From a5df18f2759e0778810729c1fabb065263d5a9ef Mon Sep 17 00:00:00 2001
From: pirovc <4673375+pirovc@users.noreply.github.com>
Date: Fri, 8 Apr 2022 09:44:24 +0200
Subject: [PATCH] v0.4.1 (#48)

* v0.4.1, no default -d, fix and improved debug -Z, fix link gtdb release 07-RS207

* threads on check missing files
---
 README.md                      |  15 ++---
 genome_updater.sh              | 100 +++++++++++++++++++--------------
 tests/integration_offline.bats |  61 +++++++++++---------
 tests/integration_online.bats  |  12 ++--
 4 files changed, 106 insertions(+), 82 deletions(-)

diff --git a/README.md b/README.md
index e717205..e18c571 100755
--- a/README.md
+++ b/README.md
@@ -199,10 +199,10 @@ or
 	┌─┐┌─┐┌┐┌┌─┐┌┬┐┌─┐    ┬ ┬┌─┐┌┬┐┌─┐┌┬┐┌─┐┬─┐
 	│ ┬├┤ ││││ ││││├┤     │ │├─┘ ││├─┤ │ ├┤ ├┬┘
 	└─┘└─┘┘└┘└─┘┴ ┴└─┘────└─┘┴  ─┴┘┴ ┴ ┴ └─┘┴└─
-	                                     v0.4.0 
+	                                     v0.4.1 
 
 	Database options:
-	 -d Database (comma-separated entries) [genbank, refseq]	Default: refseq
+	 -d Database (comma-separated entries) [genbank, refseq]
 
 	Organism options:
 	 -g Organism group (comma-separated entries) [archaea, bacteria, fungi, human, invertebrate, metagenomes, other, plant, protozoa, vertebrate_mammalian, vertebrate_other, viral]. Example: archaea,bacteria.
@@ -227,9 +227,9 @@ or
 		Default: 0
 	 -F custom filter for the assembly summary in the format colA:val1|colB:valX,valY (case insensitive). Example: -F "2:PRJNA12377,PRJNA670754|14:Partial" for column infos check ftp://ftp.ncbi.nlm.nih.gov/genomes/README_assembly_summary.txt
 		Default: ""
-	 -D Start date to keep sequences (>=), based on the sequence release date. Format YYYYMMDD. Example: -D 20201030
+	 -D Start date to keep sequences (>=), based on the sequence release date. Format YYYYMMDD. Example: 20201030
 		Default: ""
-	 -E End date to keep sequences (<=), based on the sequence release date. Format YYYYMMDD. Example: -D 20201231
+	 -E End date to keep sequences (<=), based on the sequence release date. Format YYYYMMDD. Example: 20201231
 		Default: ""
 	 -z Keep only assemblies present on the latest GTDB release
 
@@ -251,8 +251,8 @@ or
 		Default: ""
 	 -k Dry-run, no data is downloaded or updated - just checks for available sequences and changes
 	 -i Fix failed downloads or any incomplete data from a previous run, keep current version
-	 -m Check MD5 for downloaded files
-	 -t Threads
+	 -m Check MD5 of downloaded files
+	 -t Threads to parallelize download and some file operations
 		Default: 1
 
 	Misc. options:
@@ -263,7 +263,8 @@ or
 	 -n Conditional exit status. Exit Code = 1 if more than N files failed to download (integer for file number, float for percentage, 0 -> off)
 		Default: 0
 	 -V Verbose log to report successful file downloads
-	 -D Print print debug information and exit
+	 -Z Print debug information and run in debug mode
+
 
 ## References:
 
diff --git a/genome_updater.sh b/genome_updater.sh
index e7931a7..9df41b4 100755
--- a/genome_updater.sh
+++ b/genome_updater.sh
@@ -25,7 +25,7 @@ IFS=$' '
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 # THE SOFTWARE.
 
