From 325caaf668983d86c879ec6f4699631136e4c390 Mon Sep 17 00:00:00 2001
From: Tianhao-Gu <tgu@anl.gov>
Date: Tue, 2 Jul 2024 15:51:44 -0500
Subject: [PATCH] fix erros when genome folder is an integer

---
 src/loaders/compute_tools/checkm2/versions.yaml    | 5 +++++
 src/loaders/compute_tools/eggnog/versions.yaml     | 6 ++++++
 src/loaders/compute_tools/gtdb_tk/versions.yaml    | 5 +++++
 src/loaders/compute_tools/mash/versions.yaml       | 6 +++++-
 src/loaders/compute_tools/microtrait/versions.yaml | 6 +++++-
 src/loaders/compute_tools/tool_common.py           | 2 +-
 6 files changed, 27 insertions(+), 3 deletions(-)

diff --git a/src/loaders/compute_tools/checkm2/versions.yaml b/src/loaders/compute_tools/checkm2/versions.yaml
index a79ca09a..7c880d10 100644
--- a/src/loaders/compute_tools/checkm2/versions.yaml
+++ b/src/loaders/compute_tools/checkm2/versions.yaml
@@ -26,4 +26,9 @@ versions:
       - Remove the 'threads' and 'program_threads' parameters 
         and introduce the capability to specify the number of threads per tool run
       - pass job_id to the tool container and remove node_id
+    reference_db_version: 1.0.1
+  - version: 0.1.5
+    date: 2024-06-25
+    notes: |
+      - Bug fix: tool_common.py - Converted Data IDs to string format to ensure proper comparison with associated folder names
     reference_db_version: 1.0.1
\ No newline at end of file
diff --git a/src/loaders/compute_tools/eggnog/versions.yaml b/src/loaders/compute_tools/eggnog/versions.yaml
index 3bb4d81d..d044b7fc 100644
--- a/src/loaders/compute_tools/eggnog/versions.yaml
+++ b/src/loaders/compute_tools/eggnog/versions.yaml
@@ -15,4 +15,10 @@ versions:
       - Remove the 'threads' and 'program_threads' parameters 
         and introduce the capability to specify the number of threads per tool run
       - pass job_id to the tool container and remove node_id
+    reference_db_version: 5.0.2
+
+  - version: 0.1.3
+    date: 2024-06-25
+    notes: |
+      - Bug fix: tool_common.py - Converted Data IDs to string format to ensure proper comparison with associated folder names
     reference_db_version: 5.0.2
\ No newline at end of file
diff --git a/src/loaders/compute_tools/gtdb_tk/versions.yaml b/src/loaders/compute_tools/gtdb_tk/versions.yaml
index 5a92d88f..5acca88d 100644
--- a/src/loaders/compute_tools/gtdb_tk/versions.yaml
+++ b/src/loaders/compute_tools/gtdb_tk/versions.yaml
@@ -35,4 +35,9 @@ versions:
       - Remove the 'threads' and 'program_threads' parameters 
         and introduce the capability to specify the number of threads per tool run
       - pass job_id to the tool container and remove node_id
+    reference_db_version: release214
+  - version: 0.1.6
+    date: 2024-06-25
+    notes: |
+      - Bug fix: tool_common.py - Converted Data IDs to string format to ensure proper comparison with associated folder names
     reference_db_version: release214
\ No newline at end of file
diff --git a/src/loaders/compute_tools/mash/versions.yaml b/src/loaders/compute_tools/mash/versions.yaml
index 90975601..96226408 100644
--- a/src/loaders/compute_tools/mash/versions.yaml
+++ b/src/loaders/compute_tools/mash/versions.yaml
@@ -12,4 +12,8 @@ versions:
     notes: |
       - Remove the 'threads' and 'program_threads' parameters 
         and introduce the capability to specify the number of threads per tool run
-      - pass job_id to the tool container and remove node_id
\ No newline at end of file
+      - pass job_id to the tool container and remove node_id
+  - version: 0.1.4
+    date: 2024-06-25
+    notes: |
+      - Bug fix: tool_common.py - Converted Data IDs to string format to ensure proper comparison with associated folder names
\ No newline at end of file
diff --git a/src/loaders/compute_tools/microtrait/versions.yaml b/src/loaders/compute_tools/microtrait/versions.yaml
index bc844a43..16c272aa 100644
--- a/src/loaders/compute_tools/microtrait/versions.yaml
+++ b/src/loaders/compute_tools/microtrait/versions.yaml
@@ -18,4 +18,8 @@ versions:
     notes: |
       - Remove the 'threads' and 'program_threads' parameters 
         and introduce the capability to specify the number of threads per tool run
-      - pass job_id to the tool container and remove node_id
\ No newline at end of file
+      - pass job_id to the tool container and remove node_id
+  - version: 0.1.5
+    date: 2024-06-25
+    notes: |
+      - Bug fix: tool_common.py - Converted Data IDs to string format to ensure proper comparison with associated folder names
\ No newline at end of file
diff --git a/src/loaders/compute_tools/tool_common.py b/src/loaders/compute_tools/tool_common.py
index 2c612dd4..9769e2eb 100644
--- a/src/loaders/compute_tools/tool_common.py
+++ b/src/loaders/compute_tools/tool_common.py
@@ -211,7 +211,7 @@ def _get_data_ids(self):
             with self._data_id_file:
                 df = pd.read_csv(self._data_id_file, sep='\t')
                 try:
-                    data_ids = df[loader_common_names.DATA_ID_COLUMN_HEADER]
+                    data_ids = df[loader_common_names.DATA_ID_COLUMN_HEADER].astype(str).tolist()  # convert to string in case of int directory names
                 except KeyError:
                     raise ValueError(
                         f"Please ensure {loader_common_names.DATA_ID_COLUMN_HEADER} column exists in the "