Kraken2 DM: Add 2024-09-04 prebuilt databases + core_nt and gtdb_geno…

…me_reps
galaxyproject · Dec 13, 2024 · 37a815c · 37a815c
1 parent 37428f0
commit 37a815c
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 2 deletions.
diff --git a/data_managers/data_manager_build_kraken2_database/data_manager/kraken2_build_database.py b/data_managers/data_manager_build_kraken2_database/data_manager/kraken2_build_database.py
@@ -66,6 +66,8 @@ class StandardPrebuiltSizes(Enum):
     pluspfp_08gb = "pluspfp_08gb"
     pluspfp_16gb = "pluspfp_16gb"
     eupathdb48 = "eupathdb48"
+    core_nt = "core_nt"
+    gtdb_genome_reps = "gtdb_genome_reps"
 
     def __str__(self):
         return self.value
@@ -148,6 +150,8 @@ def kraken2_build_standard_prebuilt(prebuilt_db, prebuilt_date, target_directory
         'pluspfp_08gb': "PlusPFP-8 (PlusPFP with DB capped at 8 GB)",
         'pluspfp_16gb': "PlusPFP-16 (PlusPFP with DB capped at 16 GB)",
         'eupathdb48': "EuPathDB-46",
+        'core_nt': "core_nt (Very large collection, inclusive of GenBank, RefSeq, TPA and PDB)",
+        'gtdb_genome_reps': "GTDB v220 (Bacterial and archaeal)",
     }
 
     database_value = "_".join([

diff --git a/data_managers/data_manager_build_kraken2_database/data_manager/kraken2_build_database.xml b/data_managers/data_manager_build_kraken2_database/data_manager/kraken2_build_database.xml
@@ -2,7 +2,7 @@
     <description>database builder</description>
     <macros>
         <token name="@TOOL_VERSION@">2.1.3</token>
-        <token name="@VERSION_SUFFIX@">4</token>
+        <token name="@VERSION_SUFFIX@">5</token>
         <token name="@PROFILE@">22.01</token>
         <xml name="common_params">
             <param name="kmer_len" type="integer" value="35" label="K-mer length in BP" />
@@ -44,6 +44,9 @@
         <xml name="pluspfp_16gb">
             <option value="pluspfp_16gb">PlusPFP-16 (PlusPFP with DB capped at 16 GB; ~15 GB)</option>
         </xml>
+        <xml name="core_nt">
+            <option value="core_nt">core_nt (Very large collection, inclusive of GenBank, RefSeq, TPA and PDB; ~182 GB)</option>
+        </xml>
     </macros>
     <xrefs>
         <xref type="bio.tools">kraken2</xref>
@@ -110,6 +113,7 @@ python '$__tool_directory__/kraken2_build_database.py'
             <when value="standard_prebuilt">
                 <conditional name="prebuild">
                     <param name="prebuilt_date" type="select" label="Select index build date">
+                        <option value="2024-09-04">September 4, 2024</option>
                         <option value="2024-06-05">June 5, 2024</option>
                         <option value="2024-01-12">January 12, 2024</option>
                         <option value="2023-06-05">June 5, 2023</option>
@@ -119,6 +123,21 @@ python '$__tool_directory__/kraken2_build_database.py'
                         <option value="2020-12-02">December 2, 2020</option>
                         <option value="2020-09-19">September 19, 2020</option>
                     </param>
+                    <when value="2024-09-04">
+                        <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
+                            <expand macro="viral"/>
+                            <expand macro="minusb"/>
+                            <expand macro="standard"/>
+                            <expand macro="standard_08gb"/>
+                            <expand macro="standard_16gb"/>
+                            <expand macro="pluspf"/>
+                            <expand macro="pluspf_08gb"/>
+                            <expand macro="pluspf_16gb"/>
+                            <expand macro="pluspfp"/>
+                            <expand macro="pluspfp_08gb"/>
+                            <expand macro="core_nt"/>
+                        </param>
+                    </when>
                     <when value="2024-06-05">
                         <param name="prebuilt_db" type="select" label="Select a prebuilt Refseq index to download">
                             <expand macro="viral"/>
@@ -229,9 +248,14 @@ python '$__tool_directory__/kraken2_build_database.py'
             <when value="special_prebuilt">
                 <conditional name="special_prebuild">
                     <param name="special_prebuilt_db" type="select" multiple="false" label="Select pre-built database to download">
-                        <option value="eupathdb48_20230407">EuPathDB-46 (April 7, 2023)</option>
+                        <option value="gtdb_genome_reps_20241109">GTDB v220 (Bacterial and archaeal; ~497 GB) (December 13, 2024)</option>
+                        <option value="eupathdb48_20230407">EuPathDB-46 (April 18, 2023)</option>
                         <option value="eupathdb48_20201113">EuPathDB-46 (November 13, 2020)</option>
                     </param>
+                    <when value="gtdb_genome_reps_20241109">
+                        <param name="prebuilt_db" type="hidden" value="gtdb_genome_reps"/>
+                        <param name="prebuilt_date" type="hidden" value="2024-11-09"/>
+                    </when>
                     <when value="eupathdb48_20230407">
                         <param name="prebuilt_db" type="hidden" value="eupathdb48"/>
                         <param name="prebuilt_date" type="hidden" value="2023-04-07"/>