-version="0.4.0"
+version="0.4.1"
 genome_updater_args=$( printf "%q " "$@" )
 export genome_updater_args
 
@@ -44,7 +44,7 @@ use_curl=${use_curl:-0}
 # Export locale numeric to avoid errors on printf in different setups
 export LC_NUMERIC="en_US.UTF-8"
 
-gtdb_urls=( "https://data.gtdb.ecogenomic.org/releases/latest/ar122_taxonomy.tsv.gz" 
+gtdb_urls=( "https://data.gtdb.ecogenomic.org/releases/latest/ar53_taxonomy.tsv.gz" 
             "https://data.gtdb.ecogenomic.org/releases/latest/bac120_taxonomy.tsv.gz" )
 
 #activate aliases in the script
@@ -500,7 +500,7 @@ remove_files() # parameter: ${1} file, ${2} fields [assembly_accesion,url] OR fi
 check_missing_files() # ${1} file, ${2} fields [assembly_accesion,url], ${3} extension - returns assembly accession, url and filename
 {
     # Just returns if file doesn't exist or if it's zero size
-    list_files ${1} ${2} ${3} | xargs --no-run-if-empty -n3 sh -c 'if [ ! -s "'"${target_output_prefix}${files_dir}"'${2}" ]; then echo "${0}'$'\t''${1}'$'\t''${2}"; fi'
+    list_files ${1} ${2} ${3} | xargs -P "${threads}" --no-run-if-empty -n3 sh -c 'if [ ! -s "'"${target_output_prefix}${files_dir}"'${2}" ]; then echo "${0}'$'\t''${1}'$'\t''${2}"; fi'
 }
 
 check_complete_record() # parameters: ${1} file, ${2} field [assembly accession, url], ${3} extension - returns assembly accession, url
@@ -575,7 +575,7 @@ print_debug() # parameters: ${1} tools
 }
 
 # Defaults
-database="refseq"
+database=""
 organism_group=""
 species=""
 taxids=""
@@ -624,7 +624,7 @@ function showhelp {
     print_logo
     echo
     echo $'Database options:'
-    echo $' -d Database (comma-separated entries) [genbank, refseq]\tDefault: refseq'
+    echo $' -d Database (comma-separated entries) [genbank, refseq]'
     echo
     echo $'Organism options:'
     echo $' -g Organism group (comma-separated entries) [archaea, bacteria, fungi, human, invertebrate, metagenomes, other, plant, protozoa, vertebrate_mammalian, vertebrate_other, viral]. Example: archaea,bacteria.\n\tDefault: ""'
@@ -640,8 +640,8 @@ function showhelp {
     echo $' -P Number of top references for each species nodes to download. 0 for all. Selection order: RefSeq Category, Assembly level, Relation to type material, Date (most recent first)\n\tDefault: 0'
     echo $' -A Number of top references for each taxids (leaf nodes) to download. 0 for all. Selection order: RefSeq Category, Assembly level, Relation to type material, Date (most recent first)\n\tDefault: 0'
     echo $' -F custom filter for the assembly summary in the format colA:val1|colB:valX,valY (case insensitive). Example: -F "2:PRJNA12377,PRJNA670754|14:Partial" for column infos check ftp://ftp.ncbi.nlm.nih.gov/genomes/README_assembly_summary.txt\n\tDefault: ""'
-    echo $' -D Start date to keep sequences (>=), based on the sequence release date. Format YYYYMMDD. Example: -D 20201030\n\tDefault: ""'
-    echo $' -E End date to keep sequences (<=), based on the sequence release date. Format YYYYMMDD. Example: -D 20201231\n\tDefault: ""'
+    echo $' -D Start date to keep sequences (>=), based on the sequence release date. Format YYYYMMDD. Example: 20201030\n\tDefault: ""'
+    echo $' -E End date to keep sequences (<=), based on the sequence release date. Format YYYYMMDD. Example: 20201231\n\tDefault: ""'
     echo $' -z Keep only assemblies present on the latest GTDB release'
     echo
     echo $'Report options:'
@@ -657,8 +657,8 @@ function showhelp {
     echo $' -B Base label to use as the current version. Can be used to rollback to an older version or to create multiple branches from a base version. It only applies for updates. \n\tDefault: ""'
     echo $' -k Dry-run, no data is downloaded or updated - just checks for available sequences and changes'
     echo $' -i Fix failed downloads or any incomplete data from a previous run, keep current version'
-    echo $' -m Check MD5 for downloaded files'
-    echo $' -t Threads\n\tDefault: 1'
+    echo $' -m Check MD5 of downloaded files'
+    echo $' -t Threads to parallelize download and some file operations\n\tDefault: 1'
     echo
     echo $'Misc. options:'
     echo $' -x Allow the deletion of regular extra files if any found in the files folder. Symbolic links that do not belong to the current version will always be deleted.'
@@ -667,7 +667,7 @@ function showhelp {
     echo $' -w Silent output with download progress (%) and download version at the end'
     echo $' -n Conditional exit status. Exit Code = 1 if more than N files failed to download (integer for file number, float for percentage, 0 -> off)\n\tDefault: 0'
     echo $' -V Verbose log to report successful file downloads'
-    echo $' -D Print print debug information and exit'
+    echo $' -Z Print debug information and run in debug mode'
     echo
 }
 
