[TheiaProk - GAMBIT and GAMBIT_Query] Update to latest db v2.0.0-2024…

…0628 (#539) * update gambit db to v2.0.0 * remove candidate sub-speciation from merlin_tag * add missing import * update CI * remove sub-speciation from gambit_predicted_taxon * update md5sum --------- Co-authored-by: Sage Wright <[email protected]>
theiagen · Aug 16, 2024 · 1508bb5 · 1508bb5
1 parent af2f76b
commit 1508bb5
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 7 deletions.
diff --git a/tasks/taxon_id/task_gambit.wdl b/tasks/taxon_id/task_gambit.wdl
@@ -5,8 +5,8 @@ task gambit {
     File assembly
     String samplename
     String docker = "us-docker.pkg.dev/general-theiagen/staphb/gambit:1.0.0"
-    File gambit_db_genomes = "gs://gambit-databases-rp/1.3.0/gambit-metadata-1.3-231016.gdb"
-    File gambit_db_signatures = "gs://gambit-databases-rp/1.3.0/gambit-signatures-1.3-231016.gs"
+    File gambit_db_genomes = "gs://gambit-databases-rp/2.0.0/gambit-metadata-2.0.0-20240628.gdb"
+    File gambit_db_signatures = "gs://gambit-databases-rp/2.0.0/gambit-signatures-2.0.0-20240628.gs"
     Int disk_size = 20
     Int memory = 2
     Int cpu = 1
@@ -40,6 +40,7 @@ task gambit {
     python3 <<EOF
     import json
     import csv
+    import re
 
     def fmt_dist(d): return format(d, '.4f')
 
@@ -65,7 +66,13 @@ task gambit {
           if str(empty_value) == str(fmt_dist(0)):
             f.write(fmt_dist(search_item[column]))
           else:
-            f.write(search_item[column])
+            # remove candidate sub-speciation from taxon name
+            if column == 'name':
+              gambit_name = search_item[column]
+              gambit_name = re.sub(r'_[A-Za-z]+', '', gambit_name)  # This line is added to remove _X where X is any letter
+              f.write(gambit_name)
+            else:
+              f.write(search_item[column])
 
     # Predicted taxon    
     write_output('PREDICTED_TAXON', predicted, 'name', 'NA')
@@ -121,6 +128,8 @@ task gambit {
 
     try:
       merlin_tag = predicted['name']
+      # remove candidate sub-speciation from merlin_tag
+      merlin_tag = re.sub(r'_[A-Za-z]+', '', merlin_tag)  # This line is added to remove _X where X is any letter
     except:
       merlin_tag = "NA"
 

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_pe.yml
@@ -220,7 +220,7 @@
     - path: miniwdl_run/call-clean_check_reads/work/_miniwdl_inputs/0/test_1.clean.fastq.gz
     - path: miniwdl_run/call-clean_check_reads/work/_miniwdl_inputs/0/test_2.clean.fastq.gz
     - path: miniwdl_run/call-gambit/command
-      md5sum: 6f2537d9aa54eac508dc393d27772b76
+      md5sum: 8694687ae578a88817bab0865cceb652
     - path: miniwdl_run/call-gambit/inputs.json
       contains: ["assembly", "fasta", "samplename", "test"]
     - path: miniwdl_run/call-gambit/outputs.json
@@ -623,7 +623,7 @@
     - path: miniwdl_run/wdl/tasks/task_versioning.wdl
       md5sum: c20b66eea46148f4618abc038d3877b7
     - path: miniwdl_run/wdl/tasks/taxon_id/task_gambit.wdl
-      md5sum: f4826d61709d0d44b921a821d0d5706f
+      md5sum: 2aa70eab24868920f6c28843dd3b5613
     - path: miniwdl_run/wdl/tasks/taxon_id/contamination/task_kraken2.wdl
       md5sum: 0ea83681884800bda1e3c4e116f2b19d
     - path: miniwdl_run/wdl/tasks/taxon_id/contamination/task_midas.wdl

diff --git a/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml b/tests/workflows/theiaprok/test_wf_theiaprok_illumina_se.yml
@@ -212,7 +212,7 @@
       md5sum: 32c0be4fb7f3030bf9c74c0a836d4f2e
     - path: miniwdl_run/call-clean_check_reads/work/_miniwdl_inputs/0/test_1.clean.fastq.gz
     - path: miniwdl_run/call-gambit/command
-      md5sum: 6f2537d9aa54eac508dc393d27772b76
+      md5sum: 8694687ae578a88817bab0865cceb652
     - path: miniwdl_run/call-gambit/inputs.json
       contains: ["assembly", "fasta", "samplename", "test"]
     - path: miniwdl_run/call-gambit/outputs.json
@@ -586,7 +586,7 @@
     - path: miniwdl_run/wdl/tasks/task_versioning.wdl
       md5sum: c20b66eea46148f4618abc038d3877b7
     - path: miniwdl_run/wdl/tasks/taxon_id/task_gambit.wdl
-      md5sum: f4826d61709d0d44b921a821d0d5706f
+      md5sum: 2aa70eab24868920f6c28843dd3b5613
     - path: miniwdl_run/wdl/tasks/taxon_id/contamination/task_kraken2.wdl
       md5sum: 0ea83681884800bda1e3c4e116f2b19d
     - path: miniwdl_run/wdl/tasks/taxon_id/contamination/task_midas.wdl