Merge pull request #632 from Ecogenomics/staging

New release for R226

Author: pchaumeil

docs/src/announcements.rst

@@ -1,6 +1,25 @@
 Announcements
 =============
 
+GTDB-Tk 2.4.1 available
+-----------------------
+
+*April 16, 2024*
+
+* GTDB-Tk version ``2.4.1`` is now available.
+* This version of GTDB-Tk compatible with the new version of the GTDB-Tk reference package (Release 226).
+  `gtdbtk_r226_data.tar.gz <https://data.gtdb.ecogenomic.org/releases/release220/220.0/auxillary_files/gtdbtk_package/>`_.
+
+
+GTDB R226 available
+-----------------------
+
+*April 16, 2025*
+
+* GTDB Release 226 is now available and will be used from version ``2.4.1`` and up.
+* This version of GTDB-Tk is compatible with both release220 and release226 of the GTDB-Tk reference package.
+  `gtdbtk_r226_data.tar.gz <https://data.gtdb.ecogenomic.org/releases/release226/226.0/auxillary_files/>`_.
+
 GTDB-Tk 2.4.0 available
 -----------------------
 

docs/src/changelog.rst

@@ -2,6 +2,12 @@
 Change log
 ==========
 
+2.4.1
+-----
+
+Bug Fixes:
+* (`#630 <https://github.com/Ecogenomics/GTDBTk/issues/630>`_) Fixed SyntaxWarning in Python 3.12 by using raw strings for regex in HMMResultsIO.py
+
 
 2.4.0
 -----

docs/src/installing/index.rst

@@ -34,11 +34,11 @@ Hardware requirements
      - Time
    * - Archaea
      - ~60 GB
-     - ~106 GB
+     - ~140 GB
      - ~90 minutes / 1,000 genomes @ 64 CPUs
    * - Bacteria
-     - ~90GB (545 GB when using --full_tree)
-     - ~106 GB
+     - ~110GB (690 GB when using --full_tree)
+     - ~140 GB
      - ~90 minutes / 1,000 genomes @ 64 CPUs
 
 .. note::
@@ -117,7 +117,7 @@ Please cite these tools if you use GTDB-Tk in your work.
 GTDB-Tk reference data
 ----------------------
 
-GTDB-Tk requires ~110G of external data that needs to be downloaded and unarchived:
+GTDB-Tk requires ~140G of external data that needs to be downloaded and unarchived:
 
 **For full package:**
 
@@ -154,7 +154,11 @@ https://data.ace.uq.edu.au/public/gtdb/data/releases/release220/220.0/auxillary_
      - Minimum version
      - Maximum version
      - MD5
-   * - `R220 <https://data.ace.uq.edu.au/public/gtdb/data/releases/latest/auxillary_files/gtdbtk_package/full_package/gtdbtk_data.tar.gz>`_
+   * - `R226 <https://data.gtdb.ecogenomic.org/releases/release220/226.0/auxillary_files/gtdbtk_package/full_package/gtdbtk_r226_data.tar.gz>`_
+     - 2.4.0
+     - Current
+     - 24b476ea5a4ef30519d461e56cc4a27f
+   * - `R220 <https://data.gtdb.ecogenomic.org/releases/release220/220.0/auxillary_files/gtdbtk_package/full_package/gtdbtk_r220_data.tar.gz>`_
      - 2.4.0
      - Current
      - 5aafa1b9c27ceda003d75adf238ed9e0

gtdbtk/__init__.py

@@ -29,4 +29,4 @@
 __status__ = 'Production'
 __title__ = 'GTDB-Tk'
 __url__ = 'https://github.com/Ecogenomics/GTDBTk'
-__version__ = '2.4.0'
+__version__ = '2.4.1'

gtdbtk/config/common.py

@@ -12,7 +12,7 @@ class __GTDBTkCommonConfig:
     """
 
     MIN_REF_DATA_VERSION = 'r220'
-    COMPATIBLE_REF_DATA_VERSIONS = ['r220']
+    COMPATIBLE_REF_DATA_VERSIONS = ['r220','r226']
 
     BACKBONE_PPLACER_REF_PKG = 'gtdbtk_package_backbone.refpkg'
     CLASS_LEVEL_PPLACER_REF_PKG = 'gtdbtk.package.{iter}.refpkg'
@@ -332,8 +332,20 @@ def get_REF_HASHES(self,version=None):
         compatible_versions = [int(x.replace('r','')) for x in CONFIG.COMPATIBLE_REF_DATA_VERSIONS]
         if version is not None and version not in compatible_versions:
             raise ValueError(f"Version {version} is not compatible with this version of GTDB-Tk. Compatible versions are {compatible_versions}")
+        if version is None or version==226:
+            return {
+                self.PPLACER_DIR: '40d242bd5b84d4b2218586390220e6f741191804',
+                self.MASK_DIR: 'beb296ca29ffadf2d5b86f4160f2a525c5954185',
+                self.MARKER_DIR: '163f542c3f0a40f59df45d453aa235b39aa96e27',
+                self.RADII_DIR: '577663ffcd7892551a0f8cb2864fb80544c3bfcd',
+                self.MSA_FOLDER: '1d47808337897c2965cfc8a101df5e4813ce1581',
+                self.METADATA_DIR: '56b3a753b0cec1f0ac983fea4c7db32a160d00cb',
+                self.TAX_FOLDER: 'ea99878828e0baffe8977c7371ce7b7d5109c1e1',
+                self.SKANI_DIR: '8100f226743d74572fa726ac3ea358c34d5b8222',
+                self.RED_DIR: 'bd8cfe5c4f5b8fc180cebf974713194395efbc00'
+            }
 
-        if version is None or version==220:
+        elif version==220:
             return {
                 self.PPLACER_DIR: '75fdd0e093c9af6a73cb510c3d0cd2041265e093',
                 self.MASK_DIR: 'f4b8ebfa59526a7a86f09752b47e8de1efc384c7',

gtdbtk/external/pypfam/HMM/HMMResultsIO.py

@@ -331,15 +331,15 @@ def _readUnitData(self, seqId, fh, hmmRes):
             if hmmName and hmmRes.program == 'hmmsearch':
 
                 pattern1 = re.compile(r'^\s+%s\s+\d+\s+(\S+)\s+\d+' % hmmName)
-                seqId = re.sub('(\W)', r'\\\1', seqId)
+                seqId = re.sub(r'(\W)', r'\\\1', seqId)
                 # $id =~ s/\|/\\|/g;  #Escape '|', '[' and ']' characters
                 # $id =~ s/\[/\\[/g;
                 # $id =~ s/\]/\\]/g;
                 pattern2 = re.compile(r'^\s+%s\s+\d+\s+(\S+)\s+\d+' % seqId)
 
             elif seqName and hmmRes.program == 'hmmscan':
                 tmpSeqName = seqName
-                tmpSeqName = re.sub('(\W)', r'\\\1', tmpSeqName)
+                tmpSeqName = re.sub(r'(\W)', r'\\\1', tmpSeqName)
 
                 pattern1 = re.compile(r'^\s+%s\s+\d+\s+(\S+)\s+\d+' % seqId)
                 pattern2 = re.compile(r'^\s+%s\s+\d+\s+(\S+)\s+\d+' % tmpSeqName)