@@ -690,52 +690,63 @@ done
 if [ "${tool_not_found}" -eq 1 ]; then exit 1; fi
 
 OPTIND=1 # Reset getopts
-while getopts "d:g:S:T:c:l:F:o:e:R:b:B:t:f:P:A:D:E:zn:akixmurpswhDV" opt; do
+while getopts "aA:b:B:d:D:c:De:E:f:F:g:hikl:mn:o:pP:rR:sS:t:T:uVwxzZ" opt; do
   case ${opt} in
-    d) database=${OPTARG} ;;
-    g) organism_group=${OPTARG// } ;; #remove spaces
-    S) species=${OPTARG// } ;; #remove spaces
-    T) taxids=${OPTARG// } ;; #remove spaces
-    c) refseq_category=${OPTARG} ;;
-    l) assembly_level=${OPTARG} ;;
-    F) custom_filter=${OPTARG} ;;
-    o) working_dir=${OPTARG} ;;
-    e) external_assembly_summary=${OPTARG} ;;
-    R) retry_download_batch=${OPTARG} ;;
+    a) download_taxonomy=1 ;;
+    A) top_assemblies_taxids=${OPTARG} ;;
     b) label=${OPTARG} ;;
     B) rollback_label=${OPTARG} ;;
-    t) threads=${OPTARG} ;;
-    f) file_formats=${OPTARG// } ;; #remove spaces
-    P) top_assemblies_species=${OPTARG} ;;
-    A) top_assemblies_taxids=${OPTARG} ;;
+    c) refseq_category=${OPTARG} ;;
+    d) database=${OPTARG} ;;
     D) date_start=${OPTARG} ;;
+    e) external_assembly_summary=${OPTARG} ;;
     E) date_end=${OPTARG} ;;
-    z) gtdb_only=1 ;;
-    a) download_taxonomy=1 ;;
-    k) dry_run=1 ;;
+    f) file_formats=${OPTARG// } ;; #remove spaces
+    F) custom_filter=${OPTARG} ;;
+    g) organism_group=${OPTARG// } ;; #remove spaces
+    h|\?) showhelp; exit 0 ;;
     i) just_fix=1 ;;
-    x) delete_extra_files=1 ;;
+    k) dry_run=1 ;;
+    l) assembly_level=${OPTARG} ;;
     m) check_md5=1 ;;
-    u) updated_assembly_accession=1 ;;
-    r) updated_sequence_accession=1 ;;
-    p) url_list=1 ;;
     n) conditional_exit=${OPTARG} ;;
+    o) working_dir=${OPTARG} ;;
+    p) url_list=1 ;;
+    P) top_assemblies_species=${OPTARG} ;;
+    r) updated_sequence_accession=1 ;;
+    R) retry_download_batch=${OPTARG} ;;
     s) silent=1 ;;
-    w) silent_progress=1 ;;
-    D) debug_mode=1 ;;
+    S) species=${OPTARG// } ;; #remove spaces
+    t) threads=${OPTARG} ;;
+    T) taxids=${OPTARG// } ;; #remove spaces
+    u) updated_assembly_accession=1 ;;
     V) verbose_log=1 ;;
-    h|\?) showhelp; exit 0 ;;
+    w) silent_progress=1 ;;
+    x) delete_extra_files=1 ;;
+    z) gtdb_only=1 ;;
+    Z) debug_mode=1 ;;
     :) echo "Option -${OPTARG} requires an argument." >&2; exit 1 ;;
   esac
 done
-if [ ${OPTIND} -eq 1 ]; then showhelp; exit 1; fi
-shift $((OPTIND-1))
-[ "${1:-}" = "--" ] && shift
 
