grab
+# presigned URLs
+urls <- pmeta_tidy |>
+  tidyr::pivot_longer(contains("gds_indir"), names_to = "folder_type", values_to = "gds_indir") |>
+  select("SubjectID", "LibraryID", "rnasum_dataset", "folder_type", "gds_indir") |>
+  rowwise() |>
+  mutate(
+    rnasum_files = list(
+      dracarys::gds_files_list_filter_relevant(
+        gdsdir = gds_indir, token = token_prod, include = "PresignedUrl",
+        page_size = 1000, regexes = rnasum_file_regex
+      )
+    )
+  )
+
+d <- urls |>
+  tidyr::unnest(rnasum_files) |>
+  select("SubjectID", "LibraryID", dataset = "rnasum_dataset", ftype = "type", url = "presigned_url") |>
+  tidyr::pivot_wider(names_from = ftype, values_from = url)
+
+for (i in seq_len(nrow(d))) {
+  v <- "0.6.5"
+  date_str <- Sys.time() |>
+    stringr::str_replace_all("[- :]", "") |>
+    stringr::str_replace("\\..*", "")
+  report_dir <- glue("{d$SubjectID[i]}_{d$LibraryID[i]}")
+  username <- "pdiakumis"
+  gds_path <- glue("gds://development/{username}/temp/rnasum/{v}/{report_dir}/{date_str}")
+  workdir <- glue("{gds_path}/work")
+  outdir <- glue("{gds_path}/output")
+  jbody <- json_obj(
+    rnasum_version = v,
+    workdir = workdir,
+    outdir = outdir,
+    arriba_pdf = d$ArribaFusionsPdfFile[i],
+    arriba_tsv = d$ArribaFusionsTsvFile[i],
+    dataset = d$dataset[i],
+    dragen_fusions = d$DragenWtsFusionsFinalFile[i],
+    dragen_mapping_metrics = d$MapMetricsFile[i],
+    manta_tsv = d$MantaTsvFile[i],
+    pcgr_tiers_tsv = d$PcgrTiersTsvFile[i],
+    purple_gene_tsv = d$PurpleCnvGeneTsvFile[i],
+    salmon = d$DragenWtsQuantGenesSfFile[i],
+    sample_name = d$SubjectID[i],
+    subject_id = d$SubjectID[i],
+    library_id = d$LibraryID[i]
+  )
+  launch_rnasum(json_body = jbody, token = token_dev)
+}