Move more project configs to config.py and download HMM to .phyling/H…

…MM instead of user specified path.

src/phyling/config.py

@@ -1,8 +1,12 @@
 """Some project specific configuration parameters."""
-prot_aln_ext = 'aa.mfa'  # protein alignment extension
-cds_aln_ext = 'cds.mfa'  # cds alignment extension
-protein_ext = 'faa'  # protein fasta files
-cds_ext = 'cds'  # coding sequence alignment file extension
-aln_ext = 'mfa'  # general alignment extension
+from pathlib import Path
 
-default_HMM = 'HMM'  # default directory for HMM downloads
+database = "https://busco-data.ezlab.org/v5/data"
+cfg_dir = Path.home() / ".phyling"
+default_HMM = "HMM"  # default directory for HMM downloads
+
+prot_aln_ext = "aa.mfa"  # protein alignment extension
+cds_aln_ext = "cds.mfa"  # cds alignment extension
+protein_ext = "faa"  # protein fasta files
+cds_ext = "cds"  # coding sequence alignment file extension
+aln_ext = "mfa"  # general alignment extension

src/phyling/download.py

@@ -12,6 +12,8 @@
 from urllib.error import HTTPError
 from urllib.request import urlopen
 
+import phyling.config
+
 
 class Data_updater(ABC):
     """Store, update and retrieve BUSCO markers."""
@@ -92,11 +94,11 @@ class Metadata_updater(Data_updater):
     dataset names for downloading BUSCO markers.
     """
 
-    def __init__(self, database_url, cfg_dir):
+    def __init__(self, database_url):
         """Initialize metadata object and local db file."""
         super().__init__(database_url)
         self._filetype = "metadata"
-        self._data = cfg_dir / "metadata.pickle"
+        self._data = phyling.config.cfg_dir / "metadata.pickle"
         self._data_url = f"{database_url}/file_versions.tsv"
         self._metadata_md5_url = f"{database_url}/file_versions.tsv.hash"
 
@@ -189,17 +191,17 @@ def _save_data(self, data):
 #     return hasher.hexdigest()
 
 
-def download(database, cfg_dir, markerset, output, **kwargs) -> None:
+def download(markerset, **kwargs) -> None:
     """
     Help to download/update BUSCO v5 markerset to a local folder.
 
-    First it check whether the ~/.phyling/metadata.pickle is exist. A missing or outdated file will trigger the module
-    to update the metadata.
+    First it checks whether the metadata file is exist under the config folder ~/.phyling. A missing or outdated file
+    will trigger the module to download/update the metadata.
 
     Passing "list" to markerset argument will list all the available markersets. Passing a valid name to the markerset
     argument will download the markerset to the given output path.
     """
-    metadata_updater = Metadata_updater(database_url=database, cfg_dir=cfg_dir)
+    metadata_updater = Metadata_updater(database_url=phyling.config.database)
     markerset_dict = metadata_updater.updater()
 
     if markerset == "list":
@@ -211,16 +213,16 @@ def download(database, cfg_dir, markerset, output, **kwargs) -> None:
         # Adjust databases display according to the terminal size
         width, _ = shutil.get_terminal_size((80, 24))
         col = width // 40
-        url_list = [url_list[x: x + col] for x in range(0, len(url_list), col)]
+        url_list = [url_list[x : x + col] for x in range(0, len(url_list), col)]
         col_width = max(len(word) for row in url_list for word in row) + 3  # padding
         for row in url_list:
             # Print the database list
             print(" ".join(word.ljust(col_width) for word in row))
 
     else:
         hmm_markerset_updater = HMM_markerset_updater(
-            database_url=database,
-            output_dir=output,
+            database_url=phyling.config.database,
+            output_dir=Path(phyling.config.cfg_dir, phyling.config.default_HMM),
             metadata=markerset_dict,
             name=markerset,
         )

src/phyling/libphyling.py

@@ -600,9 +600,10 @@ def main(inputs, input_dir, output, markerset, evalue, method, non_trim, from_ch
         sys.exit(1)
 
     if not markerset.exists():
-        markerset = Path(phyling.config.default_HMM, markerset, "hmms")
+        markerset = Path(phyling.config.cfg_dir, phyling.config.default_HMM, markerset, "hmms")
     else:
         markerset = Path(markerset)
+    logging.info(f"Loading markerset from {markerset}")
 
     if not markerset.exists():
         logging.error(f"Markerset folder does not exist {markerset} - did you download BUSCO?")

src/phyling/phyling.py

@@ -7,7 +7,6 @@
 import sys
 import textwrap
 from pathlib import Path
-from types import SimpleNamespace
 
 try:
     from importlib.metadata import version
@@ -34,13 +33,6 @@ def parser_submodule(parser, parent_parser) -> None:
         description=textwrap.dedent(download.__doc__),
     )
     p_download.add_argument("markerset", metavar='HMM markerset or "list"', help="Name of the HMM markerset")
-    p_download.add_argument(
-        "-o",
-        "--output",
-        type=Path,
-        default=f"./{phyling.config.default_HMM}",
-        help=f'Output directory to save HMM markerset (default="./{phyling.config.default_HMM}")',
-    )
     p_download.set_defaults(func=download)
 
     p_aln = subparsers.add_parser(
@@ -166,16 +158,9 @@ def main():
     """
     logging.basicConfig(format=f"%(asctime)s {main.__name__} %(levelname)s %(message)s", level="INFO", force=True)
     logger = logging.getLogger()
-    # Create namespace object conf
-    args = dict()
-
-    args["script_path"] = Path(__file__).resolve().parent
-    args["database"] = "https://busco-data.ezlab.org/v5/data"
-    args["cfg_dir"] = Path.home() / ".phyling"
-    args = SimpleNamespace(**args)
 
     # Create config folder in $HOME/.phyling
-    args.cfg_dir.mkdir(exist_ok=True)
+    phyling.config.cfg_dir.mkdir(exist_ok=True)
 
     # Implement shared arguments between sub-menu, reference from
     # https://stackoverflow.com/questions/33645859/how-to-add-common-arguments-to-argparse-subcommands
@@ -200,7 +185,7 @@ def main():
         parser.print_help(sys.stderr)
         sys.exit(1)
 
-    parser.parse_args(namespace=args)
+    args = parser.parse_args()
 
     if args.verbose:
         logger.setLevel("DEBUG")