+# Print tools and versions
 if [ "${debug_mode}" -eq 1 ] ; then 
     print_debug tools;
-    exit 0;
+    # If debug is the only parameter, exit, otherwise set debug mode for the run (set -x)
+    if [ ${OPTIND} -eq 2 ]; then
+        exit 0;
+    else
+        set -x
+    fi
+fi
+# No params
+if [ ${OPTIND} -eq 1 ]; then 
+    showhelp; 
+    exit 1;
 fi
+shift $((OPTIND-1))
+[ "${1:-}" = "--" ] && shift
+
 ######################### General parameter validation ######################### 
 if [[ -z "${database}" ]]; then
     echo "Database is required (-d)"; exit 1;
@@ -842,7 +853,6 @@ if [[ "${MODE}" == "UPDATE" ]] || [[ "${MODE}" == "FIX" ]]; then # get existing
         if [[ -f "${rollback_assembly_summary}" ]]; then
             rm ${default_assembly_summary}
             ln -s -r "${rollback_assembly_summary}" "${default_assembly_summary}"
-
         else
             echo "Rollback label/assembly_summary.txt not found ["${rollback_assembly_summary}"]"; exit 1
         fi
@@ -928,6 +938,10 @@ else
 fi
 echolog "-------------------------------------------" "1"
 
+if [ "${debug_mode}" -eq 1 ] ; then 
+    ls -laR "${working_dir}"
+fi
+
 # new
 if [[ "${MODE}" == "NEW" ]]; then
 
@@ -983,7 +997,6 @@ if [[ "${MODE}" == "NEW" ]]; then
             fi
             echolog "" "1"
         fi
-        
     fi
     
 else # update/fix
@@ -1176,6 +1189,11 @@ if [ "${dry_run}" -eq 0 ]; then
     if [ "${silent_progress}" -eq 1 ] ; then
         echo "$(dirname $(readlink -m ${default_assembly_summary}))"
     fi
+
+    if [ "${debug_mode}" -eq 1 ] ; then 
+        ls -laR "${working_dir}"
+    fi
+
     # Exit conditional status
     exit $(exit_status ${expected_files} ${current_files})
 fi
diff --git a/tests/integration_offline.bats b/tests/integration_offline.bats
index 6d6f7ca..e7b9e7a 100644
--- a/tests/integration_offline.bats
+++ b/tests/integration_offline.bats
@@ -25,6 +25,11 @@ setup_file() {
     assert_success
 }
 
+@test "Run genome_updater.sh and show debug info" {
+    run ./genome_updater.sh -Z
+    assert_success
+}
+
 @test "DB refseq" {
     outdir=${outprefix}db-refseq/
     label="test"
@@ -59,14 +64,14 @@ setup_file() {
 @test "Organism group archaea" {
     outdir=${outprefix}og-archaea/
     label="test"
-    run ./genome_updater.sh -o archaea -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -o archaea -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 }
 
 @test "Organism group archaea and fungi" {
     outdir=${outprefix}og-archaea-fungi/
     label="test"
-    run ./genome_updater.sh -o archaea,fungi -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -o archaea,fungi -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 }
 
@@ -78,7 +83,7 @@ setup_file() {
     #echo ${txids[@]} >&3
 
     # Use third
-    run ./genome_updater.sh -S "${txids[2]}" -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -S "${txids[2]}" -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 
     # Check if output contains only used taxids
@@ -98,7 +103,7 @@ setup_file() {
     #echo ${rscat[@]} >&3
 
     # Use first
-    run ./genome_updater.sh -c "${rscat[0]}" -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -c "${rscat[0]}" -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 
     # Check if output contains only selected refseq category
@@ -119,7 +124,7 @@ setup_file() {
     #echo ${aslev[@]} >&3
 
     # Use first
-    run ./genome_updater.sh -l "${aslev[0]}" -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -l "${aslev[0]}" -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 
     # Check if output contains only selected assembly level
@@ -141,7 +146,7 @@ setup_file() {
     aslev=( $(get_values_as ${local_dir}genomes/refseq/assembly_summary_refseq.txt 12 ) )
 
     # Simulate refseq category and assembly level filter using the custom filter
-    run ./genome_updater.sh -F "5:${rscat[0]}|12:${aslev[0]}" -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -F "5:${rscat[0]}|12:${aslev[0]}" -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 
     # Check if output contains only selected refseq category
@@ -253,7 +258,7 @@ setup_file() {
 @test "Report assembly accession" {
     outdir=${outprefix}report-assembly-accession/
     label="test"
-    run ./genome_updater.sh -b ${label} -o ${outdir} -u
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -u
     sanity_check ${outdir} ${label}
 
     # Check if report was printed and has all lines reported
@@ -265,7 +270,7 @@ setup_file() {
 @test "Report sequence accession" {
     outdir=${outprefix}report-sequence-accession/
     label="test"
-    run ./genome_updater.sh -b ${label} -o ${outdir} -r
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -r
     sanity_check ${outdir} ${label}
 
