From ccd5074d873c24e0408c88cbb50e55a2feb97281 Mon Sep 17 00:00:00 2001
From: Marina Gourtovaia <mg8@sanger.ac.uk>
Date: Tue, 7 May 2024 11:27:19 +0100
Subject: [PATCH] Archive Illumina samplesheet if present

---
 Changes                                                 | 3 +++
 lib/npg_pipeline/function/run_data_to_irods_archiver.pm | 8 +++++++-
 t/20-function-run_data_to_irods_archiver.t              | 2 +-
 3 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/Changes b/Changes
index 8dcf4327..cfcb2ad2 100644
--- a/Changes
+++ b/Changes
@@ -2,6 +2,9 @@ LIST OF CHANGES
 ---------------
 
  - Removing Tidyp dependency from CI
+ - Added 'SampleSheet.csv' file from the top level of the run folder to
+   a list of archived run-level Illumina data. This file is only present
+   in MiSeq run folders.
 
 release 68.2.0
  - Added '--process_separately_lanes' to the pipeline to explicitly exclude
diff --git a/lib/npg_pipeline/function/run_data_to_irods_archiver.pm b/lib/npg_pipeline/function/run_data_to_irods_archiver.pm
index 221b3e44..1c07679e 100644
--- a/lib/npg_pipeline/function/run_data_to_irods_archiver.pm
+++ b/lib/npg_pipeline/function/run_data_to_irods_archiver.pm
@@ -19,6 +19,10 @@ override 'create' => sub {
     my $job_name_prefix = join q{_}, q{publish_run_data2irods}, $self->id_run();
     $self->assign_common_definition_attrs($ref, $job_name_prefix);
 
+    # Exclude directories, which might have copies of the top-level files
+    # that are being archived. SampleSheet.csv is expected to be present
+    # only for MiSeq runs.
+
     my $command = join q[ ],
       $PUBLISH_SCRIPT_NAME,
       q{--restart_file},     $self->restart_file_path($job_name_prefix),
@@ -27,8 +31,10 @@ override 'create' => sub {
       q{--source_directory}, $self->runfolder_path(),
       q{--include},          q['RunInfo.xml'],
       q{--include},          q['[Rr]unParameters.xml'],
+      q{--include},          q['SampleSheet.csv'],
       q{--include},          q[InterOp],
       q{--exclude},          q[Analysis],
+      q{--exclude},          q[Data],
       q{--id_run},           $self->id_run,
       q{--logconf},          $self->conf_file_path('log4perl_publish_illumina.conf');
 
@@ -95,7 +101,7 @@ Marina Gourtovaia
 
 =head1 LICENSE AND COPYRIGHT
 
-Copyright (C) 2019 Genome Research Ltd.
+Copyright (C) 2018,2019,2021,2022,2023,2024 Genome Research Ltd.
 
 This program is free software: you can redistribute it and/or modify
 it under the terms of the GNU General Public License as published by
diff --git a/t/20-function-run_data_to_irods_archiver.t b/t/20-function-run_data_to_irods_archiver.t
index 96b3d91c..a4b145a9 100644
--- a/t/20-function-run_data_to_irods_archiver.t
+++ b/t/20-function-run_data_to_irods_archiver.t
@@ -13,7 +13,7 @@ my $util = t::util->new();
 
 my $tmp_dir = $util->temp_directory();
 my $script = q{npg_publish_illumina_run.pl};
-my $inexcludes = qr/--include 'RunInfo\.xml' --include '\[Rr\]unParameters\.xml' --include InterOp --exclude Analysis/;
+my $inexcludes = qr/--include 'RunInfo\.xml' --include '\[Rr\]unParameters\.xml' --include 'SampleSheet\.csv' --include InterOp --exclude Analysis --exclude Data/;
 
 my $defaults = {
   default => {