AMRFinderPlus DM: Make reentrant

Installing databases with the AMRFinder Data Manager will cause multiple
entries into the .loc file with the same name. This does not make any
sense - if the database already exists, it should not be duplicated in the
loc file.

Author: jakobnissen

data_managers/data_manager_build_amrfinderplus/data_manager/data_manager_build_amrfinderplus.py

@@ -5,6 +5,7 @@
 from ftplib import FTP
 from io import BytesIO
 from pathlib import Path
+from typing import Optional, Set
 
 import pandas as pd
 
@@ -20,11 +21,13 @@ def __init__(
         db_name="amrfinderplus-db",
         amrfinderplus_version="latest",
         date_version=None,
+        known_databases: Optional[Set[str]] = None,
     ):
         self.data_table_name = amrfinderplus_database
         self._db_name = db_name
         self._amrfinderplus_version = amrfinderplus_version
         self._amrfinderplus_date_version = date_version
+        self._known_databases = known_databases or set()
         self.data_table_entry = None
         self.amrfinderplus_table_list = None
 
@@ -49,6 +52,11 @@ def get_data_manager(self):
         amrfinderplus_name = (
             f"V{self._amrfinderplus_version}" f"-{self._amrfinderplus_date_version}"
         )
+        # Galaxy data table values are unique row identifiers. If this value is
+        # already present, do not emit another row for the same database.
+        if amrfinderplus_value in self._known_databases:
+            self.amrfinderplus_table_list["data_tables"][self.data_table_name] = []
+            return self.amrfinderplus_table_list
         data_info = dict(
             value=amrfinderplus_value,
             name=amrfinderplus_name,
@@ -79,9 +87,10 @@ def __init__(
         date_version=None,
         amrfinderplus_db_path=None,
         test_mode=False,
+        known_databases: Optional[Set[str]] = None,
     ):
 
-        super().__init__()
+        super().__init__(known_databases=known_databases)
         self.json_file_path = json_file_path
         self._output_dir = output_dir
         self._ncbi_ftp_url = ncbi_url
@@ -98,6 +107,17 @@ def __init__(
         self.test_mode = test_mode
         self.amrfinderplus_db_path = amrfinderplus_db_path
 
+    @property
+    def amrfinderplus_value(self) -> str:
+        return (
+            f"amrfinderplus_V{self._amrfinderplus_version}"
+            f"_{self._amrfinderplus_date_version}"
+        )
+
+    @property
+    def is_known_database(self) -> bool:
+        return self.amrfinderplus_value in self._known_databases
+
     @staticmethod
     def subprocess_cmd(command, *args):
         """
@@ -116,6 +136,10 @@ def download_amrfinderplus_db(self):
         """
         Download the amrfinderplus database from the ncbi ftp server
         """
+        # Avoid overwriting files or appending duplicate .loc rows when this
+        # database value is already registered in Galaxy.
+        if self.is_known_database:
+            return
         self.amrfinderplus_db_path = f"{self._output_dir}/{self._db_name}"
         os.makedirs(self.amrfinderplus_db_path)
 
@@ -170,6 +194,8 @@ def make_hmm_profile(self):
         """
         Make the hmm profile using the AMR.LIB file previously download
         """
+        if self.is_known_database:
+            return
         hmm_file = Path(f"{self.amrfinderplus_db_path}/AMR.LIB")
         if Path.exists(hmm_file) and self.test_mode is False:
             self.subprocess_cmd("hmmpress", "-f", hmm_file)
@@ -208,6 +234,8 @@ def make_blastdb(self):
         """
         Index fasta file for blast
         """
+        if self.is_known_database:
+            return
         self.extract_filelist_makeblast()
         if self._amrfinderplus_version == "3.12":
             nucl_file_db_list = [
@@ -301,6 +329,11 @@ def parse_arguments():
         action="store_true",
         help="option to test the script with an lighted database",
     )
+    arg_parser.add_argument(
+        "--known_databases",
+        default="",
+        help="comma-separated list of installed amrfinderplus database values",
+    )
     return arg_parser.parse_args()
 
