metapackage: Default S4.2.2, require zenodo_backpack 0.3.0.

singlem.yml

@@ -28,19 +28,19 @@ dependencies:
   - sqlalchemy
   - pandas
   - bird_tool_utils_python>=0.4.1
-  - zenodo_backpack
   - smafa>=0.7.0
   - pyranges
   - polars>=0.19.3 # supplement with taxon genome length requires this, for strip_chars
   - prodigal
   - tqdm
   - pyarrow
-  # - galah >= 0.4.0 # 0.4.0 is the first version with checkm2 quality inputs, but not yet released 
+  - galah >= 0.4.0 # 0.4.0 is the first version with checkm2 quality inputs, but not yet released 
   - pytest # For testing only
   - ipython # For testing only
   - sqlparse # Don't understand why, but this isn't being installed before tests are run on GH actions. Required indirectly (by taxtastic).
   # - python-annoy optional
   # - nmslib # optional
-  # - pip:
+  - pip:
+    - zenodo_backpack>=0.3.0 # required for version parameter in metapackage download
   #   - cmake # Required so scann can be installed and run via pip. Without this, get LD_LIBRARY_PATH error when running pipe.
   #   - scann # optional

singlem/metapackage.py

@@ -14,7 +14,7 @@
 from .metapackage_read_name_store import MetapackageReadNameStore
 
 # These constants should be updated in tandem
-DATA_DEFAULT_VERSION = '3.2.1'
+DATA_DEFAULT_VERSION = '4.2.2'
 DATA_ENVIRONMENT_VARIABLE = 'SINGLEM_METAPACKAGE_PATH'
 DATA_DOI = '10.5281/zenodo.5739611'
 GTDB_DATABASE_NAME = 'Genome Taxonomy Database (GTDB)'
@@ -173,11 +173,12 @@ def download(**kwargs):
         if len(kwargs) > 0:
             raise Exception("Unexpected arguments detected: %s" % kwargs)
 
-        logging.info("Downloading data with ZenodoBackpack ..")
+        logging.info(f"Downloading data version {DATA_DEFAULT_VERSION} with ZenodoBackpack from {DATA_DOI} ..")
         backpack = zenodo_backpack.ZenodoBackpackDownloader().download_and_extract(
             output_directory,
             DATA_DOI,
-            progress_bar=True)
+            progress_bar=True,
+            version=DATA_DEFAULT_VERSION)
         logging.info("Finished downloading data")
 
         logging.info("The environment variable {} can now be set to {}".format(DATA_ENVIRONMENT_VARIABLE, output_directory))