     # Check if report was printed
@@ -276,7 +281,7 @@ setup_file() {
 @test "Report urls" {
     outdir=${outprefix}report-urls/
     label="test"
-    run ./genome_updater.sh -b ${label} -o ${outdir} -p
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -p
     sanity_check ${outdir} ${label}
 
     # Check if report was printed and has all lines reported
@@ -292,7 +297,7 @@ setup_file() {
     outdir=${outprefix}external-assembly-summary/
     label="test"
     # Get assembly_summary from -e (not directly from url)
-    run ./genome_updater.sh -b ${label} -o ${outdir} -e ${local_dir}genomes/refseq/assembly_summary_refseq.txt
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -e ${local_dir}genomes/refseq/assembly_summary_refseq.txt
     sanity_check ${outdir} ${label}
 }
 
@@ -302,17 +307,17 @@ setup_file() {
     
     # Base version with only refseq
     label1="v1"
-    run ./genome_updater.sh -b ${label1} -o ${outdir} -d refseq
+    run ./genome_updater.sh -d refseq -b ${label1} -o ${outdir} -d refseq
     sanity_check ${outdir} ${label1}
 
     # Second version with more entries (refseq,genbank)
     label2="v2"
-    run ./genome_updater.sh -b ${label2} -o ${outdir} -d refseq,genbank
+    run ./genome_updater.sh -d refseq -b ${label2} -o ${outdir} -d refseq,genbank
     sanity_check ${outdir} ${label2}
 
     # Third version with same entries (nothing to download)
     label3="v3"
-    run ./genome_updater.sh -b ${label3} -o ${outdir} -d refseq,genbank
+    run ./genome_updater.sh -d refseq -b ${label3} -o ${outdir} -d refseq,genbank
     sanity_check ${outdir} ${label3}
 
     # Check log for no updates
@@ -321,7 +326,7 @@ setup_file() {
 
     # Fourth version with the same as second but rolling back from first, re-download files
     label4="v4"
-    run ./genome_updater.sh -b ${label4} -o ${outdir} -d refseq,genbank -B v1
+    run ./genome_updater.sh -d refseq -b ${label4} -o ${outdir} -d refseq,genbank -B v1
     sanity_check ${outdir} ${label4}
 
     # Check log for updates
@@ -332,13 +337,13 @@ setup_file() {
 @test "Delete extra files" {
     outdir=${outprefix}delete-extra-files/
     label="test"
-    run ./genome_updater.sh -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
     # Create extra files
     touch "${outdir}${label}/files/EXTRA_FILE.txt"
     assert_file_exist "${outdir}${label}/files/EXTRA_FILE.txt"
     # Run to fix and delete
-    run ./genome_updater.sh -b ${label} -o ${outdir} -i -x
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -i -x
     sanity_check ${outdir} ${label}
     # File was removed
     assert_not_exist "${outdir}${label}/files/EXTRA_FILE.txt"
@@ -350,7 +355,7 @@ setup_file() {
     # update label
     label="update"
     # Update (should not not carry extra file over to new version)
-    run ./genome_updater.sh -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 
     assert_not_exist "${outdir}${label}/files/ANOTHER_EXTRA_FILE.txt"
@@ -360,14 +365,14 @@ setup_file() {
 @test "Threads" {
     outdir=${outprefix}threads/
     label="test"
-    run ./genome_updater.sh -b ${label} -o ${outdir} -t 8
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -t 8
     sanity_check ${outdir} ${label}
 }
 
 @test "Silent" {
     outdir=${outprefix}silent/
     label="test"
-    run ./genome_updater.sh -b ${label} -o ${outdir} -s
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -s
     sanity_check ${outdir} ${label}
 
     # check if printed to STDOUT
@@ -379,7 +384,7 @@ setup_file() {
     label="test"
     use_curl=1
     export use_curl
-    run ./genome_updater.sh -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 }
 
@@ -388,24 +393,24 @@ setup_file() {
     label="test"
 
     # Dry-run NEW
-    run ./genome_updater.sh -b ${label} -o ${outdir} -k
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -k
     assert_success
     assert_dir_not_exist ${outdir}
 
     # Real run NEW
-    run ./genome_updater.sh -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 
     # Remove files to simulate failure
     rm ${outdir}${label}/files/*
 