 
@@ -311,6 +344,7 @@ def main():
         date_version=all_args.db_date,
         json_file_path=all_args.data_manager_json,
         test_mode=all_args.test,
+        known_databases=set(filter(None, all_args.known_databases.split(","))),
     )
     amrfinderplus_download.read_json_input_file()
     amrfinderplus_download.download_amrfinderplus_db()

data_managers/data_manager_build_amrfinderplus/data_manager/data_manager_build_amrfinderplus.xml

@@ -6,12 +6,21 @@
     <expand macro="requirements"/>
     <command detect_errors="exit_code">
       <![CDATA[
+#set $data_table = $__app__.tool_data_tables.get('amrfinderplus_versioned_database')
+#if $data_table is not None and len($data_table.get_fields()) != 0
+    ## Pass installed values to the script so rerunning the data manager
+    ## does not append duplicate rows for an already-installed database.
+    #set $known_databases = '--known_databases=' + ','.join([row[0] for row in $data_table.get_fields()])
+#else
+    #set $known_databases = ''
+#end if
 python '$__tool_directory__/data_manager_build_amrfinderplus.py'
     '$output_file'
     --db_version '$database_list.database_version_select'
     #if $database_list.database_version_select != 'latest':
     --db_date '$database_list.database_date_select'
     #end if
+    $known_databases
     $test_data_manager
       ]]></command>
     <inputs>
@@ -49,7 +58,19 @@ python '$__tool_directory__/data_manager_build_amrfinderplus.py'
                 </assert_contents>
             </output>
         </test>
-        <!-- Test_2 DB 3.12 2024-01-31.1 -->
+        <!-- Test_2 already installed DB is not added again -->
+        <test expect_num_outputs="1">
+            <param name="test_data_manager" value="--known_databases amrfinderplus_V3.12_2024-05-02.2 --test"/>
+            <output name="output_file">
+                <assert_contents>
+                    <has_n_lines n="1"/>
+                    <has_text text="{&quot;data_tables&quot;"/>
+                    <has_text text="amrfinderplus_versioned_database"/>
+                    <has_text text='"amrfinderplus_versioned_database": []'/>
+                </assert_contents>
+            </output>
+        </test>
+        <!-- Test_3 DB 3.12 2024-01-31.1 -->
         <test expect_num_outputs="1">
             <param name="test_data_manager" value="--test"/>
             <conditional name="database_list">
@@ -66,7 +87,7 @@ python '$__tool_directory__/data_manager_build_amrfinderplus.py'
                 </assert_contents>
             </output>
         </test>
-        <!-- Test_3 DB 4.0 2025-07-16.1 -->
+        <!-- Test_4 DB 4.0 2025-07-16.1 -->
         <test expect_num_outputs="1">
             <param name="test_data_manager" value="--test"/>
             <conditional name="database_list">

data_managers/data_manager_build_amrfinderplus/test-data/amrfinderplus_versioned.loc.test

@@ -5,6 +5,4 @@
 #
 # for example
 amrfinderplus_V3.12_2024-05-02.2	V3.12-2024-05-02.2	3.12	amrfinderplus-db
-amrfinderplus_V3.12_2024-01-31.1	V3.12-2024-01-31.1	3.12	amrfinderplus-db
 amrfinderplus_V3.12_2024-05-02.2	V3.12-2024-05-02.2	3.12	/private/var/folders/7j/21czvpk170zcyjdbqrp0hmsh0000gn/T/tmp3d_yq23a/galaxy-dev/tool-data/amrfinderplus-db/amrfinderplus_V3.12_2024-05-02.2
-amrfinderplus_V3.12_2024-01-31.1	V3.12-2024-01-31.1	3.12	/private/var/folders/7j/21czvpk170zcyjdbqrp0hmsh0000gn/T/tmp3d_yq23a/galaxy-dev/tool-data/amrfinderplus-db/amrfinderplus_V3.12_2024-01-31.1