From abfb5597b187314a8c4a3cd62620e1255b1d6ee9 Mon Sep 17 00:00:00 2001
From: pdiakumis <peterdiakumis@gmail.com>
Date: Thu, 10 Oct 2024 00:45:08 +1100
Subject: [PATCH] tsov2: ignore SmallVariants_Annotated.json.gz; better dl size
 info

---
 R/fs_icav1.R |  2 +-
 R/fs_s3.R    | 11 +++++++++--
 R/tsov2.R    |  2 +-
 3 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/R/fs_icav1.R b/R/fs_icav1.R
index 270c7fe..56c6ad4 100644
--- a/R/fs_icav1.R
+++ b/R/fs_icav1.R
@@ -189,10 +189,10 @@ dr_gds_download <- function(gdsdir, outdir, token = Sys.getenv("ICA_ACCESS_TOKEN
       gdspath = .data$path
     ) |>
     dplyr::select("type", "bname", "size", "lastmodified", "file_id", "localpath", "gdspath")
-  # download recognisable dracarys files to outdir/<mirrored-cloud-path>/{bname}
   tot_size <- d |>
     dplyr::summarise(tot_size = sum(.data$size)) |>
     dplyr::pull(tot_size)
+  # download recognisable dracarys files to outdir/<mirrored-cloud-path>/{bname}
   if (!dryrun) {
     txt <- paste0(
       "{e('arrow_heading_down')} {nrow(d)} files ({tot_size}): {.file {gdsdir}}\n"
diff --git a/R/fs_s3.R b/R/fs_s3.R
index eb314ce..f2e3226 100644
--- a/R/fs_s3.R
+++ b/R/fs_s3.R
@@ -142,10 +142,16 @@ dr_s3_download <- function(s3dir, outdir, max_objects = 100, pattern = NULL,
       s3path = .data$path
     ) |>
     dplyr::select("type", "bname", "size", "lastmodified", "localpath", "s3path")
+  tot_size <- d |>
+    dplyr::summarise(tot_size = sum(.data$size)) |>
+    dplyr::pull(tot_size)
   # download recognisable dracarys files to outdir/<mirrored-cloud-path>/{bname}
   if (!dryrun) {
-    cli::cli_alert_info("{date_log()} {e('arrow_heading_down')} Downloading files from {.file {s3dir}}")
-    d |>
+    txt <- paste0(
+      "{e('arrow_heading_down')} {nrow(d)} files ({tot_size}): {.file {s3dir}}\n"
+    )
+    cli::cli_alert_info(txt)
+    res <- d |>
       dplyr::rowwise() |>
       dplyr::mutate(
         s3bucket = sub("s3://(.*?)/.*", "\\1", .data$s3path),
@@ -158,6 +164,7 @@ dr_s3_download <- function(s3dir, outdir, max_objects = 100, pattern = NULL,
         localpath = normalizePath(.data$localpath)
       ) |>
       dplyr::select("type", "bname", "size", "lastmodified", "localpath", "s3path")
+    return(res)
   } else {
     cli::cli_alert_info("{date_log()} {e('camera')} Just list relevant files from {.file {s3dir}}")
     d |>
diff --git a/R/tsov2.R b/R/tsov2.R
index 9a41204..5e3e3a7 100644
--- a/R/tsov2.R
+++ b/R/tsov2.R
@@ -78,7 +78,7 @@ Wf_tso_ctdna_tumor_only_v2 <- R6::R6Class(
         glue("{res}/{pref}_CombinedVariantOutput\\.tsv$"), "cvo",
         glue("{res}/{pref}_Fusions\\.csv$"), "fus",
         glue("{res}/{pref}_MetricsOutput\\.tsv$"), "DOWNLOAD_ONLY",
-        glue("{res}/{pref}_SmallVariants_Annotated\\.json\\.gz$"), "DOWNLOAD_ONLY",
+        # glue("{res}/{pref}_SmallVariants_Annotated\\.json\\.gz$"), "DOWNLOAD_ONLY",
         glue("{li}/SampleAnalysisResults/{pref}_SampleAnalysisResults\\.json$"), "sar"
       )
       # DragenCaller