     # Dry-run FIX
-    run ./genome_updater.sh -b ${label} -o ${outdir} -k -i
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -k -i
     assert_success
     assert_file_empty {outdir}${label}/files/
 
     # Real run FIX
-    run ./genome_updater.sh -b ${label} -o ${outdir} -i
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -i
     sanity_check ${outdir} ${label}
 }
 
@@ -414,20 +419,20 @@ setup_file() {
     label="test"
 
     # Dry-run NEW
-    run ./genome_updater.sh -b ${label} -o ${outdir} -k
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -k
     assert_success
     assert_dir_not_exist ${outdir}
 
     # Real run NEW
-    run ./genome_updater.sh -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 
     # Dry-run UPDATE (use another organism group to simulate change)
     label="update"
-    run ./genome_updater.sh -g archaea,fungi -b ${label} -o ${outdir} -k
+    run ./genome_updater.sh -d refseq -g archaea,fungi -b ${label} -o ${outdir} -k
     assert_success
 
     # Real run FIX
-    run ./genome_updater.sh -g archaea,fungi -b ${label} -o ${outdir}
+    run ./genome_updater.sh -d refseq -g archaea,fungi -b ${label} -o ${outdir}
     sanity_check ${outdir} ${label}
 }
diff --git a/tests/integration_online.bats b/tests/integration_online.bats
index 143d53c..e85df3f 100644
--- a/tests/integration_online.bats
+++ b/tests/integration_online.bats
@@ -28,7 +28,7 @@ setup_file() {
     label="test"
 
     # Protozoa in refseq is the smallest available assembly_summary at the time of writing this test (01.2022)
-    run ./genome_updater.sh -g protozoa -d refseq -b ${label} -t ${threads} -o ${outdir}
+    run ./genome_updater.sh -d refseq -g protozoa -b ${label} -t ${threads} -o ${outdir}
     sanity_check ${outdir} ${label}
 
     # Check filenames
@@ -40,14 +40,14 @@ setup_file() {
 @test "NA URL" {
     outdir=${outprefix}na-url/
     label="test"
-    run ./genome_updater.sh -b ${label} -o ${outdir} -t ${threads} -e ${files_dir}simulated/assembly_summary_na_url.txt
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -t ${threads} -e ${files_dir}simulated/assembly_summary_na_url.txt
     sanity_check ${outdir} ${label}
 }
 
 @test "All invalid URLs" {
     outdir=${outprefix}all-invalid-url/
     label="test"
-    run ./genome_updater.sh -b ${label} -o ${outdir} -t ${threads} -e ${files_dir}simulated/assembly_summary_all_invalid_url.txt
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -t ${threads} -e ${files_dir}simulated/assembly_summary_all_invalid_url.txt
     assert_success
     assert_equal $(count_files ${outdir} ${label}) 0
 }
@@ -55,7 +55,7 @@ setup_file() {
 @test "Some invalid URLs" {
     outdir=${outprefix}some-invalid-url/
     label="test"
-    run ./genome_updater.sh -b ${label} -o ${outdir} -t ${threads} -e ${files_dir}simulated/assembly_summary_some_invalid_url.txt
+    run ./genome_updater.sh -d refseq -b ${label} -o ${outdir} -t ${threads} -e ${files_dir}simulated/assembly_summary_some_invalid_url.txt
     assert_success
     assert_equal $(count_files ${outdir} ${label}) 2
 }
@@ -77,7 +77,7 @@ setup_file() {
     label="test"
 
     # 5690 Trypanosoma genus - around 6 genomes, get only one per species (01.2022)
-    run ./genome_updater.sh -g protozoa -T 5690 -P 1 -b ${label} -o ${outdir} -t ${threads} 
+    run ./genome_updater.sh -d refseq -g protozoa -T 5690 -P 1 -b ${label} -o ${outdir} -t ${threads} 
     sanity_check ${outdir} ${label}
 
     # Get counts of species taxids on output
@@ -108,7 +108,7 @@ setup_file() {
     label="test"
 
     # 5693 Trypanosoma cruzi
-    run ./genome_updater.sh -e ${files_dir}simulated/assembly_summary_gtdb.txt -b ${label} -o ${outdir} -t ${threads} -z
+    run ./genome_updater.sh -d refseq -e ${files_dir}simulated/assembly_summary_gtdb.txt -b ${label} -o ${outdir} -t ${threads} -z
     sanity_check ${outdir} ${label}
 
     # 1 out of 2 available on GTDB